diff --git a/.gitignore b/.gitignore
index 5a438dc04..8c3981be2 100644
--- a/.gitignore
+++ b/.gitignore
@@ -25,6 +25,12 @@ src/build/libadflow-f2pywrappers.f
 src/build/libadflowmodule.c
 src/build/importTest.py
 
+# unused fast_b routines generated by Tapenade
+src/adjoint/outputReverseFast/adjointextra_fast_b.f90
+src/adjoint/outputReverseFast/bcdata_fast_b.f90
+src/adjoint/outputReverseFast/oversetutilities_fast_b.f90
+src/adjoint/outputReverseFast/zipperintegrations_fast_b.f90
+
 # Regression test auxiliary files:
 input_files.tar.gz
 adflow_input_files.tar.gz
diff --git a/adflow/pyADflow.py b/adflow/pyADflow.py
index 3c7a6a284..5a2f805a2 100644
--- a/adflow/pyADflow.py
+++ b/adflow/pyADflow.py
@@ -141,6 +141,9 @@ def __init__(self, comm=None, options=None, debug=False, dtype="d"):
         # Set all internal adflow default options before we set anything from python
         self.adflow.inputparamroutines.setdefaultvalues()
 
+        # surface-roughness overwrite-dict
+        self.surfaceRoughness = dict()
+
         defSetupTime = time.time()
 
         # Initialize the inherited AeroSolver
@@ -1162,6 +1165,50 @@ def setRotationRate(self, rotCenter, rotRate, cgnsBlocks=None):
         self.adflow.preprocessingapi.updaterotationrate(rotCenter, rotations, cgnsBlocks)
         self._updateVelInfo = True
 
+    def setSurfaceRoughness(self, ks, groupName=None):
+        """
+        Sets the equivalent sandgrain roughness on the specified surface family.
+        It may only be used when the rough SA-variant (useRoughSA = True) is
+        active.
+
+        This function takes precendence over all boundary conditions set in the
+        mesh.
+
+        Parameters:
+        ----------
+        ks : float
+            Equivalent sandgrain roughness
+        groupName : str
+            Family group to use. Default to all walls if not given (None)
+        """
+
+        if not self.options["useroughsa"]:
+            raise Error(
+                "It is not possible to set a surface roughness value without "
+                "using the rough SA-variant (useRoughSA = False)"
+            )
+
+        if groupName is None:
+            groupName = self.allWallsGroup
+
+        if groupName in self.surfaceRoughness:
+            raise Error(
+                f"The roughness value for surface group '{groupName}' has "
+                f"allready been set to {self.surfaceRoughness[groupName]}"
+            )
+
+        self.surfaceRoughness[groupName] = ks
+
+    def _setSurfaceRoughness(self):
+        """
+        Set the actual roughness values. It must be postponed for as long as
+        possible. Otherwise it might get overwritten.
+        """
+
+        for groupName, ks in self.surfaceRoughness.items():
+            famList = self._getFamilyList(groupName)
+            self.adflow.preprocessingapi.updatesurfaceroughness(ks, famList)
+
     def checkPartitioning(self, nprocs):
         """This function determines the potential load balancing for
         nprocs. The intent is this function can be run in serial
@@ -3706,6 +3753,13 @@ def _setAeroProblemData(self, aeroProblem, firstCall=False):
                 self.adflow.preprocessingapi.updatemetricsalllevels()
             self.adflow.preprocessingapi.updategridvelocitiesalllevels()
 
+            # overwrite BC surface roughness
+            if len(self.surfaceRoughness) > 0:
+                self._setSurfaceRoughness()
+
+            # Propagate roughness values through volume
+            self.adflow.walldistance.updatewallroughness()
+
     def _getBCDataFromAeroProblem(self, AP):
         variables = []
         dataArray = []
@@ -5675,6 +5729,7 @@ def _getDefaultOptions():
             "useQCR": [bool, False],
             "useRotationSA": [bool, False],
             "useft2SA": [bool, True],
+            "useRoughSA": [bool, False],
             "eddyVisInfRatio": [float, 0.009],
             "useWallFunctions": [bool, False],
             "useApproxWallDistance": [bool, True],
@@ -6070,6 +6125,7 @@ def _getOptionMap(self):
             "useqcr": ["physics", "useqcr"],
             "userotationsa": ["physics", "userotationsa"],
             "useft2sa": ["physics", "useft2sa"],
+            "useroughsa": ["physics", "useroughsa"],
             "eddyvisinfratio": ["physics", "eddyvisinfratio"],
             "usewallfunctions": ["physics", "wallfunctions"],
             "walldistcutoff": ["physics", "walldistcutoff"],
diff --git a/doc/options.yaml b/doc/options.yaml
index f37f0f5da..826e31c7b 100644
--- a/doc/options.yaml
+++ b/doc/options.yaml
@@ -237,6 +237,15 @@ useft2SA:
   desc: >
     Include the ft2 term in the Spalart-Allmaras turbulence model.
 
+useRoughSA:
+  desc: >
+    Activates the rough SA variant described `here
+    <https://turbmodels.larc.nasa.gov/spalart.html#sarough>`__. To use it,
+    blockettes must be disabled ``useBlockettes: False``. The roughness value on
+    the boundary may be prescribed like a regular boundary value using the name
+    ``SandGrainRoughness``. Additionally, the ADflow method ``setSurfaceRoughness``
+    may be used.
+
 eddyVisInfRatio:
   desc: >
     Free stream value of eddy viscosity.
@@ -1613,4 +1622,3 @@ meshMaxSkewness:
 useSkewnessCheck:
   desc: >
     When set to true, ADflow computes the `skewness` of each cell and throws an error if it is above ``meshMaxSkewness``. See also ``printBadlySkewedCells``.
-
diff --git a/src/NKSolver/blockette.F90 b/src/NKSolver/blockette.F90
index 6ce106c2e..3f34686f0 100644
--- a/src/NKSolver/blockette.F90
+++ b/src/NKSolver/blockette.F90
@@ -86,7 +86,7 @@ subroutine blocketteRes(useDissApprox, useViscApprox, useUpdateIntermed, useFlow
         use initializeFlow, only: referenceState
         use section, only: sections, nSections
         use iteration, only: rFil, currentLevel
-        use haloExchange, only: exchangeCoor, whalo2
+        use haloExchange, only: exchangeCoor, whalo2, exchanged2Wall
         use wallDistance, only: updateWallDistancesQuickly
         use utils, only: setPointers, EChk
         use turbUtils, only: computeEddyViscosity
@@ -181,12 +181,20 @@ subroutine blocketteRes(useDissApprox, useViscApprox, useUpdateIntermed, useFlow
                 do nn = 1, nDom
                     call setPointers(nn, currentLevel, sps)
                     call xhalo_block()
+
+                    if (equations == RANSEquations .and. useApproxWallDistance) then
+                        call updateWallDistancesQuickly(nn, 1, sps)
+                    end if
                 end do
             end do
 
             ! Now exchange the coordinates (fine level only)
             call exchangecoor(1)
 
+            if (equations == RANSEquations .and. useApproxWallDistance) then
+                call exchanged2Wall(1)
+            end if
+
             do sps = 1, nTimeIntervalsSpectral
                 ! Update overset connectivity if necessary
                 if (oversetPresent .and. oversetUpdateMode == updateFast) then
diff --git a/src/adjoint/Makefile_tapenade b/src/adjoint/Makefile_tapenade
index d0077a2ea..fc2686635 100644
--- a/src/adjoint/Makefile_tapenade
+++ b/src/adjoint/Makefile_tapenade
@@ -132,17 +132,17 @@ flowUtils%computeEtotBlock(w, p) > \
 \
 turbUtils%computeEddyViscosity(w,rlv)>(w,rlv,rev) \
 \
-BCRoutines%bcSymm1stHalo(ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, bcData%norm) > \
-                        (ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, bcData%norm) \
+BCRoutines%bcSymm1stHalo(ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, d2wall1, d2wall2, bcData%norm) > \
+                        (ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, d2wall1, d2wall2, bcData%norm) \
 \
-BCRoutines%bcSymm2ndHalo(ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, bcData%norm) > \
-                        (ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, bcData%norm) \
+BCRoutines%bcSymm2ndHalo(ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, bd2wall0, d2wall3, cData%norm) > \
+                        (ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, bd2wall0, d2wall3, bcData%norm) \
 \
-BCRoutines%bcSymmPolar1stHalo(xx, ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2) > \
-                             (xx, ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2) \
+BCRoutines%bcSymmPolar1stHalo(xx, ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, d2wall1, d2wall2) > \
+                             (xx, ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, d2wall1, d2wall2) \
 \
-BCRoutines%bcSymmPolar2ndHalo(xx, ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3) > \
-                             (xx, ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3) \
+BCRoutines%bcSymmPolar2ndHalo(xx, ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, d2wall0, d2wall3) > \
+                             (xx, ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, d2wall0, d2wall3) \
 \
 BCRoutines%bcNSWallAdiabatic(ww0, ww1, ww2, pp0, pp1, pp2, pp3, rlv0, rlv1, rlv2, rev0, rev1, rev2, bcData%uSlip) > \
                             (ww0, ww1, ww2, pp0, pp1, pp2, pp3, rlv0, rlv1, rlv2, rev0, rev1, rev2, bcData%uSlip) \
@@ -162,11 +162,11 @@ BCRoutines%bcSubsonicInflow(ww0, ww1, ww2, pp0, pp1, pp2, rlv0, rlv1, rlv2, rev0
 BCRoutines%bcSubsonicOutflow(ww0, ww1, ww2, pp0, pp1, pp2, rlv0, rlv1, rlv2, rev0, rev1, rev2, bcData%norm, bcData%Ps) > \
                             (ww0, ww1, ww2, pp0, pp1, pp2, rlv0, rlv1, rlv2, rev0, rev1, rev2, bcData%norm, bcData%Ps) \
 \
-turbBCRoutines%applyAllTurbBCThisBlock(rev, w, bvtj1, bvtj2, bvtk1, bvtk2, bvti1, bvti2) > \
-                                      (rev, w) \
+turbBCRoutines%applyAllTurbBCThisBlock(rev, w, bvtj1, bvtj2, bvtk1, bvtk2, bvti1, bvti2, bmti1, bmti2, bmtj1, bmtj2, bmtk1, bmtk2, d2wall) > \
+                                      (rev, w, bvtj1, bvtj2, bvtk1, bvtk2, bvti1, bvti2, bmti1, bmti2, bmtj1, bmtj2, bmtk1, bmtk2, d2wall) \
 \
-turbBCRoutines%bcTurbTreatment(w, rlv, d2wall, winf) > \
-                              (w, rlv, d2wall, winf, bvtj1, bvtj2, bvtk1, bvtk2, bvti1, bvti2) \
+turbBCRoutines%bcTurbTreatment(w, rlv, d2wall, winf, bvtj1, bvtj2, bvtk1, bvtk2, bvti1, bvti2, bmti1, bmti2, bmtj1, bmtj2, bmtk1, bmtk2) > \
+                              (w, rlv, d2wall, winf, bvtj1, bvtj2, bvtk1, bvtk2, bvti1, bvti2, bmti1, bmti2, bmtj1, bmtj2, bmtk1, bmtk2) \
 \
 solverUtils%timeStep_block(w, pInfCorr, rhoInf, si, sj, sk, sFaceI, sFaceJ, sFaceK, p, radi, radj, radk, dtl, rlv, rev, vol) > \
               (w, pInfCorr, rhoInf, si, sj, sk, sFaceI, sFaceJ, sFaceK, p, radi, radj, radk, dtl, rlv, rev, vol) \
diff --git a/src/adjoint/adjointUtils.F90 b/src/adjoint/adjointUtils.F90
index 14641f51f..b2a5a85a2 100644
--- a/src/adjoint/adjointUtils.F90
+++ b/src/adjoint/adjointUtils.F90
@@ -772,8 +772,10 @@ subroutine allocDerivativeValues(level)
 
                 ! Allocate d2wall if not already done so
                 if (.not. associated(flowDoms(nn, 1, sps)%d2wall)) then
-                    allocate (flowDoms(nn, 1, sps)%d2wall(2:il, 2:jl, 2:kl))
+                    allocate (flowDoms(nn, 1, sps)%d2wall(0:ib, 0:jb, 0:kb))
                     call EChk(ierr, __FILE__, __LINE__)
+
+                    flowDoms(nn, level, sps)%d2Wall = 0.01
                 end if
 
                 ! Now allocate all valus that have a differentiable
@@ -829,7 +831,7 @@ subroutine allocDerivativeValues(level)
                     flowDomsd(nn, level, sps)%bvtj2(ie, ke, nt1:nt2), &
                     flowDomsd(nn, level, sps)%bvtk1(ie, je, nt1:nt2), &
                     flowDomsd(nn, level, sps)%bvtk2(ie, je, nt1:nt2), &
-                    flowDomsd(nn, level, sps)%d2Wall(2:il, 2:jl, 2:kl), &
+                    flowDomsd(nn, level, sps)%d2Wall(0:ib, 0:jb, 0:kb), &
                     stat=ierr)
                 call EChk(ierr, __FILE__, __LINE__)
 
diff --git a/src/adjoint/outputForward/BCData_d.f90 b/src/adjoint/outputForward/BCData_d.f90
index 5f28fb4df..9593cac1f 100644
--- a/src/adjoint/outputForward/BCData_d.f90
+++ b/src/adjoint/outputForward/BCData_d.f90
@@ -13,11 +13,28 @@ module bcdata_d
   subroutine setbcvarnamesisothermalwall()
     use cgnsnames
     use constants
+    use inputphysics, only : useroughsa
     implicit none
     nbcvar = nbcvarisothermalwall
     bcvarnames(1) = cgnstemp
+    if (useroughsa) then
+      nbcvar = nbcvar + 1
+      bcvarnames(2) = cgnssandgrainroughness
+    end if
   end subroutine setbcvarnamesisothermalwall
 
+  subroutine setbcvarnamesadiabaticwall()
+    use cgnsnames
+    use constants
+    use inputphysics, only : useroughsa
+    implicit none
+    nbcvar = nbcvaradiabaticwall
+    if (useroughsa) then
+      nbcvar = nbcvar + 1
+      bcvarnames(1) = cgnssandgrainroughness
+    end if
+  end subroutine setbcvarnamesadiabaticwall
+
   subroutine setbcvarnamessubsonicinflow()
     use constants
     use cgnsnames
@@ -329,10 +346,10 @@ subroutine unitvectorscylsystem(boco)
   end subroutine unitvectorscylsystem
 
 !  differentiation of bcdataisothermalwall in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *(*bcdata.tns_wall)
+!   variations   of useful results: *(*bcdata.tns_wall) bcvararray
 !   with respect to varying inputs: tref bcvararray
 !   rw status of diff variables: tref:in *(*bcdata.tns_wall):out
-!                bcvararray:in
+!                bcvararray:in-out
 !   plus diff mem management of: bcdata:in *bcdata.tns_wall:in
 ! ---------------------------------------------------------------
 ! routines that set the actual bcdata values from the cgns data set
@@ -350,6 +367,7 @@ subroutine bcdataisothermalwall_d(boco, bcvararray, bcvararrayd, ibeg&
     use blockpointers, only : bcfaceid, bcdata, bcdatad, nbkglobal
     use utils_d, only : terminate, sitemperature
     use flowvarrefstate, only : tref, trefd
+    use inputphysics, only : useroughsa
     use diffsizes
 !  hint: isize1ofdrfbcdata should be the size of dimension 1 of array *bcdata
     implicit none
@@ -388,6 +406,21 @@ subroutine bcdataisothermalwall_d(boco, bcvararray, bcvararrayd, ibeg&
         bcdata(boco)%tns_wall(i, j) = temp0
       end do
     end do
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+    if (useroughsa) then
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+      if (.not.bcvarpresent(1)) then
+        bcvararrayd(:, :, 2) = 0.0_8
+        bcvararray(:, :, 2) = zero
+      end if
+      do j=jbeg,jend
+        do i=ibeg,iend
+          bcdata(boco)%ksns_wall(i, j) = bcvararray(i, j, 2)
+        end do
+      end do
+    end if
   end subroutine bcdataisothermalwall_d
 
 ! ---------------------------------------------------------------
@@ -406,6 +439,7 @@ subroutine bcdataisothermalwall(boco, bcvararray, ibeg, iend, jbeg, &
     use blockpointers, only : bcfaceid, bcdata, nbkglobal
     use utils_d, only : terminate, sitemperature
     use flowvarrefstate, only : tref
+    use inputphysics, only : useroughsa
     implicit none
 !
 !      subroutine arguments.
@@ -431,8 +465,54 @@ subroutine bcdataisothermalwall(boco, bcvararray, ibeg, iend, jbeg, &
 &         tref
       end do
     end do
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+    if (useroughsa) then
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+      if (.not.bcvarpresent(1)) bcvararray(:, :, 2) = zero
+      do j=jbeg,jend
+        do i=ibeg,iend
+          bcdata(boco)%ksns_wall(i, j) = bcvararray(i, j, 2)
+        end do
+      end do
+    end if
   end subroutine bcdataisothermalwall
 
+  subroutine bcdataadiabaticwall(boco, bcvararray, ibeg, iend, jbeg, &
+&   jend)
+!
+!       tries to extract the equivalent sand grain roughness. it sets
+!       a default value of 0.0
+!
+    use constants
+    use cgnsnames
+    use inputphysics, only : useroughsa
+    use blockpointers, only : bcfaceid, bcdata, nbkglobal
+    implicit none
+!
+!      subroutine arguments.
+!
+    integer(kind=inttype) :: boco
+    integer(kind=inttype) :: ibeg, iend, jbeg, jend
+    real(kind=realtype), dimension(ibeg:iend, jbeg:jend, nbcvarmax) :: &
+&   bcvararray
+!
+!      local variables.
+!
+    integer(kind=inttype) :: i, j
+    if (useroughsa) then
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+      if (.not.bcvarpresent(1)) bcvararray(:, :, 1) = zero
+      do j=jbeg,jend
+        do i=ibeg,iend
+          bcdata(boco)%ksns_wall(i, j) = bcvararray(i, j, 1)
+        end do
+      end do
+    end if
+  end subroutine bcdataadiabaticwall
+
 !  differentiation of bcdatasubsonicinflow in forward (tangent) mode (with options i4 dr8 r8):
 !   variations   of useful results: *(*bcdata.ptinlet) *(*bcdata.ttinlet)
 !                *(*bcdata.htinlet) *(*bcdata.turbinlet)
diff --git a/src/adjoint/outputForward/BCRoutines_d.f90 b/src/adjoint/outputForward/BCRoutines_d.f90
index 3d28d0bbf..16967bf6e 100644
--- a/src/adjoint/outputForward/BCRoutines_d.f90
+++ b/src/adjoint/outputForward/BCRoutines_d.f90
@@ -156,14 +156,15 @@ subroutine applyallbc_block(secondhalo)
   end subroutine applyallbc_block
 
 !  differentiation of bcsymm1sthalo in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *rev1 *pp1 *rlv1 *ww1
-!   with respect to varying inputs: *rev1 *rev2 *pp1 *pp2 *rlv1
-!                *rlv2 *ww1 *ww2 *(*bcdata.norm)
-!   rw status of diff variables: *rev1:in-out *rev2:in *pp1:in-out
-!                *pp2:in *rlv1:in-out *rlv2:in *ww1:in-out *ww2:in
-!                *(*bcdata.norm):in
-!   plus diff mem management of: rev1:in rev2:in pp1:in pp2:in
-!                rlv1:in rlv2:in ww1:in ww2:in bcdata:in *bcdata.norm:in
+!   variations   of useful results: *rev1 *d2wall1 *pp1 *rlv1 *ww1
+!   with respect to varying inputs: *rev1 *rev2 *d2wall1 *d2wall2
+!                *pp1 *pp2 *rlv1 *rlv2 *ww1 *ww2 *(*bcdata.norm)
+!   rw status of diff variables: *rev1:in-out *rev2:in *d2wall1:in-out
+!                *d2wall2:in *pp1:in-out *pp2:in *rlv1:in-out *rlv2:in
+!                *ww1:in-out *ww2:in *(*bcdata.norm):in
+!   plus diff mem management of: rev1:in rev2:in d2wall1:in d2wall2:in
+!                pp1:in pp2:in rlv1:in rlv2:in ww1:in ww2:in bcdata:in
+!                *bcdata.norm:in
 ! ===================================================================
 !   actual implementation of each of the boundary condition routines
 ! ===================================================================
@@ -179,9 +180,11 @@ subroutine bcsymm1sthalo_d(nn)
     use constants
     use blockpointers, only : bcdata, bcdatad
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : gamma1, gamma2, ww1, ww1d, ww2, ww2d, pp1, &
 &   pp1d, pp2, pp2d, rlv1, rlv1d, rlv2, rlv2d, istart, jstart, isize, &
-&   jsize, rev1, rev1d, rev2, rev2d
+&   jsize, rev1, rev1d, rev2, rev2d, d2wall1, d2wall1d, d2wall2, &
+&   d2wall2d, ks1, ks2
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -232,6 +235,9 @@ subroutine bcsymm1sthalo_d(nn)
       gamma1(i, j) = gamma2(i, j)
       pp1d(i, j) = pp2d(i, j)
       pp1(i, j) = pp2(i, j)
+      d2wall1d(i, j) = d2wall2d(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) then
         rlv1d(i, j) = rlv2d(i, j)
         rlv1(i, j) = rlv2(i, j)
@@ -258,8 +264,10 @@ subroutine bcsymm1sthalo(nn)
     use constants
     use blockpointers, only : bcdata
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : gamma1, gamma2, ww1, ww2, pp1, pp2, rlv1, &
-&   rlv2, istart, jstart, isize, jsize, rev1, rev2
+&   rlv2, istart, jstart, isize, jsize, rev1, rev2, d2wall1, d2wall2, &
+&   ks1, ks2
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -289,6 +297,8 @@ subroutine bcsymm1sthalo(nn)
 ! laminar and eddy viscosity in the halo.
       gamma1(i, j) = gamma2(i, j)
       pp1(i, j) = pp2(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) rlv1(i, j) = rlv2(i, j)
       if (eddymodel) rev1(i, j) = rev2(i, j)
     end do
@@ -297,12 +307,11 @@ end subroutine bcsymm1sthalo
 !  differentiation of bcsymm2ndhalo in forward (tangent) mode (with options i4 dr8 r8):
 !   variations   of useful results: *rev0 *pp0 *rlv0 *ww0
 !   with respect to varying inputs: *rev0 *rev3 *pp0 *pp3 *rlv0
-!                *rlv3 *ww0 *ww3 *(*bcdata.norm)
+!                *rlv3 *ww0 *ww3
 !   rw status of diff variables: *rev0:in-out *rev3:in *pp0:in-out
 !                *pp3:in *rlv0:in-out *rlv3:in *ww0:in-out *ww3:in
-!                *(*bcdata.norm):in
 !   plus diff mem management of: rev0:in rev3:in pp0:in pp3:in
-!                rlv0:in rlv3:in ww0:in ww3:in bcdata:in *bcdata.norm:in
+!                rlv0:in rlv3:in ww0:in ww3:in
   subroutine bcsymm2ndhalo_d(nn)
 !  bcsymm2ndhalo applies the symmetry boundary conditions to a
 !  block for the 2nd halo. this routine is separate as it makes
@@ -310,9 +319,11 @@ subroutine bcsymm2ndhalo_d(nn)
     use constants
     use blockpointers, only : bcdata, bcdatad
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : gamma0, gamma3, ww0, ww0d, ww3, ww3d, pp0, &
-&   pp0d, pp3, pp3d, rlv0, rlv0d, rlv3, rlv3d, rev0, rev0d, rev3, rev3d,&
-&   istart, jstart, isize, jsize
+&   pp0d, pp3, pp3d, rlv0, rlv0d, rlv3, rlv3d, d2wall0, d2wall0d, &
+&   d2wall3, d2wall3d, ks0, ks3, rev0, rev0d, rev3, rev3d, istart, &
+&   jstart, isize, jsize
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -332,27 +343,22 @@ subroutine bcsymm2ndhalo_d(nn)
       temp = bcdata(nn)%norm(i, j, 1)
       temp0 = bcdata(nn)%norm(i, j, 2)
       temp1 = bcdata(nn)%norm(i, j, 3)
-      vnd = two*(temp*ww3d(i, j, ivx)+ww3(i, j, ivx)*bcdatad(nn)%norm(i&
-&       , j, 1)+temp0*ww3d(i, j, ivy)+ww3(i, j, ivy)*bcdatad(nn)%norm(i&
-&       , j, 2)+temp1*ww3d(i, j, ivz)+ww3(i, j, ivz)*bcdatad(nn)%norm(i&
-&       , j, 3))
-      vn = two*(ww3(i, j, ivx)*temp+ww3(i, j, ivy)*temp0+ww3(i, j, ivz)*&
-&       temp1)
+      vnd = two*(temp*ww3d(i, j, ivx)+temp0*ww3d(i, j, ivy)+temp1*ww3d(i&
+&       , j, ivz))
+      vn = two*(temp*ww3(i, j, ivx)+temp0*ww3(i, j, ivy)+temp1*ww3(i, j&
+&       , ivz))
 ! determine the flow variables in the halo cell.
       ww0d(i, j, irho) = ww3d(i, j, irho)
       ww0(i, j, irho) = ww3(i, j, irho)
       temp1 = bcdata(nn)%norm(i, j, 1)
-      ww0d(i, j, ivx) = ww3d(i, j, ivx) - temp1*vnd - vn*bcdatad(nn)%&
-&       norm(i, j, 1)
-      ww0(i, j, ivx) = ww3(i, j, ivx) - vn*temp1
+      ww0d(i, j, ivx) = ww3d(i, j, ivx) - temp1*vnd
+      ww0(i, j, ivx) = ww3(i, j, ivx) - temp1*vn
       temp1 = bcdata(nn)%norm(i, j, 2)
-      ww0d(i, j, ivy) = ww3d(i, j, ivy) - temp1*vnd - vn*bcdatad(nn)%&
-&       norm(i, j, 2)
-      ww0(i, j, ivy) = ww3(i, j, ivy) - vn*temp1
+      ww0d(i, j, ivy) = ww3d(i, j, ivy) - temp1*vnd
+      ww0(i, j, ivy) = ww3(i, j, ivy) - temp1*vn
       temp1 = bcdata(nn)%norm(i, j, 3)
-      ww0d(i, j, ivz) = ww3d(i, j, ivz) - temp1*vnd - vn*bcdatad(nn)%&
-&       norm(i, j, 3)
-      ww0(i, j, ivz) = ww3(i, j, ivz) - vn*temp1
+      ww0d(i, j, ivz) = ww3d(i, j, ivz) - temp1*vnd
+      ww0(i, j, ivz) = ww3(i, j, ivz) - temp1*vn
       ww0d(i, j, irhoe) = ww3d(i, j, irhoe)
       ww0(i, j, irhoe) = ww3(i, j, irhoe)
 ! set the pressure and gamma and possibly the
@@ -360,6 +366,9 @@ subroutine bcsymm2ndhalo_d(nn)
       gamma0(i, j) = gamma3(i, j)
       pp0d(i, j) = pp3d(i, j)
       pp0(i, j) = pp3(i, j)
+      d2wall0d(i, j) = 0.0_8
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) then
         rlv0d(i, j) = rlv3d(i, j)
         rlv0(i, j) = rlv3(i, j)
@@ -378,8 +387,10 @@ subroutine bcsymm2ndhalo(nn)
     use constants
     use blockpointers, only : bcdata
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : gamma0, gamma3, ww0, ww3, pp0, pp3, rlv0, &
-&   rlv3, rev0, rev3, istart, jstart, isize, jsize
+&   rlv3, d2wall0, d2wall3, ks0, ks3, rev0, rev3, istart, jstart, isize,&
+&   jsize
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -406,19 +417,23 @@ subroutine bcsymm2ndhalo(nn)
 ! laminar and eddy viscosity in the halo.
       gamma0(i, j) = gamma3(i, j)
       pp0(i, j) = pp3(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) rlv0(i, j) = rlv3(i, j)
       if (eddymodel) rev0(i, j) = rev3(i, j)
     end do
   end subroutine bcsymm2ndhalo
 
 !  differentiation of bcsymmpolar1sthalo in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *rev1 *pp1 *rlv1 *ww1
-!   with respect to varying inputs: *xx *rev1 *rev2 *pp1 *pp2 *rlv1
-!                *rlv2 *ww1 *ww2
-!   rw status of diff variables: *xx:in *rev1:in-out *rev2:in *pp1:in-out
-!                *pp2:in *rlv1:in-out *rlv2:in *ww1:in-out *ww2:in
-!   plus diff mem management of: xx:in rev1:in rev2:in pp1:in pp2:in
-!                rlv1:in rlv2:in ww1:in ww2:in
+!   variations   of useful results: *rev1 *d2wall1 *pp1 *rlv1 *ww1
+!   with respect to varying inputs: *xx *rev1 *rev2 *d2wall1 *d2wall2
+!                *pp1 *pp2 *rlv1 *rlv2 *ww1 *ww2
+!   rw status of diff variables: *xx:in *rev1:in-out *rev2:in *d2wall1:in-out
+!                *d2wall2:in *pp1:in-out *pp2:in *rlv1:in-out *rlv2:in
+!                *ww1:in-out *ww2:in
+!   plus diff mem management of: xx:in rev1:in rev2:in d2wall1:in
+!                d2wall2:in pp1:in pp2:in rlv1:in rlv2:in ww1:in
+!                ww2:in
   subroutine bcsymmpolar1sthalo_d(nn)
 ! bcsymmpolar applies the polar symmetry boundary conditions to a
 ! singular line of a block. it is assumed that the pointers in
@@ -427,9 +442,10 @@ subroutine bcsymmpolar1sthalo_d(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 1st level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : ww1, ww1d, ww2, ww2d, pp1, pp1d, pp2, pp2d,&
 &   rlv1, rlv1d, rlv2, rlv2d, rev1, rev1d, rev2, rev2d, xx, xxd, istart,&
-&   jstart, isize, jsize
+&   jstart, isize, jsize, d2wall1, d2wall1d, d2wall2, d2wall2d, ks1, ks2
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -514,6 +530,9 @@ subroutine bcsymmpolar1sthalo_d(nn)
 ! eddy viscosity in the halo.
       pp1d(i, j) = pp2d(i, j)
       pp1(i, j) = pp2(i, j)
+      d2wall1d(i, j) = d2wall2d(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) then
         rlv1d(i, j) = rlv2d(i, j)
         rlv1(i, j) = rlv2(i, j)
@@ -533,8 +552,9 @@ subroutine bcsymmpolar1sthalo(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 1st level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2,&
-&   xx, istart, jstart, isize, jsize
+&   xx, istart, jstart, isize, jsize, d2wall1, d2wall2, ks1, ks2
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -590,19 +610,23 @@ subroutine bcsymmpolar1sthalo(nn)
 ! set the pressure and possibly the laminar and
 ! eddy viscosity in the halo.
       pp1(i, j) = pp2(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) rlv1(i, j) = rlv2(i, j)
       if (eddymodel) rev1(i, j) = rev2(i, j)
     end do
   end subroutine bcsymmpolar1sthalo
 
 !  differentiation of bcsymmpolar2ndhalo in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *rev0 *pp0 *rlv0 *ww0
-!   with respect to varying inputs: *xx *rev0 *rev3 *pp0 *pp3 *rlv0
-!                *rlv3 *ww0 *ww3
-!   rw status of diff variables: *xx:in *rev0:in-out *rev3:in *pp0:in-out
-!                *pp3:in *rlv0:in-out *rlv3:in *ww0:in-out *ww3:in
-!   plus diff mem management of: xx:in rev0:in rev3:in pp0:in pp3:in
-!                rlv0:in rlv3:in ww0:in ww3:in
+!   variations   of useful results: *rev0 *d2wall0 *pp0 *rlv0 *ww0
+!   with respect to varying inputs: *xx *rev0 *d2wall0 *rev3 *d2wall3
+!                *pp0 *pp3 *rlv0 *rlv3 *ww0 *ww3
+!   rw status of diff variables: *xx:in *rev0:in-out *d2wall0:in-out
+!                *rev3:in *d2wall3:in *pp0:in-out *pp3:in *rlv0:in-out
+!                *rlv3:in *ww0:in-out *ww3:in
+!   plus diff mem management of: xx:in rev0:in d2wall0:in rev3:in
+!                d2wall3:in pp0:in pp3:in rlv0:in rlv3:in ww0:in
+!                ww3:in
   subroutine bcsymmpolar2ndhalo_d(nn)
 ! bcsymmpolar applies the polar symmetry boundary conditions to a
 ! singular line of a block. it is assumed that the pointers in
@@ -611,9 +635,11 @@ subroutine bcsymmpolar2ndhalo_d(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 2nd level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : ww0, ww0d, ww3, ww3d, pp0, pp0d, pp3, pp3d,&
-&   rlv0, rlv0d, rlv3, rlv3d, rev0, rev0d, rev3, rev3d, xx, xxd, istart,&
-&   jstart, isize, jsize
+&   rlv0, rlv0d, rlv3, rlv3d, rev0, rev0d, rev3, rev3d, d2wall0, &
+&   d2wall0d, d2wall3, d2wall3d, ks0, ks3, xx, xxd, istart, jstart, &
+&   isize, jsize
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -696,6 +722,9 @@ subroutine bcsymmpolar2ndhalo_d(nn)
 ! eddy viscosity in the halo.
       pp0d(i, j) = pp3d(i, j)
       pp0(i, j) = pp3(i, j)
+      d2wall0d(i, j) = d2wall3d(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) then
         rlv0d(i, j) = rlv3d(i, j)
         rlv0(i, j) = rlv3(i, j)
@@ -715,8 +744,9 @@ subroutine bcsymmpolar2ndhalo(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 2nd level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_d, only : ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3,&
-&   xx, istart, jstart, isize, jsize
+&   d2wall0, d2wall3, ks0, ks3, xx, istart, jstart, isize, jsize
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -770,6 +800,8 @@ subroutine bcsymmpolar2ndhalo(nn)
 ! set the pressure and possibly the laminar and
 ! eddy viscosity in the halo.
       pp0(i, j) = pp3(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) rlv0(i, j) = rlv3(i, j)
       if (eddymodel) rev0(i, j) = rev3(i, j)
     end do
diff --git a/src/adjoint/outputForward/sa_d.f90 b/src/adjoint/outputForward/sa_d.f90
index 7d076aa76..c97a02d0e 100644
--- a/src/adjoint/outputForward/sa_d.f90
+++ b/src/adjoint/outputForward/sa_d.f90
@@ -40,6 +40,8 @@ subroutine sasource_d()
     real(kind=realtype), parameter :: f23=two*third
 ! local variables.
     integer(kind=inttype) :: i, j, k, nn, ii
+    real(kind=realtype) :: distrough
+    real(kind=realtype) :: distroughd
     real(kind=realtype) :: fv1, fv2, ft2
     real(kind=realtype) :: fv1d, fv2d, ft2d
     real(kind=realtype) :: ss, sst, nu, dist2inv, chi, chi2, chi3
@@ -363,12 +365,22 @@ subroutine sasource_d()
             temp9 = rlv(i, j, k)/temp10
             nud = (rlvd(i, j, k)-temp9*wd(i, j, k, irho))/temp10
             nu = temp9
-            temp10 = one/(d2wall(i, j, k)*d2wall(i, j, k))
-            dist2invd = -(temp10*2*d2walld(i, j, k)/d2wall(i, j, k))
-            dist2inv = temp10
             temp10 = w(i, j, k, itu1)/nu
             chid = (wd(i, j, k, itu1)-temp10*nud)/nu
             chi = temp10
+            if (.not.useroughsa) then
+              temp10 = one/(d2wall(i, j, k)*d2wall(i, j, k))
+              dist2invd = -(temp10*2*d2walld(i, j, k)/d2wall(i, j, k))
+              dist2inv = temp10
+            else
+              distroughd = d2walld(i, j, k)
+              distrough = d2wall(i, j, k) + 0.03_realtype*ks(i, j, k)
+              temp10 = one/(distrough*distrough)
+              dist2invd = -(temp10*2*distroughd/distrough)
+              dist2inv = temp10
+              chid = chid - ks(i, j, k)*rsacr1*distroughd/distrough**2
+              chi = chi + rsacr1*ks(i, j, k)/distrough
+            end if
             chi2d = 2*chi*chid
             chi2 = chi*chi
             chi3d = chi2*chid + chi*chi2d
@@ -376,9 +388,17 @@ subroutine sasource_d()
             temp10 = chi3/(cv13+chi3)
             fv1d = (1.0-temp10)*chi3d/(cv13+chi3)
             fv1 = temp10
-            temp10 = chi/(one+chi*fv1)
-            fv2d = -((chid-temp10*(fv1*chid+chi*fv1d))/(one+chi*fv1))
-            fv2 = one - temp10
+            if (.not.useroughsa) then
+              temp10 = chi/(one+chi*fv1)
+              fv2d = -((chid-temp10*(fv1*chid+chi*fv1d))/(one+chi*fv1))
+              fv2 = one - temp10
+            else
+              temp10 = w(i, j, k, itu1)
+              temp9 = w(i, j, k, itu1)/(nu+temp10*fv1)
+              fv2d = -((wd(i, j, k, itu1)-temp9*(nud+fv1*wd(i, j, k, &
+&               itu1)+temp10*fv1d))/(nu+temp10*fv1))
+              fv2 = one - temp9
+            end if
 ! the function ft2, which is designed to keep a laminar
 ! solution laminar. when running in fully turbulent mode
 ! this function should be set to 0.0.
@@ -491,6 +511,7 @@ subroutine sasource()
     real(kind=realtype), parameter :: f23=two*third
 ! local variables.
     integer(kind=inttype) :: i, j, k, nn, ii
+    real(kind=realtype) :: distrough
     real(kind=realtype) :: fv1, fv2, ft2
     real(kind=realtype) :: ss, sst, nu, dist2inv, chi, chi2, chi3
     real(kind=realtype) :: rr, gg, gg6, termfw, fwsa, term1, term2
@@ -604,12 +625,22 @@ subroutine sasource()
 ! and nu) and the functions fv1 and fv2. the latter corrects
 ! the production term near a viscous wall.
             nu = rlv(i, j, k)/w(i, j, k, irho)
-            dist2inv = one/d2wall(i, j, k)**2
             chi = w(i, j, k, itu1)/nu
+            if (.not.useroughsa) then
+              dist2inv = one/d2wall(i, j, k)**2
+            else
+              distrough = d2wall(i, j, k) + 0.03_realtype*ks(i, j, k)
+              dist2inv = one/distrough**2
+              chi = chi + rsacr1*ks(i, j, k)/distrough
+            end if
             chi2 = chi*chi
             chi3 = chi*chi2
             fv1 = chi3/(chi3+cv13)
-            fv2 = one - chi/(one+chi*fv1)
+            if (.not.useroughsa) then
+              fv2 = one - chi/(one+chi*fv1)
+            else
+              fv2 = one - w(i, j, k, itu1)/(nu+w(i, j, k, itu1)*fv1)
+            end if
 ! the function ft2, which is designed to keep a laminar
 ! solution laminar. when running in fully turbulent mode
 ! this function should be set to 0.0.
diff --git a/src/adjoint/outputForward/turbBCRoutines_d.f90 b/src/adjoint/outputForward/turbBCRoutines_d.f90
index 6f41ad314..758cddfd9 100644
--- a/src/adjoint/outputForward/turbBCRoutines_d.f90
+++ b/src/adjoint/outputForward/turbBCRoutines_d.f90
@@ -7,12 +7,16 @@ module turbbcroutines_d
 contains
 !  differentiation of applyallturbbcthisblock in forward (tangent) mode (with options i4 dr8 r8):
 !   variations   of useful results: *rev *w
-!   with respect to varying inputs: *rev *bvtj1 *bvtj2 *w *bvtk1
-!                *bvtk2 *bvti1 *bvti2
+!   with respect to varying inputs: *rev *bvtj1 *bvtj2 *w *bmtk1
+!                *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2 *bvti1
+!                *bvti2 *bmtj1 *bmtj2
 !   rw status of diff variables: *rev:in-out *bvtj1:in *bvtj2:in
-!                *w:in-out *bvtk1:in *bvtk2:in *bvti1:in *bvti2:in
+!                *w:in-out *bmtk1:in *bmtk2:in *bvtk1:in *bvtk2:in
+!                *d2wall:in *bmti1:in *bmti2:in *bvti1:in *bvti2:in
+!                *bmtj1:in *bmtj2:in
 !   plus diff mem management of: rev:in bvtj1:in bvtj2:in w:in
-!                bvtk1:in bvtk2:in bvti1:in bvti2:in
+!                bmtk1:in bmtk2:in bvtk1:in bvtk2:in d2wall:in
+!                bmti1:in bmti2:in bvti1:in bvti2:in bmtj1:in bmtj2:in
 !      ==================================================================
   subroutine applyallturbbcthisblock_d(secondhalo)
 !
@@ -36,6 +40,7 @@ subroutine applyallturbbcthisblock_d(secondhalo)
     real(kind=realtype), dimension(:, :, :, :), pointer :: bmt
     real(kind=realtype), dimension(:, :, :), pointer :: bvt, ww1, ww2
     real(kind=realtype) :: temp
+    real(kind=realtype) :: temp0
 ! loop over the boundary condition subfaces of this block.
 bocos:do nn=1,nbocos
 ! loop over the faces and set the state in
@@ -49,9 +54,11 @@ subroutine applyallturbbcthisblock_d(secondhalo)
                 wd(1, i, j, l) = bvti1d(i, j, l)
                 w(1, i, j, l) = bvti1(i, j, l)
                 do m=nt1,nt2
-                  temp = bmti1(i, j, l, m)
-                  wd(1, i, j, l) = wd(1, i, j, l) - temp*wd(2, i, j, m)
-                  w(1, i, j, l) = w(1, i, j, l) - temp*w(2, i, j, m)
+                  temp = w(2, i, j, m)
+                  temp0 = bmti1(i, j, l, m)
+                  wd(1, i, j, l) = wd(1, i, j, l) - temp*bmti1d(i, j, l&
+&                   , m) - temp0*wd(2, i, j, m)
+                  w(1, i, j, l) = w(1, i, j, l) - temp0*temp
                 end do
               end do
             end do
@@ -63,10 +70,11 @@ subroutine applyallturbbcthisblock_d(secondhalo)
                 wd(ie, i, j, l) = bvti2d(i, j, l)
                 w(ie, i, j, l) = bvti2(i, j, l)
                 do m=nt1,nt2
+                  temp0 = w(il, i, j, m)
                   temp = bmti2(i, j, l, m)
-                  wd(ie, i, j, l) = wd(ie, i, j, l) - temp*wd(il, i, j, &
-&                   m)
-                  w(ie, i, j, l) = w(ie, i, j, l) - temp*w(il, i, j, m)
+                  wd(ie, i, j, l) = wd(ie, i, j, l) - temp0*bmti2d(i, j&
+&                   , l, m) - temp*wd(il, i, j, m)
+                  w(ie, i, j, l) = w(ie, i, j, l) - temp*temp0
                 end do
               end do
             end do
@@ -78,9 +86,11 @@ subroutine applyallturbbcthisblock_d(secondhalo)
                 wd(i, 1, j, l) = bvtj1d(i, j, l)
                 w(i, 1, j, l) = bvtj1(i, j, l)
                 do m=nt1,nt2
+                  temp0 = w(i, 2, j, m)
                   temp = bmtj1(i, j, l, m)
-                  wd(i, 1, j, l) = wd(i, 1, j, l) - temp*wd(i, 2, j, m)
-                  w(i, 1, j, l) = w(i, 1, j, l) - temp*w(i, 2, j, m)
+                  wd(i, 1, j, l) = wd(i, 1, j, l) - temp0*bmtj1d(i, j, l&
+&                   , m) - temp*wd(i, 2, j, m)
+                  w(i, 1, j, l) = w(i, 1, j, l) - temp*temp0
                 end do
               end do
             end do
@@ -92,10 +102,11 @@ subroutine applyallturbbcthisblock_d(secondhalo)
                 wd(i, je, j, l) = bvtj2d(i, j, l)
                 w(i, je, j, l) = bvtj2(i, j, l)
                 do m=nt1,nt2
+                  temp0 = w(i, jl, j, m)
                   temp = bmtj2(i, j, l, m)
-                  wd(i, je, j, l) = wd(i, je, j, l) - temp*wd(i, jl, j, &
-&                   m)
-                  w(i, je, j, l) = w(i, je, j, l) - temp*w(i, jl, j, m)
+                  wd(i, je, j, l) = wd(i, je, j, l) - temp0*bmtj2d(i, j&
+&                   , l, m) - temp*wd(i, jl, j, m)
+                  w(i, je, j, l) = w(i, je, j, l) - temp*temp0
                 end do
               end do
             end do
@@ -107,9 +118,11 @@ subroutine applyallturbbcthisblock_d(secondhalo)
                 wd(i, j, 1, l) = bvtk1d(i, j, l)
                 w(i, j, 1, l) = bvtk1(i, j, l)
                 do m=nt1,nt2
+                  temp0 = w(i, j, 2, m)
                   temp = bmtk1(i, j, l, m)
-                  wd(i, j, 1, l) = wd(i, j, 1, l) - temp*wd(i, j, 2, m)
-                  w(i, j, 1, l) = w(i, j, 1, l) - temp*w(i, j, 2, m)
+                  wd(i, j, 1, l) = wd(i, j, 1, l) - temp0*bmtk1d(i, j, l&
+&                   , m) - temp*wd(i, j, 2, m)
+                  w(i, j, 1, l) = w(i, j, 1, l) - temp*temp0
                 end do
               end do
             end do
@@ -121,10 +134,11 @@ subroutine applyallturbbcthisblock_d(secondhalo)
                 wd(i, j, ke, l) = bvtk2d(i, j, l)
                 w(i, j, ke, l) = bvtk2(i, j, l)
                 do m=nt1,nt2
+                  temp0 = w(i, j, kl, m)
                   temp = bmtk2(i, j, l, m)
-                  wd(i, j, ke, l) = wd(i, j, ke, l) - temp*wd(i, j, kl, &
-&                   m)
-                  w(i, j, ke, l) = w(i, j, ke, l) - temp*w(i, j, kl, m)
+                  wd(i, j, ke, l) = wd(i, j, ke, l) - temp0*bmtk2d(i, j&
+&                   , l, m) - temp*wd(i, j, kl, m)
+                  w(i, j, ke, l) = w(i, j, ke, l) - temp*temp0
                 end do
               end do
             end do
@@ -407,8 +421,8 @@ end subroutine bceddynowall
 
 !  differentiation of bceddywall in forward (tangent) mode (with options i4 dr8 r8):
 !   variations   of useful results: *rev
-!   with respect to varying inputs: *rev
-!   plus diff mem management of: rev:in
+!   with respect to varying inputs: *rev *d2wall
+!   plus diff mem management of: rev:in d2wall:in
   subroutine bceddywall_d(nn)
 !
 !       bceddywall sets the eddy viscosity in the halo cells of
@@ -427,6 +441,8 @@ subroutine bceddywall_d(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j
+    real(kind=realtype) :: fact
+    real(kind=realtype) :: factd
 ! determine the face id on which the subface is located and
 ! loop over the faces of the subface and set the eddy viscosity
 ! in the halo cells.
@@ -434,43 +450,49 @@ subroutine bceddywall_d(nn)
     case (imin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          revd(1, i, j) = -revd(2, i, j)
-          rev(1, i, j) = -rev(2, i, j)
+          call saroughfact_d(2, i, j, fact, factd)
+          revd(1, i, j) = rev(2, i, j)*factd + fact*revd(2, i, j)
+          rev(1, i, j) = fact*rev(2, i, j)
         end do
       end do
     case (imax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          revd(ie, i, j) = -revd(il, i, j)
-          rev(ie, i, j) = -rev(il, i, j)
+          call saroughfact_d(il, i, j, fact, factd)
+          revd(ie, i, j) = rev(il, i, j)*factd + fact*revd(il, i, j)
+          rev(ie, i, j) = fact*rev(il, i, j)
         end do
       end do
     case (jmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          revd(i, 1, j) = -revd(i, 2, j)
-          rev(i, 1, j) = -rev(i, 2, j)
+          call saroughfact_d(i, 2, j, fact, factd)
+          revd(i, 1, j) = rev(i, 2, j)*factd + fact*revd(i, 2, j)
+          rev(i, 1, j) = fact*rev(i, 2, j)
         end do
       end do
     case (jmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          revd(i, je, j) = -revd(i, jl, j)
-          rev(i, je, j) = -rev(i, jl, j)
+          call saroughfact_d(i, jl, j, fact, factd)
+          revd(i, je, j) = rev(i, jl, j)*factd + fact*revd(i, jl, j)
+          rev(i, je, j) = fact*rev(i, jl, j)
         end do
       end do
     case (kmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          revd(i, j, 1) = -revd(i, j, 2)
-          rev(i, j, 1) = -rev(i, j, 2)
+          call saroughfact_d(i, j, 2, fact, factd)
+          revd(i, j, 1) = rev(i, j, 2)*factd + fact*revd(i, j, 2)
+          rev(i, j, 1) = fact*rev(i, j, 2)
         end do
       end do
     case (kmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          revd(i, j, ke) = -revd(i, j, kl)
-          rev(i, j, ke) = -rev(i, j, kl)
+          call saroughfact_d(i, j, kl, fact, factd)
+          revd(i, j, ke) = rev(i, j, kl)*factd + fact*revd(i, j, kl)
+          rev(i, j, ke) = fact*rev(i, j, kl)
         end do
       end do
     end select
@@ -494,6 +516,7 @@ subroutine bceddywall(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j
+    real(kind=realtype) :: fact
 ! determine the face id on which the subface is located and
 ! loop over the faces of the subface and set the eddy viscosity
 ! in the halo cells.
@@ -501,49 +524,58 @@ subroutine bceddywall(nn)
     case (imin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(1, i, j) = -rev(2, i, j)
+          call saroughfact(2, i, j, fact)
+          rev(1, i, j) = fact*rev(2, i, j)
         end do
       end do
     case (imax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(ie, i, j) = -rev(il, i, j)
+          call saroughfact(il, i, j, fact)
+          rev(ie, i, j) = fact*rev(il, i, j)
         end do
       end do
     case (jmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, 1, j) = -rev(i, 2, j)
+          call saroughfact(i, 2, j, fact)
+          rev(i, 1, j) = fact*rev(i, 2, j)
         end do
       end do
     case (jmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, je, j) = -rev(i, jl, j)
+          call saroughfact(i, jl, j, fact)
+          rev(i, je, j) = fact*rev(i, jl, j)
         end do
       end do
     case (kmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, j, 1) = -rev(i, j, 2)
+          call saroughfact(i, j, 2, fact)
+          rev(i, j, 1) = fact*rev(i, j, 2)
         end do
       end do
     case (kmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, j, ke) = -rev(i, j, kl)
+          call saroughfact(i, j, kl, fact)
+          rev(i, j, ke) = fact*rev(i, j, kl)
         end do
       end do
     end select
   end subroutine bceddywall
 
 !  differentiation of bcturbfarfield in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *bvtj1 *bvtj2 *bvtk1 *bvtk2
-!                *bvti1 *bvti2
-!   with respect to varying inputs: winf *bvtj1 *bvtj2 *bvtk1 *bvtk2
-!                *bvti1 *bvti2
-!   plus diff mem management of: bvtj1:in bvtj2:in bvtk1:in bvtk2:in
-!                bvti1:in bvti2:in
+!   variations   of useful results: *bvtj1 *bvtj2 *bmtk1 *bmtk2
+!                *bvtk1 *bvtk2 *bmti1 *bmti2 *bvti1 *bvti2 *bmtj1
+!                *bmtj2
+!   with respect to varying inputs: winf *bvtj1 *bvtj2 *bmtk1 *bmtk2
+!                *bvtk1 *bvtk2 *bmti1 *bmti2 *bvti1 *bvti2 *bmtj1
+!                *bmtj2
+!   plus diff mem management of: bvtj1:in bvtj2:in bmtk1:in bmtk2:in
+!                bvtk1:in bvtk2:in bmti1:in bmti2:in bvti1:in bvti2:in
+!                bmtj1:in bmtj2:in
   subroutine bcturbfarfield_d(nn)
 !
 !       bcturbfarfield applies the implicit treatment of the
@@ -584,16 +616,22 @@ subroutine bcturbfarfield_d(nn)
           do l=nt1,nt2
             select case  (bcfaceid(nn)) 
             case (imin) 
+              bmti1d(i, j, l, l) = 0.0_8
               bmti1(i, j, l, l) = -one
             case (imax) 
+              bmti2d(i, j, l, l) = 0.0_8
               bmti2(i, j, l, l) = -one
             case (jmin) 
+              bmtj1d(i, j, l, l) = 0.0_8
               bmtj1(i, j, l, l) = -one
             case (jmax) 
+              bmtj2d(i, j, l, l) = 0.0_8
               bmtj2(i, j, l, l) = -one
             case (kmin) 
+              bmtk1d(i, j, l, l) = 0.0_8
               bmtk1(i, j, l, l) = -one
             case (kmax) 
+              bmtk2d(i, j, l, l) = 0.0_8
               bmtk2(i, j, l, l) = -one
             end select
           end do
@@ -908,6 +946,65 @@ subroutine bcturboutflow(nn)
     end do
   end subroutine bcturboutflow
 
+!  differentiation of bcturbsymm in forward (tangent) mode (with options i4 dr8 r8):
+!   variations   of useful results: *bmtk1 *bmtk2 *bmti1 *bmti2
+!                *bmtj1 *bmtj2
+!   with respect to varying inputs: *bmtk1 *bmtk2 *bmti1 *bmti2
+!                *bmtj1 *bmtj2
+!   plus diff mem management of: bmtk1:in bmtk2:in bmti1:in bmti2:in
+!                bmtj1:in bmtj2:in
+  subroutine bcturbsymm_d(nn)
+!
+!       bcturbsymm applies the implicit treatment of the symmetry
+!       boundary condition (or inviscid wall) to subface nn. as the
+!       symmetry boundary condition is independent of the turbulence
+!       model, this routine is valid for all models. it is assumed
+!       that the pointers in blockpointers are already set to the
+!       correct block on the correct grid level.
+!
+    use constants
+    use blockpointers
+    use flowvarrefstate
+    implicit none
+!
+!      subroutine arguments.
+!
+    integer(kind=inttype), intent(in) :: nn
+!
+!      local variables.
+!
+    integer(kind=inttype) :: i, j, l
+! loop over the faces of the subfaces and set the values of bmt
+! for an implicit treatment. for a symmetry face this means
+! that the halo value is set to the internal value.
+    do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+      do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+        do l=nt1,nt2
+          select case  (bcfaceid(nn)) 
+          case (imin) 
+            bmti1d(i, j, l, l) = 0.0_8
+            bmti1(i, j, l, l) = -one
+          case (imax) 
+            bmti2d(i, j, l, l) = 0.0_8
+            bmti2(i, j, l, l) = -one
+          case (jmin) 
+            bmtj1d(i, j, l, l) = 0.0_8
+            bmtj1(i, j, l, l) = -one
+          case (jmax) 
+            bmtj2d(i, j, l, l) = 0.0_8
+            bmtj2(i, j, l, l) = -one
+          case (kmin) 
+            bmtk1d(i, j, l, l) = 0.0_8
+            bmtk1(i, j, l, l) = -one
+          case (kmax) 
+            bmtk2d(i, j, l, l) = 0.0_8
+            bmtk2(i, j, l, l) = -one
+          end select
+        end do
+      end do
+    end do
+  end subroutine bcturbsymm_d
+
   subroutine bcturbsymm(nn)
 !
 !       bcturbsymm applies the implicit treatment of the symmetry
@@ -955,14 +1052,19 @@ subroutine bcturbsymm(nn)
   end subroutine bcturbsymm
 
 !  differentiation of bcturbtreatment in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *bvtj1 *bvtj2 *bvtk1 *bvtk2
-!                *bvti1 *bvti2
-!   with respect to varying inputs: winf *w *rlv *d2wall
-!   rw status of diff variables: winf:in *bvtj1:out *bvtj2:out
-!                *w:in *rlv:in *bvtk1:out *bvtk2:out *d2wall:in
-!                *bvti1:out *bvti2:out
-!   plus diff mem management of: bvtj1:in bvtj2:in w:in rlv:in
-!                bvtk1:in bvtk2:in d2wall:in bvti1:in bvti2:in
+!   variations   of useful results: *bvtj1 *bvtj2 *bmtk1 *bmtk2
+!                *bvtk1 *bvtk2 *bmti1 *bmti2 *bvti1 *bvti2 *bmtj1
+!                *bmtj2
+!   with respect to varying inputs: winf *bvtj1 *bvtj2 *w *bmtk1
+!                *rlv *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2
+!                *bvti1 *bvti2 *bmtj1 *bmtj2
+!   rw status of diff variables: winf:in *bvtj1:in-out *bvtj2:in-out
+!                *w:in *bmtk1:in-out *rlv:in *bmtk2:in-out *bvtk1:in-out
+!                *bvtk2:in-out *d2wall:in *bmti1:in-out *bmti2:in-out
+!                *bvti1:in-out *bvti2:in-out *bmtj1:in-out *bmtj2:in-out
+!   plus diff mem management of: bvtj1:in bvtj2:in w:in bmtk1:in
+!                rlv:in bmtk2:in bvtk1:in bvtk2:in d2wall:in bmti1:in
+!                bmti2:in bvti1:in bvti2:in bmtj1:in bmtj2:in
   subroutine bcturbtreatment_d()
 !
 !       bcturbtreatment sets the arrays bmti1, bvti1, etc, such that
@@ -991,7 +1093,9 @@ subroutine bcturbtreatment_d()
       do j=1,je
         do l=nt1,nt2
           do m=nt1,nt2
+            bmti1d(j, k, l, m) = 0.0_8
             bmti1(j, k, l, m) = zero
+            bmti2d(j, k, l, m) = 0.0_8
             bmti2(j, k, l, m) = zero
           end do
           bvti1d(j, k, l) = 0.0_8
@@ -1005,7 +1109,9 @@ subroutine bcturbtreatment_d()
       do i=1,ie
         do l=nt1,nt2
           do m=nt1,nt2
+            bmtj1d(i, k, l, m) = 0.0_8
             bmtj1(i, k, l, m) = zero
+            bmtj2d(i, k, l, m) = 0.0_8
             bmtj2(i, k, l, m) = zero
           end do
           bvtj1d(i, k, l) = 0.0_8
@@ -1019,7 +1125,9 @@ subroutine bcturbtreatment_d()
       do i=1,ie
         do l=nt1,nt2
           do m=nt1,nt2
+            bmtk1d(i, j, l, m) = 0.0_8
             bmtk1(i, j, l, m) = zero
+            bmtk2d(i, j, l, m) = 0.0_8
             bmtk2(i, j, l, m) = zero
           end do
           bvtk1d(i, j, l) = 0.0_8
@@ -1029,12 +1137,6 @@ subroutine bcturbtreatment_d()
         end do
       end do
     end do
-    if (associated(bvtj1d)) bvtj1d = 0.0_8
-    if (associated(bvtj2d)) bvtj2d = 0.0_8
-    if (associated(bvtk1d)) bvtk1d = 0.0_8
-    if (associated(bvtk2d)) bvtk2d = 0.0_8
-    if (associated(bvti1d)) bvti1d = 0.0_8
-    if (associated(bvti2d)) bvti2d = 0.0_8
 ! loop over the boundary condition subfaces of this block.
 bocos:do nn=1,nbocos
 ! determine the kind of boundary condition for this subface.
@@ -1049,7 +1151,7 @@ subroutine bcturbtreatment_d()
       case (symm, symmpolar, eulerwall) 
 ! symmetry, polar symmetry or inviscid wall. treatment of
 ! the turbulent equations is identical.
-        call bcturbsymm(nn)
+        call bcturbsymm_d(nn)
 !=============================================================
       case (farfield) 
 ! farfield. the kind of boundary condition to be applied,
@@ -1160,12 +1262,15 @@ subroutine bcturbtreatment()
   end subroutine bcturbtreatment
 
 !  differentiation of bcturbwall in forward (tangent) mode (with options i4 dr8 r8):
-!   variations   of useful results: *bvtj1 *bvtj2 *bvtk1 *bvtk2
-!                *bvti1 *bvti2
-!   with respect to varying inputs: *bvtj1 *bvtj2 *w *rlv *bvtk1
-!                *bvtk2 *d2wall *bvti1 *bvti2
-!   plus diff mem management of: bvtj1:in bvtj2:in w:in rlv:in
-!                bvtk1:in bvtk2:in d2wall:in bvti1:in bvti2:in
+!   variations   of useful results: *bvtj1 *bvtj2 *bmtk1 *bmtk2
+!                *bvtk1 *bvtk2 *bmti1 *bmti2 *bvti1 *bvti2 *bmtj1
+!                *bmtj2
+!   with respect to varying inputs: *bvtj1 *bvtj2 *w *bmtk1 *rlv
+!                *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2 *bvti1
+!                *bvti2 *bmtj1 *bmtj2
+!   plus diff mem management of: bvtj1:in bvtj2:in w:in bmtk1:in
+!                rlv:in bmtk2:in bvtk1:in bvtk2:in d2wall:in bmti1:in
+!                bmti2:in bvti1:in bvti2:in bmtj1:in bmtj2:in
   subroutine bcturbwall_d(nn)
 !
 !       bcturbwall applies the implicit treatment of the viscous
@@ -1188,8 +1293,8 @@ subroutine bcturbwall_d(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j, ii, jj, iimax, jjmax
-    real(kind=realtype) :: tmpd, tmpe, tmpf, nu
-    real(kind=realtype) :: tmpdd, nud
+    real(kind=realtype) :: tmpd, tmpe, tmpf, nu, fact
+    real(kind=realtype) :: tmpdd, nud, factd
     real(kind=realtype), dimension(:, :, :, :), pointer :: bmt
     real(kind=realtype), dimension(:, :, :), pointer :: bvt, ww2
     real(kind=realtype), dimension(:, :), pointer :: rlv2, dd2wall
@@ -1220,37 +1325,49 @@ subroutine bcturbwall_d(nn)
       case (imin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti1(i, j, itu1, itu1) = one
+            call saroughfact_d(2, i, j, fact, factd)
+            bmti1d(i, j, itu1, itu1) = -factd
+            bmti1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (imax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti2(i, j, itu1, itu1) = one
+            call saroughfact_d(il, i, j, fact, factd)
+            bmti2d(i, j, itu1, itu1) = -factd
+            bmti2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj1(i, j, itu1, itu1) = one
+            call saroughfact_d(i, 2, j, fact, factd)
+            bmtj1d(i, j, itu1, itu1) = -factd
+            bmtj1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj2(i, j, itu1, itu1) = one
+            call saroughfact_d(i, jl, j, fact, factd)
+            bmtj2d(i, j, itu1, itu1) = -factd
+            bmtj2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk1(i, j, itu1, itu1) = one
+            call saroughfact_d(i, j, 2, fact, factd)
+            bmtk1d(i, j, itu1, itu1) = -factd
+            bmtk1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk2(i, j, itu1, itu1) = one
+            call saroughfact_d(i, j, kl, fact, factd)
+            bmtk2d(i, j, itu1, itu1) = -factd
+            bmtk2(i, j, itu1, itu1) = -fact
           end do
         end do
       end select
@@ -1299,7 +1416,9 @@ subroutine bcturbwall_d(nn)
             temp0 = one/(rkwbeta1*(d2wall(2, ii, jj)*d2wall(2, ii, jj)))
             tmpdd = -(temp0*2*d2walld(2, ii, jj)/d2wall(2, ii, jj))
             tmpd = temp0
+            bmti1d(i, j, itu1, itu1) = 0.0_8
             bmti1(i, j, itu1, itu1) = one
+            bmti1d(i, j, itu2, itu2) = 0.0_8
             bmti1(i, j, itu2, itu2) = one
             bvti1d(i, j, itu2) = two*60.0_realtype*(tmpd*nud+nu*tmpdd)
             bvti1(i, j, itu2) = two*60.0_realtype*nu*tmpd
@@ -1338,7 +1457,9 @@ subroutine bcturbwall_d(nn)
 &             ))
             tmpdd = -(temp0*2*d2walld(il, ii, jj)/d2wall(il, ii, jj))
             tmpd = temp0
+            bmti2d(i, j, itu1, itu1) = 0.0_8
             bmti2(i, j, itu1, itu1) = one
+            bmti2d(i, j, itu2, itu2) = 0.0_8
             bmti2(i, j, itu2, itu2) = one
             bvti2d(i, j, itu2) = two*60.0_realtype*(tmpd*nud+nu*tmpdd)
             bvti2(i, j, itu2) = two*60.0_realtype*nu*tmpd
@@ -1376,7 +1497,9 @@ subroutine bcturbwall_d(nn)
             temp0 = one/(rkwbeta1*(d2wall(ii, 2, jj)*d2wall(ii, 2, jj)))
             tmpdd = -(temp0*2*d2walld(ii, 2, jj)/d2wall(ii, 2, jj))
             tmpd = temp0
+            bmtj1d(i, j, itu1, itu1) = 0.0_8
             bmtj1(i, j, itu1, itu1) = one
+            bmtj1d(i, j, itu2, itu2) = 0.0_8
             bmtj1(i, j, itu2, itu2) = one
             bvtj1d(i, j, itu2) = two*60.0_realtype*(tmpd*nud+nu*tmpdd)
             bvtj1(i, j, itu2) = two*60.0_realtype*nu*tmpd
@@ -1415,7 +1538,9 @@ subroutine bcturbwall_d(nn)
 &             ))
             tmpdd = -(temp0*2*d2walld(ii, jl, jj)/d2wall(ii, jl, jj))
             tmpd = temp0
+            bmtj2d(i, j, itu1, itu1) = 0.0_8
             bmtj2(i, j, itu1, itu1) = one
+            bmtj2d(i, j, itu2, itu2) = 0.0_8
             bmtj2(i, j, itu2, itu2) = one
             bvtj2d(i, j, itu2) = two*60.0_realtype*(tmpd*nud+nu*tmpdd)
             bvtj2(i, j, itu2) = two*60.0_realtype*nu*tmpd
@@ -1453,7 +1578,9 @@ subroutine bcturbwall_d(nn)
             temp0 = one/(rkwbeta1*(d2wall(ii, jj, 2)*d2wall(ii, jj, 2)))
             tmpdd = -(temp0*2*d2walld(ii, jj, 2)/d2wall(ii, jj, 2))
             tmpd = temp0
+            bmtk1d(i, j, itu1, itu1) = 0.0_8
             bmtk1(i, j, itu1, itu1) = one
+            bmtk1d(i, j, itu2, itu2) = 0.0_8
             bmtk1(i, j, itu2, itu2) = one
             bvtk1d(i, j, itu2) = two*60.0_realtype*(tmpd*nud+nu*tmpdd)
             bvtk1(i, j, itu2) = two*60.0_realtype*nu*tmpd
@@ -1492,7 +1619,9 @@ subroutine bcturbwall_d(nn)
 &             ))
             tmpdd = -(temp0*2*d2walld(ii, jj, kl)/d2wall(ii, jj, kl))
             tmpd = temp0
+            bmtk2d(i, j, itu1, itu1) = 0.0_8
             bmtk2(i, j, itu1, itu1) = one
+            bmtk2d(i, j, itu2, itu2) = 0.0_8
             bmtk2(i, j, itu2, itu2) = one
             bvtk2d(i, j, itu2) = two*60.0_realtype*(tmpd*nud+nu*tmpdd)
             bvtk2(i, j, itu2) = two*60.0_realtype*nu*tmpd
@@ -1507,42 +1636,54 @@ subroutine bcturbwall_d(nn)
       case (imin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+            bmti1d(i, j, itu1, itu1) = 0.0_8
             bmti1(i, j, itu1, itu1) = one
+            bmti1d(i, j, itu2, itu2) = 0.0_8
             bmti1(i, j, itu2, itu2) = one
           end do
         end do
       case (imax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+            bmti2d(i, j, itu1, itu1) = 0.0_8
             bmti2(i, j, itu1, itu1) = one
+            bmti2d(i, j, itu2, itu2) = 0.0_8
             bmti2(i, j, itu2, itu2) = one
           end do
         end do
       case (jmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+            bmtj1d(i, j, itu1, itu1) = 0.0_8
             bmtj1(i, j, itu1, itu1) = one
+            bmtj1d(i, j, itu2, itu2) = 0.0_8
             bmtj1(i, j, itu2, itu2) = one
           end do
         end do
       case (jmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+            bmtj2d(i, j, itu1, itu1) = 0.0_8
             bmtj2(i, j, itu1, itu1) = one
+            bmtj2d(i, j, itu2, itu2) = 0.0_8
             bmtj2(i, j, itu2, itu2) = one
           end do
         end do
       case (kmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+            bmtk1d(i, j, itu1, itu1) = 0.0_8
             bmtk1(i, j, itu1, itu1) = one
+            bmtk1d(i, j, itu2, itu2) = 0.0_8
             bmtk1(i, j, itu2, itu2) = one
           end do
         end do
       case (kmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+            bmtk2d(i, j, itu1, itu1) = 0.0_8
             bmtk2(i, j, itu1, itu1) = one
+            bmtk2d(i, j, itu2, itu2) = 0.0_8
             bmtk2(i, j, itu2, itu2) = one
           end do
         end do
@@ -1573,7 +1714,7 @@ subroutine bcturbwall(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j, ii, jj, iimax, jjmax
-    real(kind=realtype) :: tmpd, tmpe, tmpf, nu
+    real(kind=realtype) :: tmpd, tmpe, tmpf, nu, fact
     real(kind=realtype), dimension(:, :, :, :), pointer :: bmt
     real(kind=realtype), dimension(:, :, :), pointer :: bvt, ww2
     real(kind=realtype), dimension(:, :), pointer :: rlv2, dd2wall
@@ -1602,37 +1743,43 @@ subroutine bcturbwall(nn)
       case (imin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti1(i, j, itu1, itu1) = one
+            call saroughfact(2, i, j, fact)
+            bmti1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (imax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti2(i, j, itu1, itu1) = one
+            call saroughfact(il, i, j, fact)
+            bmti2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj1(i, j, itu1, itu1) = one
+            call saroughfact(i, 2, j, fact)
+            bmtj1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj2(i, j, itu1, itu1) = one
+            call saroughfact(i, jl, j, fact)
+            bmtj2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk1(i, j, itu1, itu1) = one
+            call saroughfact(i, j, 2, fact)
+            bmtk1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk2(i, j, itu1, itu1) = one
+            call saroughfact(i, j, kl, fact)
+            bmtk2(i, j, itu1, itu1) = -fact
           end do
         end do
       end select
@@ -2224,5 +2371,68 @@ subroutine turbbcnswall(secondhalo)
     end do bocos
   end subroutine turbbcnswall
 
+!  differentiation of saroughfact in forward (tangent) mode (with options i4 dr8 r8):
+!   variations   of useful results: fact
+!   with respect to varying inputs: *d2wall
+!   plus diff mem management of: d2wall:in
+  subroutine saroughfact_d(i, j, k, fact, factd)
+! returns either the regular sa-boundary condition
+! or the modified roughness-boundary condition
+    use constants
+    use inputphysics, only : useroughsa
+    use blockpointers, only : ks, d2wall, d2walld, il, jl, kl
+    implicit none
+! local variablse
+    integer(kind=inttype), intent(in) :: i, j, k
+    real(kind=realtype), intent(out) :: fact
+    real(kind=realtype), intent(out) :: factd
+    real(kind=realtype) :: temp
+    real(kind=realtype) :: temp0
+    if (.not.useroughsa) then
+      fact = -one
+      factd = 0.0_8
+      return
+    else
+      temp = ks(i, j, k) + d2wall(i, j, k)/0.03_realtype
+      temp0 = (ks(i, j, k)-d2wall(i, j, k)/0.03_realtype)/temp
+      factd = -((1.0/0.03_realtype+temp0/0.03_realtype)*d2walld(i, j, k)&
+&       /temp)
+      fact = temp0
+      if (ks(i, j, k) .eq. 0.01 .or. d2wall(i, j, k) .eq. 0.01) print*, &
+&                                                               i, j, k&
+&                                                               , fact, &
+&                                                               d2wall(i&
+&                                                               , j, k)&
+&                                                               , ks(i, &
+&                                                               j, k)
+    end if
+  end subroutine saroughfact_d
+
+  subroutine saroughfact(i, j, k, fact)
+! returns either the regular sa-boundary condition
+! or the modified roughness-boundary condition
+    use constants
+    use inputphysics, only : useroughsa
+    use blockpointers, only : ks, d2wall, il, jl, kl
+    implicit none
+! local variablse
+    integer(kind=inttype), intent(in) :: i, j, k
+    real(kind=realtype), intent(out) :: fact
+    if (.not.useroughsa) then
+      fact = -one
+      return
+    else
+      fact = (ks(i, j, k)-d2wall(i, j, k)/0.03_realtype)/(ks(i, j, k)+&
+&       d2wall(i, j, k)/0.03_realtype)
+      if (ks(i, j, k) .eq. 0.01 .or. d2wall(i, j, k) .eq. 0.01) print*, &
+&                                                               i, j, k&
+&                                                               , fact, &
+&                                                               d2wall(i&
+&                                                               , j, k)&
+&                                                               , ks(i, &
+&                                                               j, k)
+    end if
+  end subroutine saroughfact
+
 end module turbbcroutines_d
 
diff --git a/src/adjoint/outputForward/turbUtils_d.f90 b/src/adjoint/outputForward/turbUtils_d.f90
index 274215a60..d2cb16708 100644
--- a/src/adjoint/outputForward/turbUtils_d.f90
+++ b/src/adjoint/outputForward/turbUtils_d.f90
@@ -884,8 +884,9 @@ subroutine turbadvection_d(madv, nadv, offset, qq)
     use constants
     use blockpointers, only : nx, ny, nz, il, jl, kl, vol, vold, &
 &   sfacei, sfaceid, sfacej, sfacejd, sfacek, sfacekd, w, wd, si, sid, &
-&   sj, sjd, sk, skd, addgridvelocities, bmti1, bmti2, bmtj1, bmtj2, &
-&   bmtk1, bmtk2, scratch, scratchd
+&   sj, sjd, sk, skd, addgridvelocities, bmti1, bmti1d, bmti2, bmti2d, &
+&   bmtj1, bmtj1d, bmtj2, bmtj2d, bmtk1, bmtk1d, bmtk2, bmtk2d, scratch,&
+&   scratchd
     use inputdiscretization, only : orderturb
     use iteration, only : groundlevel
     use turbmod, only : secondord
diff --git a/src/adjoint/outputForward/utils_d.f90 b/src/adjoint/outputForward/utils_d.f90
index ac9f6f0e4..09821228d 100644
--- a/src/adjoint/outputForward/utils_d.f90
+++ b/src/adjoint/outputForward/utils_d.f90
@@ -732,12 +732,13 @@ subroutine setbcpointers(nn, spatialpointers)
     use blockpointers, only : w, p, rlv, rev, gamma, x, d2wall, si, sj&
 &   , sk, s, globalcell, bcdata, nx, il, ie, ib, ny, jl, je, jb, nz, kl,&
 &   ke, kb, bcfaceid, addgridvelocities, sfacei, sfacej, sfacek, &
-&   addgridvelocities
+&   addgridvelocities, d2wall, ks
     use bcpointers_d, only : ww0, ww1, ww2, ww3, pp0, pp1, pp2, pp3, &
 &   rlv0, rlv1, rlv2, rlv3, rev0, rev1, rev2, rev3, gamma0, gamma1, &
 &   gamma2, gamma3, gcp, xx, ss, ssi, ssj, ssk, dd2wall, sface, istart, &
-&   iend, jstart, jend, isize, jsize
-    use inputphysics, only : cpmodel, equations
+&   iend, jstart, jend, isize, jsize, d2wall0, d2wall1, d2wall2, d2wall3&
+&   , ks0, ks1, ks2, ks3
+    use inputphysics, only : cpmodel, equations, useroughsa
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -772,6 +773,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(2, 1:, 1:)
       rev1 => rev(1, 1:, 1:)
       rev0 => rev(0, 1:, 1:)
+      d2wall3 => d2wall(3, 1:, 1:)
+      d2wall2 => d2wall(2, 1:, 1:)
+      d2wall1 => d2wall(1, 1:, 1:)
+      d2wall0 => d2wall(0, 1:, 1:)
+      if (useroughsa) then
+        ks3 => ks(3, 1:, 1:)
+        ks2 => ks(2, 1:, 1:)
+        ks1 => ks(1, 1:, 1:)
+        ks0 => ks(0, 1:, 1:)
+      end if
       gamma3 => gamma(3, 1:, 1:)
       gamma2 => gamma(2, 1:, 1:)
       gamma1 => gamma(1, 1:, 1:)
@@ -795,6 +806,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(il, 1:, 1:)
       rev1 => rev(ie, 1:, 1:)
       rev0 => rev(ib, 1:, 1:)
+      d2wall3 => d2wall(nx, 1:, 1:)
+      d2wall2 => d2wall(il, 1:, 1:)
+      d2wall1 => d2wall(ie, 1:, 1:)
+      d2wall0 => d2wall(ib, 1:, 1:)
+      if (useroughsa) then
+        ks3 => ks(nx, 1:, 1:)
+        ks2 => ks(il, 1:, 1:)
+        ks1 => ks(ie, 1:, 1:)
+        ks0 => ks(ib, 1:, 1:)
+      end if
       gamma3 => gamma(nx, 1:, 1:)
       gamma2 => gamma(il, 1:, 1:)
       gamma1 => gamma(ie, 1:, 1:)
@@ -818,6 +839,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 2, 1:)
       rev1 => rev(1:, 1, 1:)
       rev0 => rev(1:, 0, 1:)
+      d2wall3 => d2wall(1:, 3, 1:)
+      d2wall2 => d2wall(1:, 2, 1:)
+      d2wall1 => d2wall(1:, 1, 1:)
+      d2wall0 => d2wall(1:, 0, 1:)
+      if (useroughsa) then
+        ks3 => ks(1:, 3, 1:)
+        ks2 => ks(1:, 2, 1:)
+        ks1 => ks(1:, 1, 1:)
+        ks0 => ks(1:, 0, 1:)
+      end if
       gamma3 => gamma(1:, 3, 1:)
       gamma2 => gamma(1:, 2, 1:)
       gamma1 => gamma(1:, 1, 1:)
@@ -841,6 +872,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, jl, 1:)
       rev1 => rev(1:, je, 1:)
       rev0 => rev(1:, jb, 1:)
+      d2wall3 => d2wall(1:, ny, 1:)
+      d2wall2 => d2wall(1:, jl, 1:)
+      d2wall1 => d2wall(1:, je, 1:)
+      d2wall0 => d2wall(1:, jb, 1:)
+      if (useroughsa) then
+        ks3 => ks(1:, ny, 1:)
+        ks2 => ks(1:, jl, 1:)
+        ks1 => ks(1:, je, 1:)
+        ks0 => ks(1:, jb, 1:)
+      end if
       gamma3 => gamma(1:, ny, 1:)
       gamma2 => gamma(1:, jl, 1:)
       gamma1 => gamma(1:, je, 1:)
@@ -864,6 +905,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 1:, 2)
       rev1 => rev(1:, 1:, 1)
       rev0 => rev(1:, 1:, 0)
+      d2wall3 => d2wall(1:, 1:, 3)
+      d2wall2 => d2wall(1:, 1:, 2)
+      d2wall1 => d2wall(1:, 1:, 1)
+      d2wall0 => d2wall(1:, 1:, 0)
+      if (useroughsa) then
+        ks3 => ks(1:, 1:, 3)
+        ks2 => ks(1:, 1:, 2)
+        ks1 => ks(1:, 1:, 1)
+        ks0 => ks(1:, 1:, 0)
+      end if
       gamma3 => gamma(1:, 1:, 3)
       gamma2 => gamma(1:, 1:, 2)
       gamma1 => gamma(1:, 1:, 1)
@@ -887,6 +938,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 1:, kl)
       rev1 => rev(1:, 1:, ke)
       rev0 => rev(1:, 1:, kb)
+      d2wall3 => d2wall(1:, 1:, nz)
+      d2wall2 => d2wall(1:, 1:, kl)
+      d2wall1 => d2wall(1:, 1:, ke)
+      d2wall0 => d2wall(1:, 1:, kb)
+      if (useroughsa) then
+        ks3 => ks(1:, 1:, nz)
+        ks2 => ks(1:, 1:, kl)
+        ks1 => ks(1:, 1:, ke)
+        ks0 => ks(1:, 1:, kb)
+      end if
       gamma3 => gamma(1:, 1:, nz)
       gamma2 => gamma(1:, 1:, kl)
       gamma1 => gamma(1:, 1:, ke)
diff --git a/src/adjoint/outputReverse/BCData_b.f90 b/src/adjoint/outputReverse/BCData_b.f90
index f3684fe6a..4d7ef5267 100644
--- a/src/adjoint/outputReverse/BCData_b.f90
+++ b/src/adjoint/outputReverse/BCData_b.f90
@@ -13,11 +13,28 @@ module bcdata_b
   subroutine setbcvarnamesisothermalwall()
     use cgnsnames
     use constants
+    use inputphysics, only : useroughsa
     implicit none
     nbcvar = nbcvarisothermalwall
     bcvarnames(1) = cgnstemp
+    if (useroughsa) then
+      nbcvar = nbcvar + 1
+      bcvarnames(2) = cgnssandgrainroughness
+    end if
   end subroutine setbcvarnamesisothermalwall
 
+  subroutine setbcvarnamesadiabaticwall()
+    use cgnsnames
+    use constants
+    use inputphysics, only : useroughsa
+    implicit none
+    nbcvar = nbcvaradiabaticwall
+    if (useroughsa) then
+      nbcvar = nbcvar + 1
+      bcvarnames(1) = cgnssandgrainroughness
+    end if
+  end subroutine setbcvarnamesadiabaticwall
+
   subroutine setbcvarnamessubsonicinflow()
     use constants
     use cgnsnames
@@ -346,6 +363,7 @@ subroutine bcdataisothermalwall_b(boco, bcvararray, bcvararrayd, ibeg&
     use blockpointers, only : bcfaceid, bcdata, bcdatad, nbkglobal
     use utils_b, only : terminate, sitemperature
     use flowvarrefstate, only : tref, trefd
+    use inputphysics, only : useroughsa
     implicit none
 !
 !      subroutine arguments.
@@ -395,6 +413,7 @@ subroutine bcdataisothermalwall(boco, bcvararray, ibeg, iend, jbeg, &
     use blockpointers, only : bcfaceid, bcdata, nbkglobal
     use utils_b, only : terminate, sitemperature
     use flowvarrefstate, only : tref
+    use inputphysics, only : useroughsa
     implicit none
 !
 !      subroutine arguments.
@@ -420,8 +439,54 @@ subroutine bcdataisothermalwall(boco, bcvararray, ibeg, iend, jbeg, &
 &         tref
       end do
     end do
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+    if (useroughsa) then
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+      if (.not.bcvarpresent(1)) bcvararray(:, :, 2) = zero
+      do j=jbeg,jend
+        do i=ibeg,iend
+          bcdata(boco)%ksns_wall(i, j) = bcvararray(i, j, 2)
+        end do
+      end do
+    end if
   end subroutine bcdataisothermalwall
 
+  subroutine bcdataadiabaticwall(boco, bcvararray, ibeg, iend, jbeg, &
+&   jend)
+!
+!       tries to extract the equivalent sand grain roughness. it sets
+!       a default value of 0.0
+!
+    use constants
+    use cgnsnames
+    use inputphysics, only : useroughsa
+    use blockpointers, only : bcfaceid, bcdata, nbkglobal
+    implicit none
+!
+!      subroutine arguments.
+!
+    integer(kind=inttype) :: boco
+    integer(kind=inttype) :: ibeg, iend, jbeg, jend
+    real(kind=realtype), dimension(ibeg:iend, jbeg:jend, nbcvarmax) :: &
+&   bcvararray
+!
+!      local variables.
+!
+    integer(kind=inttype) :: i, j
+    if (useroughsa) then
+! set a value of 0 if it was not possible to determine the
+! sand grain roughness
+      if (.not.bcvarpresent(1)) bcvararray(:, :, 1) = zero
+      do j=jbeg,jend
+        do i=ibeg,iend
+          bcdata(boco)%ksns_wall(i, j) = bcvararray(i, j, 1)
+        end do
+      end do
+    end if
+  end subroutine bcdataadiabaticwall
+
 !  differentiation of bcdatasubsonicinflow in reverse (adjoint) mode (with options noisize i4 dr8 r8):
 !   gradient     of useful results: muref tref winf href pref rhoref
 !                *(*bcdata.ptinlet) *(*bcdata.ttinlet) *(*bcdata.htinlet)
diff --git a/src/adjoint/outputReverse/BCRoutines_b.f90 b/src/adjoint/outputReverse/BCRoutines_b.f90
index cdbf2ebf7..e8ed35f16 100644
--- a/src/adjoint/outputReverse/BCRoutines_b.f90
+++ b/src/adjoint/outputReverse/BCRoutines_b.f90
@@ -156,15 +156,16 @@ subroutine applyallbc_block(secondhalo)
   end subroutine applyallbc_block
 
 !  differentiation of bcsymm1sthalo in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: *rev1 *rev2 *pp1 *pp2 *rlv1
-!                *rlv2 *ww1 *ww2 *(*bcdata.norm)
-!   with respect to varying inputs: *rev1 *rev2 *pp1 *pp2 *rlv1
-!                *rlv2 *ww1 *ww2 *(*bcdata.norm)
-!   rw status of diff variables: *rev1:in-out *rev2:incr *pp1:in-out
-!                *pp2:incr *rlv1:in-out *rlv2:incr *ww1:in-out
-!                *ww2:incr *(*bcdata.norm):incr
-!   plus diff mem management of: rev1:in rev2:in pp1:in pp2:in
-!                rlv1:in rlv2:in ww1:in ww2:in bcdata:in *bcdata.norm:in
+!   gradient     of useful results: *rev1 *rev2 *d2wall1 *d2wall2
+!                *pp1 *pp2 *rlv1 *rlv2 *ww1 *ww2 *(*bcdata.norm)
+!   with respect to varying inputs: *rev1 *rev2 *d2wall1 *d2wall2
+!                *pp1 *pp2 *rlv1 *rlv2 *ww1 *ww2 *(*bcdata.norm)
+!   rw status of diff variables: *rev1:in-out *rev2:incr *d2wall1:in-out
+!                *d2wall2:incr *pp1:in-out *pp2:incr *rlv1:in-out
+!                *rlv2:incr *ww1:in-out *ww2:incr *(*bcdata.norm):incr
+!   plus diff mem management of: rev1:in rev2:in d2wall1:in d2wall2:in
+!                pp1:in pp2:in rlv1:in rlv2:in ww1:in ww2:in bcdata:in
+!                *bcdata.norm:in
 ! ===================================================================
 !   actual implementation of each of the boundary condition routines
 ! ===================================================================
@@ -180,9 +181,11 @@ subroutine bcsymm1sthalo_b(nn)
     use constants
     use blockpointers, only : bcdata, bcdatad
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : gamma1, gamma2, ww1, ww1d, ww2, ww2d, pp1, &
 &   pp1d, pp2, pp2d, rlv1, rlv1d, rlv2, rlv2d, istart, jstart, isize, &
-&   jsize, rev1, rev1d, rev2, rev2d
+&   jsize, rev1, rev1d, rev2, rev2d, d2wall1, d2wall1d, d2wall2, &
+&   d2wall2d, ks1, ks2
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -220,6 +223,8 @@ subroutine bcsymm1sthalo_b(nn)
         rlv2d(i, j) = rlv2d(i, j) + rlv1d(i, j)
         rlv1d(i, j) = 0.0_8
       end if
+      d2wall2d(i, j) = d2wall2d(i, j) + d2wall1d(i, j)
+      d2wall1d(i, j) = 0.0_8
       pp2d(i, j) = pp2d(i, j) + pp1d(i, j)
       pp1d(i, j) = 0.0_8
       ww2d(i, j, irhoe) = ww2d(i, j, irhoe) + ww1d(i, j, irhoe)
@@ -267,8 +272,10 @@ subroutine bcsymm1sthalo(nn)
     use constants
     use blockpointers, only : bcdata
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : gamma1, gamma2, ww1, ww2, pp1, pp2, rlv1, &
-&   rlv2, istart, jstart, isize, jsize, rev1, rev2
+&   rlv2, istart, jstart, isize, jsize, rev1, rev2, d2wall1, d2wall2, &
+&   ks1, ks2
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -298,6 +305,8 @@ subroutine bcsymm1sthalo(nn)
 ! laminar and eddy viscosity in the halo.
       gamma1(i, j) = gamma2(i, j)
       pp1(i, j) = pp2(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) rlv1(i, j) = rlv2(i, j)
       if (eddymodel) rev1(i, j) = rev2(i, j)
     end do
@@ -305,14 +314,14 @@ end subroutine bcsymm1sthalo
 
 !  differentiation of bcsymm2ndhalo in reverse (adjoint) mode (with options noisize i4 dr8 r8):
 !   gradient     of useful results: *rev0 *rev3 *pp0 *pp3 *rlv0
-!                *rlv3 *ww0 *ww3 *(*bcdata.norm)
+!                *rlv3 *ww0 *ww3
 !   with respect to varying inputs: *rev0 *rev3 *pp0 *pp3 *rlv0
-!                *rlv3 *ww0 *ww3 *(*bcdata.norm)
+!                *rlv3 *ww0 *ww3
 !   rw status of diff variables: *rev0:in-out *rev3:incr *pp0:in-out
 !                *pp3:incr *rlv0:in-out *rlv3:incr *ww0:in-out
-!                *ww3:incr *(*bcdata.norm):incr
+!                *ww3:incr
 !   plus diff mem management of: rev0:in rev3:in pp0:in pp3:in
-!                rlv0:in rlv3:in ww0:in ww3:in bcdata:in *bcdata.norm:in
+!                rlv0:in rlv3:in ww0:in ww3:in
   subroutine bcsymm2ndhalo_b(nn)
 !  bcsymm2ndhalo applies the symmetry boundary conditions to a
 !  block for the 2nd halo. this routine is separate as it makes
@@ -320,9 +329,11 @@ subroutine bcsymm2ndhalo_b(nn)
     use constants
     use blockpointers, only : bcdata, bcdatad
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : gamma0, gamma3, ww0, ww0d, ww3, ww3d, pp0, &
-&   pp0d, pp3, pp3d, rlv0, rlv0d, rlv3, rlv3d, rev0, rev0d, rev3, rev3d,&
-&   istart, jstart, isize, jsize
+&   pp0d, pp3, pp3d, rlv0, rlv0d, rlv3, rlv3d, d2wall0, d2wall0d, &
+&   d2wall3, d2wall3d, ks0, ks3, rev0, rev0d, rev3, rev3d, istart, &
+&   jstart, isize, jsize
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -337,9 +348,6 @@ subroutine bcsymm2ndhalo_b(nn)
     do ii=0,isize*jsize-1
       i = mod(ii, isize) + istart
       j = ii/isize + jstart
-      vn = two*(ww3(i, j, ivx)*bcdata(nn)%norm(i, j, 1)+ww3(i, j, ivy)*&
-&       bcdata(nn)%norm(i, j, 2)+ww3(i, j, ivz)*bcdata(nn)%norm(i, j, 3)&
-&       )
 ! determine the flow variables in the halo cell.
 ! set the pressure and gamma and possibly the
 ! laminar and eddy viscosity in the halo.
@@ -363,29 +371,19 @@ subroutine bcsymm2ndhalo_b(nn)
       ww0d(i, j, irhoe) = 0.0_8
       ww3d(i, j, ivz) = ww3d(i, j, ivz) + ww0d(i, j, ivz)
       vnd = -(bcdata(nn)%norm(i, j, 3)*ww0d(i, j, ivz))
-      bcdatad(nn)%norm(i, j, 3) = bcdatad(nn)%norm(i, j, 3) - vn*ww0d(i&
-&       , j, ivz)
       ww0d(i, j, ivz) = 0.0_8
       ww3d(i, j, ivy) = ww3d(i, j, ivy) + ww0d(i, j, ivy)
       vnd = vnd - bcdata(nn)%norm(i, j, 2)*ww0d(i, j, ivy)
-      bcdatad(nn)%norm(i, j, 2) = bcdatad(nn)%norm(i, j, 2) - vn*ww0d(i&
-&       , j, ivy)
       ww0d(i, j, ivy) = 0.0_8
       ww3d(i, j, ivx) = ww3d(i, j, ivx) + ww0d(i, j, ivx)
       vnd = vnd - bcdata(nn)%norm(i, j, 1)*ww0d(i, j, ivx)
-      tempd = two*vnd
-      bcdatad(nn)%norm(i, j, 1) = bcdatad(nn)%norm(i, j, 1) + ww3(i, j, &
-&       ivx)*tempd - vn*ww0d(i, j, ivx)
       ww0d(i, j, ivx) = 0.0_8
       ww3d(i, j, irho) = ww3d(i, j, irho) + ww0d(i, j, irho)
       ww0d(i, j, irho) = 0.0_8
+      tempd = two*vnd
       ww3d(i, j, ivx) = ww3d(i, j, ivx) + bcdata(nn)%norm(i, j, 1)*tempd
       ww3d(i, j, ivy) = ww3d(i, j, ivy) + bcdata(nn)%norm(i, j, 2)*tempd
-      bcdatad(nn)%norm(i, j, 2) = bcdatad(nn)%norm(i, j, 2) + ww3(i, j, &
-&       ivy)*tempd
       ww3d(i, j, ivz) = ww3d(i, j, ivz) + bcdata(nn)%norm(i, j, 3)*tempd
-      bcdatad(nn)%norm(i, j, 3) = bcdatad(nn)%norm(i, j, 3) + ww3(i, j, &
-&       ivz)*tempd
     end do
   end subroutine bcsymm2ndhalo_b
 
@@ -396,8 +394,10 @@ subroutine bcsymm2ndhalo(nn)
     use constants
     use blockpointers, only : bcdata
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : gamma0, gamma3, ww0, ww3, pp0, pp3, rlv0, &
-&   rlv3, rev0, rev3, istart, jstart, isize, jsize
+&   rlv3, d2wall0, d2wall3, ks0, ks3, rev0, rev3, istart, jstart, isize,&
+&   jsize
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -424,21 +424,24 @@ subroutine bcsymm2ndhalo(nn)
 ! laminar and eddy viscosity in the halo.
       gamma0(i, j) = gamma3(i, j)
       pp0(i, j) = pp3(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) rlv0(i, j) = rlv3(i, j)
       if (eddymodel) rev0(i, j) = rev3(i, j)
     end do
   end subroutine bcsymm2ndhalo
 
 !  differentiation of bcsymmpolar1sthalo in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: *xx *rev1 *rev2 *pp1 *pp2 *rlv1
-!                *rlv2 *ww1 *ww2
-!   with respect to varying inputs: *xx *rev1 *rev2 *pp1 *pp2 *rlv1
-!                *rlv2 *ww1 *ww2
+!   gradient     of useful results: *xx *rev1 *rev2 *d2wall1 *d2wall2
+!                *pp1 *pp2 *rlv1 *rlv2 *ww1 *ww2
+!   with respect to varying inputs: *xx *rev1 *rev2 *d2wall1 *d2wall2
+!                *pp1 *pp2 *rlv1 *rlv2 *ww1 *ww2
 !   rw status of diff variables: *xx:incr *rev1:in-out *rev2:incr
-!                *pp1:in-out *pp2:incr *rlv1:in-out *rlv2:incr
-!                *ww1:in-out *ww2:incr
-!   plus diff mem management of: xx:in rev1:in rev2:in pp1:in pp2:in
-!                rlv1:in rlv2:in ww1:in ww2:in
+!                *d2wall1:in-out *d2wall2:incr *pp1:in-out *pp2:incr
+!                *rlv1:in-out *rlv2:incr *ww1:in-out *ww2:incr
+!   plus diff mem management of: xx:in rev1:in rev2:in d2wall1:in
+!                d2wall2:in pp1:in pp2:in rlv1:in rlv2:in ww1:in
+!                ww2:in
   subroutine bcsymmpolar1sthalo_b(nn)
 ! bcsymmpolar applies the polar symmetry boundary conditions to a
 ! singular line of a block. it is assumed that the pointers in
@@ -447,9 +450,10 @@ subroutine bcsymmpolar1sthalo_b(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 1st level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : ww1, ww1d, ww2, ww2d, pp1, pp1d, pp2, pp2d,&
 &   rlv1, rlv1d, rlv2, rlv2d, rev1, rev1d, rev2, rev2d, xx, xxd, istart,&
-&   jstart, isize, jsize
+&   jstart, isize, jsize, d2wall1, d2wall1d, d2wall2, d2wall2d, ks1, ks2
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -514,6 +518,8 @@ subroutine bcsymmpolar1sthalo_b(nn)
         rlv2d(i, j) = rlv2d(i, j) + rlv1d(i, j)
         rlv1d(i, j) = 0.0_8
       end if
+      d2wall2d(i, j) = d2wall2d(i, j) + d2wall1d(i, j)
+      d2wall1d(i, j) = 0.0_8
       pp2d(i, j) = pp2d(i, j) + pp1d(i, j)
       pp1d(i, j) = 0.0_8
       ww2d(i, j, irhoe) = ww2d(i, j, irhoe) + ww1d(i, j, irhoe)
@@ -569,8 +575,9 @@ subroutine bcsymmpolar1sthalo(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 1st level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2,&
-&   xx, istart, jstart, isize, jsize
+&   xx, istart, jstart, isize, jsize, d2wall1, d2wall2, ks1, ks2
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -622,21 +629,24 @@ subroutine bcsymmpolar1sthalo(nn)
 ! set the pressure and possibly the laminar and
 ! eddy viscosity in the halo.
       pp1(i, j) = pp2(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) rlv1(i, j) = rlv2(i, j)
       if (eddymodel) rev1(i, j) = rev2(i, j)
     end do
   end subroutine bcsymmpolar1sthalo
 
 !  differentiation of bcsymmpolar2ndhalo in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: *xx *rev0 *rev3 *pp0 *pp3 *rlv0
-!                *rlv3 *ww0 *ww3
-!   with respect to varying inputs: *xx *rev0 *rev3 *pp0 *pp3 *rlv0
-!                *rlv3 *ww0 *ww3
-!   rw status of diff variables: *xx:incr *rev0:in-out *rev3:incr
-!                *pp0:in-out *pp3:incr *rlv0:in-out *rlv3:incr
-!                *ww0:in-out *ww3:incr
-!   plus diff mem management of: xx:in rev0:in rev3:in pp0:in pp3:in
-!                rlv0:in rlv3:in ww0:in ww3:in
+!   gradient     of useful results: *xx *rev0 *d2wall0 *rev3 *d2wall3
+!                *pp0 *pp3 *rlv0 *rlv3 *ww0 *ww3
+!   with respect to varying inputs: *xx *rev0 *d2wall0 *rev3 *d2wall3
+!                *pp0 *pp3 *rlv0 *rlv3 *ww0 *ww3
+!   rw status of diff variables: *xx:incr *rev0:in-out *d2wall0:in-out
+!                *rev3:incr *d2wall3:incr *pp0:in-out *pp3:incr
+!                *rlv0:in-out *rlv3:incr *ww0:in-out *ww3:incr
+!   plus diff mem management of: xx:in rev0:in d2wall0:in rev3:in
+!                d2wall3:in pp0:in pp3:in rlv0:in rlv3:in ww0:in
+!                ww3:in
   subroutine bcsymmpolar2ndhalo_b(nn)
 ! bcsymmpolar applies the polar symmetry boundary conditions to a
 ! singular line of a block. it is assumed that the pointers in
@@ -645,9 +655,11 @@ subroutine bcsymmpolar2ndhalo_b(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 2nd level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : ww0, ww0d, ww3, ww3d, pp0, pp0d, pp3, pp3d,&
-&   rlv0, rlv0d, rlv3, rlv3d, rev0, rev0d, rev3, rev3d, xx, xxd, istart,&
-&   jstart, isize, jsize
+&   rlv0, rlv0d, rlv3, rlv3d, rev0, rev0d, rev3, rev3d, d2wall0, &
+&   d2wall0d, d2wall3, d2wall3d, ks0, ks3, xx, xxd, istart, jstart, &
+&   isize, jsize
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -712,6 +724,8 @@ subroutine bcsymmpolar2ndhalo_b(nn)
         rlv3d(i, j) = rlv3d(i, j) + rlv0d(i, j)
         rlv0d(i, j) = 0.0_8
       end if
+      d2wall3d(i, j) = d2wall3d(i, j) + d2wall0d(i, j)
+      d2wall0d(i, j) = 0.0_8
       pp3d(i, j) = pp3d(i, j) + pp0d(i, j)
       pp0d(i, j) = 0.0_8
       ww3d(i, j, irhoe) = ww3d(i, j, irhoe) + ww0d(i, j, irhoe)
@@ -767,8 +781,9 @@ subroutine bcsymmpolar2ndhalo(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 2nd level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers_b, only : ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3,&
-&   xx, istart, jstart, isize, jsize
+&   d2wall0, d2wall3, ks0, ks3, xx, istart, jstart, isize, jsize
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -818,6 +833,8 @@ subroutine bcsymmpolar2ndhalo(nn)
 ! set the pressure and possibly the laminar and
 ! eddy viscosity in the halo.
       pp0(i, j) = pp3(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) rlv0(i, j) = rlv3(i, j)
       if (eddymodel) rev0(i, j) = rev3(i, j)
     end do
diff --git a/src/adjoint/outputReverse/sa_b.f90 b/src/adjoint/outputReverse/sa_b.f90
index e960263e3..83b5c3b31 100644
--- a/src/adjoint/outputReverse/sa_b.f90
+++ b/src/adjoint/outputReverse/sa_b.f90
@@ -42,6 +42,8 @@ subroutine sasource_b()
     real(kind=realtype), parameter :: f23=two*third
 ! local variables.
     integer(kind=inttype) :: i, j, k, nn, ii
+    real(kind=realtype) :: distrough
+    real(kind=realtype) :: distroughd
     real(kind=realtype) :: fv1, fv2, ft2
     real(kind=realtype) :: fv1d, fv2d, ft2d
     real(kind=realtype) :: ss, sst, nu, dist2inv, chi, chi2, chi3
@@ -184,12 +186,26 @@ subroutine sasource_b()
 ! and nu) and the functions fv1 and fv2. the latter corrects
 ! the production term near a viscous wall.
         nu = rlv(i, j, k)/w(i, j, k, irho)
-        dist2inv = one/d2wall(i, j, k)**2
         chi = w(i, j, k, itu1)/nu
+        if (.not.useroughsa) then
+          dist2inv = one/d2wall(i, j, k)**2
+          call pushcontrol1b(0)
+        else
+          distrough = d2wall(i, j, k) + 0.03_realtype*ks(i, j, k)
+          dist2inv = one/distrough**2
+          chi = chi + rsacr1*ks(i, j, k)/distrough
+          call pushcontrol1b(1)
+        end if
         chi2 = chi*chi
         chi3 = chi*chi2
         fv1 = chi3/(chi3+cv13)
-        fv2 = one - chi/(one+chi*fv1)
+        if (.not.useroughsa) then
+          fv2 = one - chi/(one+chi*fv1)
+          call pushcontrol1b(0)
+        else
+          fv2 = one - w(i, j, k, itu1)/(nu+w(i, j, k, itu1)*fv1)
+          call pushcontrol1b(1)
+        end if
 ! the function ft2, which is designed to keep a laminar
 ! solution laminar. when running in fully turbulent mode
 ! this function should be set to 0.0.
@@ -311,18 +327,37 @@ subroutine sasource_b()
         else
           chi2d = 0.0_8
         end if
-        tempd = -(fv2d/(one+chi*fv1))
-        chid = tempd
-        tempd0 = -(chi*tempd/(one+chi*fv1))
-        fv1d = chi*tempd0
+        call popcontrol1b(branch)
+        if (branch .eq. 0) then
+          tempd = -(fv2d/(one+chi*fv1))
+          tempd0 = -(chi*tempd/(one+chi*fv1))
+          chid = tempd + fv1*tempd0
+          fv1d = chi*tempd0
+          nud = 0.0_8
+        else
+          temp0 = w(i, j, k, itu1)
+          temp = nu + temp0*fv1
+          tempd = w(i, j, k, itu1)*fv2d/temp**2
+          wd(i, j, k, itu1) = wd(i, j, k, itu1) + fv1*tempd - fv2d/temp
+          nud = tempd
+          fv1d = temp0*tempd
+          chid = 0.0_8
+        end if
         tempd = fv1d/(cv13+chi3)
         chi3d = (1.0-chi3/(cv13+chi3))*tempd
         chi2d = chi2d + chi*chi3d
-        chid = chid + fv1*tempd0 + chi2*chi3d + 2*chi*chi2d
+        chid = chid + chi2*chi3d + 2*chi*chi2d
+        call popcontrol1b(branch)
+        if (branch .eq. 0) then
+          temp = d2wall(i, j, k)
+          d2walld(i, j, k) = d2walld(i, j, k) - 2*one*dist2invd/temp**3
+        else
+          distroughd = -(rsacr1*ks(i, j, k)*chid/distrough**2) - 2*one*&
+&           dist2invd/distrough**3
+          d2walld(i, j, k) = d2walld(i, j, k) + distroughd
+        end if
         wd(i, j, k, itu1) = wd(i, j, k, itu1) + chid/nu
-        nud = -(w(i, j, k, itu1)*chid/nu**2)
-        temp = d2wall(i, j, k)
-        d2walld(i, j, k) = d2walld(i, j, k) - 2*one*dist2invd/temp**3
+        nud = nud - w(i, j, k, itu1)*chid/nu**2
         temp = w(i, j, k, irho)
         rlvd(i, j, k) = rlvd(i, j, k) + nud/temp
         wd(i, j, k, irho) = wd(i, j, k, irho) - rlv(i, j, k)*nud/temp**2
@@ -511,6 +546,7 @@ subroutine sasource()
     real(kind=realtype), parameter :: f23=two*third
 ! local variables.
     integer(kind=inttype) :: i, j, k, nn, ii
+    real(kind=realtype) :: distrough
     real(kind=realtype) :: fv1, fv2, ft2
     real(kind=realtype) :: ss, sst, nu, dist2inv, chi, chi2, chi3
     real(kind=realtype) :: rr, gg, gg6, termfw, fwsa, term1, term2
@@ -627,12 +663,22 @@ subroutine sasource()
 ! and nu) and the functions fv1 and fv2. the latter corrects
 ! the production term near a viscous wall.
         nu = rlv(i, j, k)/w(i, j, k, irho)
-        dist2inv = one/d2wall(i, j, k)**2
         chi = w(i, j, k, itu1)/nu
+        if (.not.useroughsa) then
+          dist2inv = one/d2wall(i, j, k)**2
+        else
+          distrough = d2wall(i, j, k) + 0.03_realtype*ks(i, j, k)
+          dist2inv = one/distrough**2
+          chi = chi + rsacr1*ks(i, j, k)/distrough
+        end if
         chi2 = chi*chi
         chi3 = chi*chi2
         fv1 = chi3/(chi3+cv13)
-        fv2 = one - chi/(one+chi*fv1)
+        if (.not.useroughsa) then
+          fv2 = one - chi/(one+chi*fv1)
+        else
+          fv2 = one - w(i, j, k, itu1)/(nu+w(i, j, k, itu1)*fv1)
+        end if
 ! the function ft2, which is designed to keep a laminar
 ! solution laminar. when running in fully turbulent mode
 ! this function should be set to 0.0.
diff --git a/src/adjoint/outputReverse/turbBCRoutines_b.f90 b/src/adjoint/outputReverse/turbBCRoutines_b.f90
index 589ef941c..c16bc2266 100644
--- a/src/adjoint/outputReverse/turbBCRoutines_b.f90
+++ b/src/adjoint/outputReverse/turbBCRoutines_b.f90
@@ -6,13 +6,19 @@ module turbbcroutines_b
 
 contains
 !  differentiation of applyallturbbcthisblock in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: *rev *w
-!   with respect to varying inputs: *rev *bvtj1 *bvtj2 *w *bvtk1
-!                *bvtk2 *bvti1 *bvti2
-!   rw status of diff variables: *rev:in-out *bvtj1:out *bvtj2:out
-!                *w:in-out *bvtk1:out *bvtk2:out *bvti1:out *bvti2:out
+!   gradient     of useful results: *rev *bvtj1 *bvtj2 *w *bmtk1
+!                *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2 *bvti1
+!                *bvti2 *bmtj1 *bmtj2
+!   with respect to varying inputs: *rev *bvtj1 *bvtj2 *w *bmtk1
+!                *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2 *bvti1
+!                *bvti2 *bmtj1 *bmtj2
+!   rw status of diff variables: *rev:in-out *bvtj1:incr *bvtj2:incr
+!                *w:in-out *bmtk1:incr *bmtk2:incr *bvtk1:incr
+!                *bvtk2:incr *d2wall:incr *bmti1:incr *bmti2:incr
+!                *bvti1:incr *bvti2:incr *bmtj1:incr *bmtj2:incr
 !   plus diff mem management of: rev:in bvtj1:in bvtj2:in w:in
-!                bvtk1:in bvtk2:in bvti1:in bvti2:in
+!                bmtk1:in bmtk2:in bvtk1:in bvtk2:in d2wall:in
+!                bmti1:in bmti2:in bvti1:in bvti2:in bmtj1:in bmtj2:in
 !      ==================================================================
   subroutine applyallturbbcthisblock_b(secondhalo)
 !
@@ -41,6 +47,7 @@ subroutine applyallturbbcthisblock_b(secondhalo)
     real(kind=realtype) :: tmpd0
     real(kind=realtype) :: tmp1
     real(kind=realtype) :: tmpd1
+    integer :: branch
     integer :: ad_from
     integer :: ad_to
     integer :: ad_from0
@@ -65,7 +72,6 @@ subroutine applyallturbbcthisblock_b(secondhalo)
     integer :: ad_to9
     integer :: ad_from10
     integer :: ad_to10
-    integer :: branch
 ! loop over the boundary condition subfaces of this block.
 bocos:do nn=1,nbocos
 ! loop over the faces and set the state in
@@ -78,7 +84,17 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           ad_from0 = bcdata(nn)%jcbeg
           do j=ad_from0,bcdata(nn)%jcend
             ad_from = bcdata(nn)%icbeg
-            i = bcdata(nn)%icend + 1
+            do i=ad_from,bcdata(nn)%icend
+              do l=nt1,nt2
+                call pushreal8(w(1, i, j, l))
+                w(1, i, j, l) = bvti1(i, j, l)
+                do m=nt1,nt2
+                  call pushreal8(w(1, i, j, l))
+                  w(1, i, j, l) = w(1, i, j, l) - bmti1(i, j, l, m)*w(2&
+&                   , i, j, m)
+                end do
+              end do
+            end do
             call pushinteger4(i - 1)
             call pushinteger4(ad_from)
           end do
@@ -89,7 +105,18 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           ad_from2 = bcdata(nn)%jcbeg
           do j=ad_from2,bcdata(nn)%jcend
             ad_from1 = bcdata(nn)%icbeg
-            i = bcdata(nn)%icend + 1
+            do i=ad_from1,bcdata(nn)%icend
+              do l=nt1,nt2
+                call pushreal8(w(ie, i, j, l))
+                w(ie, i, j, l) = bvti2(i, j, l)
+                do m=nt1,nt2
+                  tmp = w(ie, i, j, l) - bmti2(i, j, l, m)*w(il, i, j, m&
+&                   )
+                  call pushreal8(w(ie, i, j, l))
+                  w(ie, i, j, l) = tmp
+                end do
+              end do
+            end do
             call pushinteger4(i - 1)
             call pushinteger4(ad_from1)
           end do
@@ -100,7 +127,17 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           ad_from4 = bcdata(nn)%jcbeg
           do j=ad_from4,bcdata(nn)%jcend
             ad_from3 = bcdata(nn)%icbeg
-            i = bcdata(nn)%icend + 1
+            do i=ad_from3,bcdata(nn)%icend
+              do l=nt1,nt2
+                call pushreal8(w(i, 1, j, l))
+                w(i, 1, j, l) = bvtj1(i, j, l)
+                do m=nt1,nt2
+                  call pushreal8(w(i, 1, j, l))
+                  w(i, 1, j, l) = w(i, 1, j, l) - bmtj1(i, j, l, m)*w(i&
+&                   , 2, j, m)
+                end do
+              end do
+            end do
             call pushinteger4(i - 1)
             call pushinteger4(ad_from3)
           end do
@@ -111,7 +148,18 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           ad_from6 = bcdata(nn)%jcbeg
           do j=ad_from6,bcdata(nn)%jcend
             ad_from5 = bcdata(nn)%icbeg
-            i = bcdata(nn)%icend + 1
+            do i=ad_from5,bcdata(nn)%icend
+              do l=nt1,nt2
+                call pushreal8(w(i, je, j, l))
+                w(i, je, j, l) = bvtj2(i, j, l)
+                do m=nt1,nt2
+                  tmp0 = w(i, je, j, l) - bmtj2(i, j, l, m)*w(i, jl, j, &
+&                   m)
+                  call pushreal8(w(i, je, j, l))
+                  w(i, je, j, l) = tmp0
+                end do
+              end do
+            end do
             call pushinteger4(i - 1)
             call pushinteger4(ad_from5)
           end do
@@ -122,7 +170,17 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           ad_from8 = bcdata(nn)%jcbeg
           do j=ad_from8,bcdata(nn)%jcend
             ad_from7 = bcdata(nn)%icbeg
-            i = bcdata(nn)%icend + 1
+            do i=ad_from7,bcdata(nn)%icend
+              do l=nt1,nt2
+                call pushreal8(w(i, j, 1, l))
+                w(i, j, 1, l) = bvtk1(i, j, l)
+                do m=nt1,nt2
+                  call pushreal8(w(i, j, 1, l))
+                  w(i, j, 1, l) = w(i, j, 1, l) - bmtk1(i, j, l, m)*w(i&
+&                   , j, 2, m)
+                end do
+              end do
+            end do
             call pushinteger4(i - 1)
             call pushinteger4(ad_from7)
           end do
@@ -133,7 +191,18 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           ad_from10 = bcdata(nn)%jcbeg
           do j=ad_from10,bcdata(nn)%jcend
             ad_from9 = bcdata(nn)%icbeg
-            i = bcdata(nn)%icend + 1
+            do i=ad_from9,bcdata(nn)%icend
+              do l=nt1,nt2
+                call pushreal8(w(i, j, ke, l))
+                w(i, j, ke, l) = bvtk2(i, j, l)
+                do m=nt1,nt2
+                  tmp1 = w(i, j, ke, l) - bmtk2(i, j, l, m)*w(i, j, kl, &
+&                   m)
+                  call pushreal8(w(i, j, ke, l))
+                  w(i, j, ke, l) = tmp1
+                end do
+              end do
+            end do
             call pushinteger4(i - 1)
             call pushinteger4(ad_from9)
           end do
@@ -150,8 +219,21 @@ subroutine applyallturbbcthisblock_b(secondhalo)
       if (eddymodel) then
         if (bctype(nn) .eq. nswalladiabatic .or. bctype(nn) .eq. &
 &           nswallisothermal) then
+! viscous wall boundary condition. eddy viscosity is
+! zero at the wall.
+          if (associated(rev)) then
+            call pushreal8array(rev, size(rev, 1)*size(rev, 2)*size(rev&
+&                         , 3))
+            call pushcontrol1b(1)
+          else
+            call pushcontrol1b(0)
+          end if
+          call bceddywall(nn)
           call pushcontrol2b(0)
         else
+! any boundary condition but viscous wall. a homogeneous
+! neumann condition is applied to the eddy viscosity.
+          call bceddynowall(nn)
           call pushcontrol2b(1)
         end if
       else
@@ -160,22 +242,32 @@ subroutine applyallturbbcthisblock_b(secondhalo)
 ! extrapolate the turbulent variables in case a second halo
 ! is needed.
       if (secondhalo) then
+        if (associated(w)) then
+          call pushreal8array(w, size(w, 1)*size(w, 2)*size(w, 3)*size(w&
+&                       , 4))
+          call pushcontrol1b(1)
+        else
+          call pushcontrol1b(0)
+        end if
+        call turb2ndhalo(nn)
         call pushcontrol1b(1)
       else
         call pushcontrol1b(0)
       end if
     end do bocos
-    if (associated(bvtj1d)) bvtj1d = 0.0_8
-    if (associated(bvtj2d)) bvtj2d = 0.0_8
-    if (associated(bvtk1d)) bvtk1d = 0.0_8
-    if (associated(bvtk2d)) bvtk2d = 0.0_8
-    if (associated(bvti1d)) bvti1d = 0.0_8
-    if (associated(bvti2d)) bvti2d = 0.0_8
     do nn=nbocos,1,-1
       call popcontrol1b(branch)
-      if (branch .ne. 0) call turb2ndhalo_b(nn)
+      if (branch .ne. 0) then
+        call popcontrol1b(branch)
+        if (branch .eq. 1) call popreal8array(w, size(w, 1)*size(w, 2)*&
+&                                       size(w, 3)*size(w, 4))
+        call turb2ndhalo_b(nn)
+      end if
       call popcontrol2b(branch)
       if (branch .eq. 0) then
+        call popcontrol1b(branch)
+        if (branch .eq. 1) call popreal8array(rev, size(rev, 1)*size(rev&
+&                                       , 2)*size(rev, 3))
         call bceddywall_b(nn)
       else if (branch .eq. 1) then
         call bceddynowall_b(nn)
@@ -192,9 +284,13 @@ subroutine applyallturbbcthisblock_b(secondhalo)
               do i=ad_to,ad_from,-1
                 do l=nt2,nt1,-1
                   do m=nt2,nt1,-1
+                    call popreal8(w(1, i, j, l))
+                    bmti1d(i, j, l, m) = bmti1d(i, j, l, m) - w(2, i, j&
+&                     , m)*wd(1, i, j, l)
                     wd(2, i, j, m) = wd(2, i, j, m) - bmti1(i, j, l, m)*&
 &                     wd(1, i, j, l)
                   end do
+                  call popreal8(w(1, i, j, l))
                   bvti1d(i, j, l) = bvti1d(i, j, l) + wd(1, i, j, l)
                   wd(1, i, j, l) = 0.0_8
                 end do
@@ -209,11 +305,15 @@ subroutine applyallturbbcthisblock_b(secondhalo)
               do i=ad_to1,ad_from1,-1
                 do l=nt2,nt1,-1
                   do m=nt2,nt1,-1
+                    call popreal8(w(ie, i, j, l))
                     tmpd = wd(ie, i, j, l)
                     wd(ie, i, j, l) = tmpd
+                    bmti2d(i, j, l, m) = bmti2d(i, j, l, m) - w(il, i, j&
+&                     , m)*tmpd
                     wd(il, i, j, m) = wd(il, i, j, m) - bmti2(i, j, l, m&
 &                     )*tmpd
                   end do
+                  call popreal8(w(ie, i, j, l))
                   bvti2d(i, j, l) = bvti2d(i, j, l) + wd(ie, i, j, l)
                   wd(ie, i, j, l) = 0.0_8
                 end do
@@ -231,9 +331,13 @@ subroutine applyallturbbcthisblock_b(secondhalo)
             do i=ad_to3,ad_from3,-1
               do l=nt2,nt1,-1
                 do m=nt2,nt1,-1
+                  call popreal8(w(i, 1, j, l))
+                  bmtj1d(i, j, l, m) = bmtj1d(i, j, l, m) - w(i, 2, j, m&
+&                   )*wd(i, 1, j, l)
                   wd(i, 2, j, m) = wd(i, 2, j, m) - bmtj1(i, j, l, m)*wd&
 &                   (i, 1, j, l)
                 end do
+                call popreal8(w(i, 1, j, l))
                 bvtj1d(i, j, l) = bvtj1d(i, j, l) + wd(i, 1, j, l)
                 wd(i, 1, j, l) = 0.0_8
               end do
@@ -248,11 +352,15 @@ subroutine applyallturbbcthisblock_b(secondhalo)
             do i=ad_to5,ad_from5,-1
               do l=nt2,nt1,-1
                 do m=nt2,nt1,-1
+                  call popreal8(w(i, je, j, l))
                   tmpd0 = wd(i, je, j, l)
                   wd(i, je, j, l) = tmpd0
+                  bmtj2d(i, j, l, m) = bmtj2d(i, j, l, m) - w(i, jl, j, &
+&                   m)*tmpd0
                   wd(i, jl, j, m) = wd(i, jl, j, m) - bmtj2(i, j, l, m)*&
 &                   tmpd0
                 end do
+                call popreal8(w(i, je, j, l))
                 bvtj2d(i, j, l) = bvtj2d(i, j, l) + wd(i, je, j, l)
                 wd(i, je, j, l) = 0.0_8
               end do
@@ -268,9 +376,13 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           do i=ad_to7,ad_from7,-1
             do l=nt2,nt1,-1
               do m=nt2,nt1,-1
+                call popreal8(w(i, j, 1, l))
+                bmtk1d(i, j, l, m) = bmtk1d(i, j, l, m) - w(i, j, 2, m)*&
+&                 wd(i, j, 1, l)
                 wd(i, j, 2, m) = wd(i, j, 2, m) - bmtk1(i, j, l, m)*wd(i&
 &                 , j, 1, l)
               end do
+              call popreal8(w(i, j, 1, l))
               bvtk1d(i, j, l) = bvtk1d(i, j, l) + wd(i, j, 1, l)
               wd(i, j, 1, l) = 0.0_8
             end do
@@ -285,11 +397,15 @@ subroutine applyallturbbcthisblock_b(secondhalo)
           do i=ad_to9,ad_from9,-1
             do l=nt2,nt1,-1
               do m=nt2,nt1,-1
+                call popreal8(w(i, j, ke, l))
                 tmpd1 = wd(i, j, ke, l)
                 wd(i, j, ke, l) = tmpd1
+                bmtk2d(i, j, l, m) = bmtk2d(i, j, l, m) - w(i, j, kl, m)&
+&                 *tmpd1
                 wd(i, j, kl, m) = wd(i, j, kl, m) - bmtk2(i, j, l, m)*&
 &                 tmpd1
               end do
+              call popreal8(w(i, j, ke, l))
               bvtk2d(i, j, l) = bvtk2d(i, j, l) + wd(i, j, ke, l)
               wd(i, j, ke, l) = 0.0_8
             end do
@@ -562,9 +678,9 @@ subroutine bceddynowall(nn)
   end subroutine bceddynowall
 
 !  differentiation of bceddywall in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: *rev
-!   with respect to varying inputs: *rev
-!   plus diff mem management of: rev:in
+!   gradient     of useful results: *rev *d2wall
+!   with respect to varying inputs: *rev *d2wall
+!   plus diff mem management of: rev:in d2wall:in
   subroutine bceddywall_b(nn)
 !
 !       bceddywall sets the eddy viscosity in the halo cells of
@@ -583,6 +699,8 @@ subroutine bceddywall_b(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j
+    real(kind=realtype) :: fact
+    real(kind=realtype) :: factd
     real(kind=realtype) :: tmp
     real(kind=realtype) :: tmpd
     real(kind=realtype) :: tmp0
@@ -594,48 +712,123 @@ subroutine bceddywall_b(nn)
 ! in the halo cells.
     select case  (bcfaceid(nn)) 
     case (imin) 
+      do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+        do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+          call pushreal8(fact)
+          call saroughfact(2, i, j, fact)
+          call pushreal8(rev(1, i, j))
+          rev(1, i, j) = fact*rev(2, i, j)
+        end do
+      end do
       do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
         do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
-          revd(2, i, j) = revd(2, i, j) - revd(1, i, j)
+          call popreal8(rev(1, i, j))
+          factd = rev(2, i, j)*revd(1, i, j)
+          revd(2, i, j) = revd(2, i, j) + fact*revd(1, i, j)
           revd(1, i, j) = 0.0_8
+          call popreal8(fact)
+          call saroughfact_b(2, i, j, fact, factd)
         end do
       end do
     case (imax) 
+      do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+        do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+          call pushreal8(fact)
+          call saroughfact(il, i, j, fact)
+          tmp = fact*rev(il, i, j)
+          call pushreal8(rev(ie, i, j))
+          rev(ie, i, j) = tmp
+        end do
+      end do
       do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
         do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+          call popreal8(rev(ie, i, j))
           tmpd = revd(ie, i, j)
           revd(ie, i, j) = 0.0_8
-          revd(il, i, j) = revd(il, i, j) - tmpd
+          factd = rev(il, i, j)*tmpd
+          revd(il, i, j) = revd(il, i, j) + fact*tmpd
+          call popreal8(fact)
+          call saroughfact_b(il, i, j, fact, factd)
         end do
       end do
     case (jmin) 
+      do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+        do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+          call pushreal8(fact)
+          call saroughfact(i, 2, j, fact)
+          call pushreal8(rev(i, 1, j))
+          rev(i, 1, j) = fact*rev(i, 2, j)
+        end do
+      end do
       do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
         do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
-          revd(i, 2, j) = revd(i, 2, j) - revd(i, 1, j)
+          call popreal8(rev(i, 1, j))
+          factd = rev(i, 2, j)*revd(i, 1, j)
+          revd(i, 2, j) = revd(i, 2, j) + fact*revd(i, 1, j)
           revd(i, 1, j) = 0.0_8
+          call popreal8(fact)
+          call saroughfact_b(i, 2, j, fact, factd)
         end do
       end do
     case (jmax) 
+      do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+        do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+          call pushreal8(fact)
+          call saroughfact(i, jl, j, fact)
+          tmp0 = fact*rev(i, jl, j)
+          call pushreal8(rev(i, je, j))
+          rev(i, je, j) = tmp0
+        end do
+      end do
       do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
         do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+          call popreal8(rev(i, je, j))
           tmpd0 = revd(i, je, j)
           revd(i, je, j) = 0.0_8
-          revd(i, jl, j) = revd(i, jl, j) - tmpd0
+          factd = rev(i, jl, j)*tmpd0
+          revd(i, jl, j) = revd(i, jl, j) + fact*tmpd0
+          call popreal8(fact)
+          call saroughfact_b(i, jl, j, fact, factd)
         end do
       end do
     case (kmin) 
+      do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+        do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+          call pushreal8(fact)
+          call saroughfact(i, j, 2, fact)
+          call pushreal8(rev(i, j, 1))
+          rev(i, j, 1) = fact*rev(i, j, 2)
+        end do
+      end do
       do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
         do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
-          revd(i, j, 2) = revd(i, j, 2) - revd(i, j, 1)
+          call popreal8(rev(i, j, 1))
+          factd = rev(i, j, 2)*revd(i, j, 1)
+          revd(i, j, 2) = revd(i, j, 2) + fact*revd(i, j, 1)
           revd(i, j, 1) = 0.0_8
+          call popreal8(fact)
+          call saroughfact_b(i, j, 2, fact, factd)
         end do
       end do
     case (kmax) 
+      do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+        do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+          call pushreal8(fact)
+          call saroughfact(i, j, kl, fact)
+          tmp1 = fact*rev(i, j, kl)
+          call pushreal8(rev(i, j, ke))
+          rev(i, j, ke) = tmp1
+        end do
+      end do
       do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
         do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+          call popreal8(rev(i, j, ke))
           tmpd1 = revd(i, j, ke)
           revd(i, j, ke) = 0.0_8
-          revd(i, j, kl) = revd(i, j, kl) - tmpd1
+          factd = rev(i, j, kl)*tmpd1
+          revd(i, j, kl) = revd(i, j, kl) + fact*tmpd1
+          call popreal8(fact)
+          call saroughfact_b(i, j, kl, fact, factd)
         end do
       end do
     end select
@@ -659,6 +852,7 @@ subroutine bceddywall(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j
+    real(kind=realtype) :: fact
 ! determine the face id on which the subface is located and
 ! loop over the faces of the subface and set the eddy viscosity
 ! in the halo cells.
@@ -666,49 +860,58 @@ subroutine bceddywall(nn)
     case (imin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(1, i, j) = -rev(2, i, j)
+          call saroughfact(2, i, j, fact)
+          rev(1, i, j) = fact*rev(2, i, j)
         end do
       end do
     case (imax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(ie, i, j) = -rev(il, i, j)
+          call saroughfact(il, i, j, fact)
+          rev(ie, i, j) = fact*rev(il, i, j)
         end do
       end do
     case (jmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, 1, j) = -rev(i, 2, j)
+          call saroughfact(i, 2, j, fact)
+          rev(i, 1, j) = fact*rev(i, 2, j)
         end do
       end do
     case (jmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, je, j) = -rev(i, jl, j)
+          call saroughfact(i, jl, j, fact)
+          rev(i, je, j) = fact*rev(i, jl, j)
         end do
       end do
     case (kmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, j, 1) = -rev(i, j, 2)
+          call saroughfact(i, j, 2, fact)
+          rev(i, j, 1) = fact*rev(i, j, 2)
         end do
       end do
     case (kmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, j, ke) = -rev(i, j, kl)
+          call saroughfact(i, j, kl, fact)
+          rev(i, j, ke) = fact*rev(i, j, kl)
         end do
       end do
     end select
   end subroutine bceddywall
 
 !  differentiation of bcturbfarfield in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: winf *bvtj1 *bvtj2 *bvtk1 *bvtk2
-!                *bvti1 *bvti2
-!   with respect to varying inputs: winf *bvtj1 *bvtj2 *bvtk1 *bvtk2
-!                *bvti1 *bvti2
-!   plus diff mem management of: bvtj1:in bvtj2:in bvtk1:in bvtk2:in
-!                bvti1:in bvti2:in
+!   gradient     of useful results: winf *bvtj1 *bvtj2 *bmtk1 *bmtk2
+!                *bvtk1 *bvtk2 *bmti1 *bmti2 *bvti1 *bvti2 *bmtj1
+!                *bmtj2
+!   with respect to varying inputs: winf *bvtj1 *bvtj2 *bmtk1 *bmtk2
+!                *bvtk1 *bvtk2 *bmti1 *bmti2 *bvti1 *bvti2 *bmtj1
+!                *bmtj2
+!   plus diff mem management of: bvtj1:in bvtj2:in bmtk1:in bmtk2:in
+!                bvtk1:in bvtk2:in bmti1:in bmti2:in bvti1:in bvti2:in
+!                bmtj1:in bmtj2:in
   subroutine bcturbfarfield_b(nn)
 !
 !       bcturbfarfield applies the implicit treatment of the
@@ -745,6 +948,26 @@ subroutine bcturbfarfield_b(nn)
 ! determine whether we are dealing with an inflow or
 ! outflow boundary here.
         if (dot .gt. zero) then
+! outflow. simply extrapolation or zero neumann bc
+! of the turbulent variables.
+          do l=nt1,nt2
+            select case  (bcfaceid(nn)) 
+            case (imin) 
+              call pushcontrol3b(5)
+            case (imax) 
+              call pushcontrol3b(4)
+            case (jmin) 
+              call pushcontrol3b(3)
+            case (jmax) 
+              call pushcontrol3b(2)
+            case (kmin) 
+              call pushcontrol3b(1)
+            case (kmax) 
+              call pushcontrol3b(0)
+            case default
+              call pushcontrol3b(6)
+            end select
+          end do
           call pushcontrol1b(1)
         else
 ! inflow. turbulent variables are prescribed.
@@ -800,6 +1023,27 @@ subroutine bcturbfarfield_b(nn)
               bvti1d(i, j, l) = 0.0_8
             end if
           end do
+        else
+          do l=nt2,nt1,-1
+            call popcontrol3b(branch)
+            if (branch .lt. 3) then
+              if (branch .eq. 0) then
+                bmtk2d(i, j, l, l) = 0.0_8
+              else if (branch .eq. 1) then
+                bmtk1d(i, j, l, l) = 0.0_8
+              else
+                bmtj2d(i, j, l, l) = 0.0_8
+              end if
+            else if (branch .lt. 5) then
+              if (branch .eq. 3) then
+                bmtj1d(i, j, l, l) = 0.0_8
+              else
+                bmti2d(i, j, l, l) = 0.0_8
+              end if
+            else if (branch .eq. 5) then
+              bmti1d(i, j, l, l) = 0.0_8
+            end if
+          end do
         end if
       end do
     end do
@@ -1111,6 +1355,86 @@ subroutine bcturboutflow(nn)
     end do
   end subroutine bcturboutflow
 
+!  differentiation of bcturbsymm in reverse (adjoint) mode (with options noisize i4 dr8 r8):
+!   gradient     of useful results: *bmtk1 *bmtk2 *bmti1 *bmti2
+!                *bmtj1 *bmtj2
+!   with respect to varying inputs: *bmtk1 *bmtk2 *bmti1 *bmti2
+!                *bmtj1 *bmtj2
+!   plus diff mem management of: bmtk1:in bmtk2:in bmti1:in bmti2:in
+!                bmtj1:in bmtj2:in
+  subroutine bcturbsymm_b(nn)
+!
+!       bcturbsymm applies the implicit treatment of the symmetry
+!       boundary condition (or inviscid wall) to subface nn. as the
+!       symmetry boundary condition is independent of the turbulence
+!       model, this routine is valid for all models. it is assumed
+!       that the pointers in blockpointers are already set to the
+!       correct block on the correct grid level.
+!
+    use constants
+    use blockpointers
+    use flowvarrefstate
+    implicit none
+!
+!      subroutine arguments.
+!
+    integer(kind=inttype), intent(in) :: nn
+!
+!      local variables.
+!
+    integer(kind=inttype) :: i, j, l
+    integer :: branch
+! loop over the faces of the subfaces and set the values of bmt
+! for an implicit treatment. for a symmetry face this means
+! that the halo value is set to the internal value.
+    do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
+      do i=bcdata(nn)%icbeg,bcdata(nn)%icend
+        do l=nt1,nt2
+          select case  (bcfaceid(nn)) 
+          case (imin) 
+            call pushcontrol3b(5)
+          case (imax) 
+            call pushcontrol3b(4)
+          case (jmin) 
+            call pushcontrol3b(3)
+          case (jmax) 
+            call pushcontrol3b(2)
+          case (kmin) 
+            call pushcontrol3b(1)
+          case (kmax) 
+            call pushcontrol3b(0)
+          case default
+            call pushcontrol3b(6)
+          end select
+        end do
+      end do
+    end do
+    do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+      do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+        do l=nt2,nt1,-1
+          call popcontrol3b(branch)
+          if (branch .lt. 3) then
+            if (branch .eq. 0) then
+              bmtk2d(i, j, l, l) = 0.0_8
+            else if (branch .eq. 1) then
+              bmtk1d(i, j, l, l) = 0.0_8
+            else
+              bmtj2d(i, j, l, l) = 0.0_8
+            end if
+          else if (branch .lt. 5) then
+            if (branch .eq. 3) then
+              bmtj1d(i, j, l, l) = 0.0_8
+            else
+              bmti2d(i, j, l, l) = 0.0_8
+            end if
+          else if (branch .eq. 5) then
+            bmti1d(i, j, l, l) = 0.0_8
+          end if
+        end do
+      end do
+    end do
+  end subroutine bcturbsymm_b
+
   subroutine bcturbsymm(nn)
 !
 !       bcturbsymm applies the implicit treatment of the symmetry
@@ -1158,15 +1482,20 @@ subroutine bcturbsymm(nn)
   end subroutine bcturbsymm
 
 !  differentiation of bcturbtreatment in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: winf *bvtj1 *bvtj2 *w *rlv
-!                *bvtk1 *bvtk2 *d2wall *bvti1 *bvti2
-!   with respect to varying inputs: winf *bvtj1 *bvtj2 *w *rlv
-!                *bvtk1 *bvtk2 *d2wall *bvti1 *bvti2
+!   gradient     of useful results: winf *bvtj1 *bvtj2 *w *bmtk1
+!                *rlv *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2
+!                *bvti1 *bvti2 *bmtj1 *bmtj2
+!   with respect to varying inputs: winf *bvtj1 *bvtj2 *w *bmtk1
+!                *rlv *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2
+!                *bvti1 *bvti2 *bmtj1 *bmtj2
 !   rw status of diff variables: winf:incr *bvtj1:in-out *bvtj2:in-out
-!                *w:incr *rlv:incr *bvtk1:in-out *bvtk2:in-out
-!                *d2wall:incr *bvti1:in-out *bvti2:in-out
-!   plus diff mem management of: bvtj1:in bvtj2:in w:in rlv:in
-!                bvtk1:in bvtk2:in d2wall:in bvti1:in bvti2:in
+!                *w:incr *bmtk1:in-out *rlv:incr *bmtk2:in-out
+!                *bvtk1:in-out *bvtk2:in-out *d2wall:incr *bmti1:in-out
+!                *bmti2:in-out *bvti1:in-out *bvti2:in-out *bmtj1:in-out
+!                *bmtj2:in-out
+!   plus diff mem management of: bvtj1:in bvtj2:in w:in bmtk1:in
+!                rlv:in bmtk2:in bvtk1:in bvtk2:in d2wall:in bmti1:in
+!                bmti2:in bvti1:in bvti2:in bmtj1:in bmtj2:in
   subroutine bcturbtreatment_b()
 !
 !       bcturbtreatment sets the arrays bmti1, bvti1, etc, such that
@@ -1194,21 +1523,21 @@ subroutine bcturbtreatment_b()
 ! determine the kind of boundary condition for this subface.
       select case  (bctype(nn)) 
       case (nswalladiabatic, nswallisothermal) 
-        call pushcontrol2b(2)
+        call pushcontrol3b(3)
       case (symm, symmpolar, eulerwall) 
-        call pushcontrol2b(3)
+        call pushcontrol3b(2)
       case (farfield) 
-        call pushcontrol2b(1)
+        call pushcontrol3b(1)
       case (slidinginterface, oversetouterbound, domaininterfaceall, &
 &     domaininterfacerhouvw, domaininterfacep, domaininterfacerho, &
 &     domaininterfacetotal) 
-        call pushcontrol2b(0)
+        call pushcontrol3b(0)
       case default
-        call pushcontrol2b(3)
+        call pushcontrol3b(4)
       end select
     end do bocos
     do nn=nbocos,1,-1
-      call popcontrol2b(branch)
+      call popcontrol3b(branch)
       if (branch .lt. 2) then
         if (branch .eq. 0) then
           call bcturbinterface_b(nn)
@@ -1216,6 +1545,8 @@ subroutine bcturbtreatment_b()
           call bcturbfarfield_b(nn)
         end if
       else if (branch .eq. 2) then
+        call bcturbsymm_b(nn)
+      else if (branch .eq. 3) then
         call bcturbwall_b(nn)
       end if
     end do
@@ -1224,6 +1555,10 @@ subroutine bcturbtreatment_b()
         do l=nt2,nt1,-1
           bvtk2d(i, j, l) = 0.0_8
           bvtk1d(i, j, l) = 0.0_8
+          do m=nt2,nt1,-1
+            bmtk2d(i, j, l, m) = 0.0_8
+            bmtk1d(i, j, l, m) = 0.0_8
+          end do
         end do
       end do
     end do
@@ -1232,6 +1567,10 @@ subroutine bcturbtreatment_b()
         do l=nt2,nt1,-1
           bvtj2d(i, k, l) = 0.0_8
           bvtj1d(i, k, l) = 0.0_8
+          do m=nt2,nt1,-1
+            bmtj2d(i, k, l, m) = 0.0_8
+            bmtj1d(i, k, l, m) = 0.0_8
+          end do
         end do
       end do
     end do
@@ -1240,6 +1579,10 @@ subroutine bcturbtreatment_b()
         do l=nt2,nt1,-1
           bvti2d(j, k, l) = 0.0_8
           bvti1d(j, k, l) = 0.0_8
+          do m=nt2,nt1,-1
+            bmti2d(j, k, l, m) = 0.0_8
+            bmti1d(j, k, l, m) = 0.0_8
+          end do
         end do
       end do
     end do
@@ -1338,12 +1681,15 @@ subroutine bcturbtreatment()
   end subroutine bcturbtreatment
 
 !  differentiation of bcturbwall in reverse (adjoint) mode (with options noisize i4 dr8 r8):
-!   gradient     of useful results: *bvtj1 *bvtj2 *w *rlv *bvtk1
-!                *bvtk2 *d2wall *bvti1 *bvti2
-!   with respect to varying inputs: *bvtj1 *bvtj2 *w *rlv *bvtk1
-!                *bvtk2 *d2wall *bvti1 *bvti2
-!   plus diff mem management of: bvtj1:in bvtj2:in w:in rlv:in
-!                bvtk1:in bvtk2:in d2wall:in bvti1:in bvti2:in
+!   gradient     of useful results: *bvtj1 *bvtj2 *w *bmtk1 *rlv
+!                *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2 *bvti1
+!                *bvti2 *bmtj1 *bmtj2
+!   with respect to varying inputs: *bvtj1 *bvtj2 *w *bmtk1 *rlv
+!                *bmtk2 *bvtk1 *bvtk2 *d2wall *bmti1 *bmti2 *bvti1
+!                *bvti2 *bmtj1 *bmtj2
+!   plus diff mem management of: bvtj1:in bvtj2:in w:in bmtk1:in
+!                rlv:in bmtk2:in bvtk1:in bvtk2:in d2wall:in bmti1:in
+!                bmti2:in bvti1:in bvti2:in bmtj1:in bmtj2:in
   subroutine bcturbwall_b(nn)
 !
 !       bcturbwall applies the implicit treatment of the viscous
@@ -1366,8 +1712,8 @@ subroutine bcturbwall_b(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j, ii, jj, iimax, jjmax
-    real(kind=realtype) :: tmpd, tmpe, tmpf, nu
-    real(kind=realtype) :: tmpdd, nud
+    real(kind=realtype) :: tmpd, tmpe, tmpf, nu, fact
+    real(kind=realtype) :: tmpdd, nud, factd
     real(kind=realtype), dimension(:, :, :, :), pointer :: bmt
     real(kind=realtype), dimension(:, :, :), pointer :: bvt, ww2
     real(kind=realtype), dimension(:, :), pointer :: rlv2, dd2wall
@@ -1393,7 +1739,59 @@ subroutine bcturbwall_b(nn)
 ! implicit treatment.
     select case  (turbmodel) 
     case (spalartallmaras, spalartallmarasedwards) 
-
+! spalart-allmaras type of model. value at the wall is zero,
+! so simply negate the internal value.
+      select case  (bcfaceid(nn)) 
+      case (imin) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            factd = -bmti1d(i, j, itu1, itu1)
+            bmti1d(i, j, itu1, itu1) = 0.0_8
+            call saroughfact_b(2, i, j, fact, factd)
+          end do
+        end do
+      case (imax) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            factd = -bmti2d(i, j, itu1, itu1)
+            bmti2d(i, j, itu1, itu1) = 0.0_8
+            call saroughfact_b(il, i, j, fact, factd)
+          end do
+        end do
+      case (jmin) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            factd = -bmtj1d(i, j, itu1, itu1)
+            bmtj1d(i, j, itu1, itu1) = 0.0_8
+            call saroughfact_b(i, 2, j, fact, factd)
+          end do
+        end do
+      case (jmax) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            factd = -bmtj2d(i, j, itu1, itu1)
+            bmtj2d(i, j, itu1, itu1) = 0.0_8
+            call saroughfact_b(i, jl, j, fact, factd)
+          end do
+        end do
+      case (kmin) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            factd = -bmtk1d(i, j, itu1, itu1)
+            bmtk1d(i, j, itu1, itu1) = 0.0_8
+            call saroughfact_b(i, j, 2, fact, factd)
+          end do
+        end do
+      case (kmax) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            factd = -bmtk2d(i, j, itu1, itu1)
+            bmtk2d(i, j, itu1, itu1) = 0.0_8
+            call saroughfact_b(i, j, kl, fact, factd)
+          end do
+        end do
+      end select
+!        ================================================================
     case (komegawilcox, komegamodified, mentersst) 
 ! k-omega type of models. k is zero on the wall and thus the
 ! halo value is the negative of the first internal cell.
@@ -1449,6 +1847,8 @@ subroutine bcturbwall_b(nn)
             bvti1d(i, j, itu2) = 0.0_8
             nud = tmpd*tempd
             tmpdd = nu*tempd
+            bmti1d(i, j, itu2, itu2) = 0.0_8
+            bmti1d(i, j, itu1, itu1) = 0.0_8
             temp = rkwbeta1*(d2wall(2, ii, jj)*d2wall(2, ii, jj))
             d2walld(2, ii, jj) = d2walld(2, ii, jj) - 2*d2wall(2, ii, jj&
 &             )*rkwbeta1*one*tmpdd/temp**2
@@ -1513,6 +1913,8 @@ subroutine bcturbwall_b(nn)
             bvti2d(i, j, itu2) = 0.0_8
             nud = tmpd*tempd
             tmpdd = nu*tempd
+            bmti2d(i, j, itu2, itu2) = 0.0_8
+            bmti2d(i, j, itu1, itu1) = 0.0_8
             temp = rkwbeta1*(d2wall(il, ii, jj)*d2wall(il, ii, jj))
             d2walld(il, ii, jj) = d2walld(il, ii, jj) - 2*d2wall(il, ii&
 &             , jj)*rkwbeta1*one*tmpdd/temp**2
@@ -1577,6 +1979,8 @@ subroutine bcturbwall_b(nn)
             bvtj1d(i, j, itu2) = 0.0_8
             nud = tmpd*tempd
             tmpdd = nu*tempd
+            bmtj1d(i, j, itu2, itu2) = 0.0_8
+            bmtj1d(i, j, itu1, itu1) = 0.0_8
             temp = rkwbeta1*(d2wall(ii, 2, jj)*d2wall(ii, 2, jj))
             d2walld(ii, 2, jj) = d2walld(ii, 2, jj) - 2*d2wall(ii, 2, jj&
 &             )*rkwbeta1*one*tmpdd/temp**2
@@ -1641,6 +2045,8 @@ subroutine bcturbwall_b(nn)
             bvtj2d(i, j, itu2) = 0.0_8
             nud = tmpd*tempd
             tmpdd = nu*tempd
+            bmtj2d(i, j, itu2, itu2) = 0.0_8
+            bmtj2d(i, j, itu1, itu1) = 0.0_8
             temp = rkwbeta1*(d2wall(ii, jl, jj)*d2wall(ii, jl, jj))
             d2walld(ii, jl, jj) = d2walld(ii, jl, jj) - 2*d2wall(ii, jl&
 &             , jj)*rkwbeta1*one*tmpdd/temp**2
@@ -1705,6 +2111,8 @@ subroutine bcturbwall_b(nn)
             bvtk1d(i, j, itu2) = 0.0_8
             nud = tmpd*tempd
             tmpdd = nu*tempd
+            bmtk1d(i, j, itu2, itu2) = 0.0_8
+            bmtk1d(i, j, itu1, itu1) = 0.0_8
             temp = rkwbeta1*(d2wall(ii, jj, 2)*d2wall(ii, jj, 2))
             d2walld(ii, jj, 2) = d2walld(ii, jj, 2) - 2*d2wall(ii, jj, 2&
 &             )*rkwbeta1*one*tmpdd/temp**2
@@ -1769,6 +2177,8 @@ subroutine bcturbwall_b(nn)
             bvtk2d(i, j, itu2) = 0.0_8
             nud = tmpd*tempd
             tmpdd = nu*tempd
+            bmtk2d(i, j, itu2, itu2) = 0.0_8
+            bmtk2d(i, j, itu1, itu1) = 0.0_8
             temp = rkwbeta1*(d2wall(ii, jj, kl)*d2wall(ii, jj, kl))
             d2walld(ii, jj, kl) = d2walld(ii, jj, kl) - 2*d2wall(ii, jj&
 &             , kl)*rkwbeta1*one*tmpdd/temp**2
@@ -1791,6 +2201,54 @@ subroutine bcturbwall_b(nn)
           end if
         end do
       end select
+!        ================================================================
+    case (ktau) 
+! k-tau model. both k and tau are zero at the wall, so the
+! negative value of the internal cell is taken for the halo.
+      select case  (bcfaceid(nn)) 
+      case (imin) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            bmti1d(i, j, itu2, itu2) = 0.0_8
+            bmti1d(i, j, itu1, itu1) = 0.0_8
+          end do
+        end do
+      case (imax) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            bmti2d(i, j, itu2, itu2) = 0.0_8
+            bmti2d(i, j, itu1, itu1) = 0.0_8
+          end do
+        end do
+      case (jmin) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            bmtj1d(i, j, itu2, itu2) = 0.0_8
+            bmtj1d(i, j, itu1, itu1) = 0.0_8
+          end do
+        end do
+      case (jmax) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            bmtj2d(i, j, itu2, itu2) = 0.0_8
+            bmtj2d(i, j, itu1, itu1) = 0.0_8
+          end do
+        end do
+      case (kmin) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            bmtk1d(i, j, itu2, itu2) = 0.0_8
+            bmtk1d(i, j, itu1, itu1) = 0.0_8
+          end do
+        end do
+      case (kmax) 
+        do j=bcdata(nn)%jcend,bcdata(nn)%jcbeg,-1
+          do i=bcdata(nn)%icend,bcdata(nn)%icbeg,-1
+            bmtk2d(i, j, itu2, itu2) = 0.0_8
+            bmtk2d(i, j, itu1, itu1) = 0.0_8
+          end do
+        end do
+      end select
 !        ================================================================
     end select
   end subroutine bcturbwall_b
@@ -1817,7 +2275,7 @@ subroutine bcturbwall(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j, ii, jj, iimax, jjmax
-    real(kind=realtype) :: tmpd, tmpe, tmpf, nu
+    real(kind=realtype) :: tmpd, tmpe, tmpf, nu, fact
     real(kind=realtype), dimension(:, :, :, :), pointer :: bmt
     real(kind=realtype), dimension(:, :, :), pointer :: bvt, ww2
     real(kind=realtype), dimension(:, :), pointer :: rlv2, dd2wall
@@ -1846,37 +2304,43 @@ subroutine bcturbwall(nn)
       case (imin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti1(i, j, itu1, itu1) = one
+            call saroughfact(2, i, j, fact)
+            bmti1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (imax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti2(i, j, itu1, itu1) = one
+            call saroughfact(il, i, j, fact)
+            bmti2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj1(i, j, itu1, itu1) = one
+            call saroughfact(i, 2, j, fact)
+            bmtj1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj2(i, j, itu1, itu1) = one
+            call saroughfact(i, jl, j, fact)
+            bmtj2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk1(i, j, itu1, itu1) = one
+            call saroughfact(i, j, 2, fact)
+            bmtk1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk2(i, j, itu1, itu1) = one
+            call saroughfact(i, j, kl, fact)
+            bmtk2(i, j, itu1, itu1) = -fact
           end do
         end do
       end select
@@ -2547,5 +3011,55 @@ subroutine turbbcnswall(secondhalo)
     end do bocos
   end subroutine turbbcnswall
 
+!  differentiation of saroughfact in reverse (adjoint) mode (with options noisize i4 dr8 r8):
+!   gradient     of useful results: *d2wall fact
+!   with respect to varying inputs: *d2wall
+!   plus diff mem management of: d2wall:in
+  subroutine saroughfact_b(i, j, k, fact, factd)
+! returns either the regular sa-boundary condition
+! or the modified roughness-boundary condition
+    use constants
+    use inputphysics, only : useroughsa
+    use blockpointers, only : ks, d2wall, d2walld, il, jl, kl
+    implicit none
+! local variablse
+    integer(kind=inttype), intent(in) :: i, j, k
+    real(kind=realtype) :: fact
+    real(kind=realtype) :: factd
+    real(kind=realtype) :: temp
+    if (useroughsa) then
+      temp = ks(i, j, k) + d2wall(i, j, k)/0.03_realtype
+      d2walld(i, j, k) = d2walld(i, j, k) - (1.0/(0.03_realtype*temp)+(&
+&       ks(i, j, k)-d2wall(i, j, k)/0.03_realtype)/(0.03_realtype*temp**&
+&       2))*factd
+    end if
+  end subroutine saroughfact_b
+
+  subroutine saroughfact(i, j, k, fact)
+! returns either the regular sa-boundary condition
+! or the modified roughness-boundary condition
+    use constants
+    use inputphysics, only : useroughsa
+    use blockpointers, only : ks, d2wall, il, jl, kl
+    implicit none
+! local variablse
+    integer(kind=inttype), intent(in) :: i, j, k
+    real(kind=realtype), intent(out) :: fact
+    if (.not.useroughsa) then
+      fact = -one
+      return
+    else
+      fact = (ks(i, j, k)-d2wall(i, j, k)/0.03_realtype)/(ks(i, j, k)+&
+&       d2wall(i, j, k)/0.03_realtype)
+      if (ks(i, j, k) .eq. 0.01 .or. d2wall(i, j, k) .eq. 0.01) print*, &
+&                                                               i, j, k&
+&                                                               , fact, &
+&                                                               d2wall(i&
+&                                                               , j, k)&
+&                                                               , ks(i, &
+&                                                               j, k)
+    end if
+  end subroutine saroughfact
+
 end module turbbcroutines_b
 
diff --git a/src/adjoint/outputReverse/turbUtils_b.f90 b/src/adjoint/outputReverse/turbUtils_b.f90
index 50ebc309c..d42719519 100644
--- a/src/adjoint/outputReverse/turbUtils_b.f90
+++ b/src/adjoint/outputReverse/turbUtils_b.f90
@@ -906,8 +906,9 @@ subroutine turbadvection_b(madv, nadv, offset, qq)
     use constants
     use blockpointers, only : nx, ny, nz, il, jl, kl, vol, vold, &
 &   sfacei, sfaceid, sfacej, sfacejd, sfacek, sfacekd, w, wd, si, sid, &
-&   sj, sjd, sk, skd, addgridvelocities, bmti1, bmti2, bmtj1, bmtj2, &
-&   bmtk1, bmtk2, scratch, scratchd
+&   sj, sjd, sk, skd, addgridvelocities, bmti1, bmti1d, bmti2, bmti2d, &
+&   bmtj1, bmtj1d, bmtj2, bmtj2d, bmtk1, bmtk1d, bmtk2, bmtk2d, scratch,&
+&   scratchd
     use inputdiscretization, only : orderturb
     use iteration, only : groundlevel
     use turbmod, only : secondord
diff --git a/src/adjoint/outputReverse/utils_b.f90 b/src/adjoint/outputReverse/utils_b.f90
index 3ad24e4bb..9d3f1d1f8 100644
--- a/src/adjoint/outputReverse/utils_b.f90
+++ b/src/adjoint/outputReverse/utils_b.f90
@@ -725,12 +725,13 @@ subroutine setbcpointers(nn, spatialpointers)
     use blockpointers, only : w, p, rlv, rev, gamma, x, d2wall, si, sj&
 &   , sk, s, globalcell, bcdata, nx, il, ie, ib, ny, jl, je, jb, nz, kl,&
 &   ke, kb, bcfaceid, addgridvelocities, sfacei, sfacej, sfacek, &
-&   addgridvelocities
+&   addgridvelocities, d2wall, ks
     use bcpointers_b, only : ww0, ww1, ww2, ww3, pp0, pp1, pp2, pp3, &
 &   rlv0, rlv1, rlv2, rlv3, rev0, rev1, rev2, rev3, gamma0, gamma1, &
 &   gamma2, gamma3, gcp, xx, ss, ssi, ssj, ssk, dd2wall, sface, istart, &
-&   iend, jstart, jend, isize, jsize
-    use inputphysics, only : cpmodel, equations
+&   iend, jstart, jend, isize, jsize, d2wall0, d2wall1, d2wall2, d2wall3&
+&   , ks0, ks1, ks2, ks3
+    use inputphysics, only : cpmodel, equations, useroughsa
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -765,6 +766,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(2, 1:, 1:)
       rev1 => rev(1, 1:, 1:)
       rev0 => rev(0, 1:, 1:)
+      d2wall3 => d2wall(3, 1:, 1:)
+      d2wall2 => d2wall(2, 1:, 1:)
+      d2wall1 => d2wall(1, 1:, 1:)
+      d2wall0 => d2wall(0, 1:, 1:)
+      if (useroughsa) then
+        ks3 => ks(3, 1:, 1:)
+        ks2 => ks(2, 1:, 1:)
+        ks1 => ks(1, 1:, 1:)
+        ks0 => ks(0, 1:, 1:)
+      end if
       gamma3 => gamma(3, 1:, 1:)
       gamma2 => gamma(2, 1:, 1:)
       gamma1 => gamma(1, 1:, 1:)
@@ -788,6 +799,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(il, 1:, 1:)
       rev1 => rev(ie, 1:, 1:)
       rev0 => rev(ib, 1:, 1:)
+      d2wall3 => d2wall(nx, 1:, 1:)
+      d2wall2 => d2wall(il, 1:, 1:)
+      d2wall1 => d2wall(ie, 1:, 1:)
+      d2wall0 => d2wall(ib, 1:, 1:)
+      if (useroughsa) then
+        ks3 => ks(nx, 1:, 1:)
+        ks2 => ks(il, 1:, 1:)
+        ks1 => ks(ie, 1:, 1:)
+        ks0 => ks(ib, 1:, 1:)
+      end if
       gamma3 => gamma(nx, 1:, 1:)
       gamma2 => gamma(il, 1:, 1:)
       gamma1 => gamma(ie, 1:, 1:)
@@ -811,6 +832,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 2, 1:)
       rev1 => rev(1:, 1, 1:)
       rev0 => rev(1:, 0, 1:)
+      d2wall3 => d2wall(1:, 3, 1:)
+      d2wall2 => d2wall(1:, 2, 1:)
+      d2wall1 => d2wall(1:, 1, 1:)
+      d2wall0 => d2wall(1:, 0, 1:)
+      if (useroughsa) then
+        ks3 => ks(1:, 3, 1:)
+        ks2 => ks(1:, 2, 1:)
+        ks1 => ks(1:, 1, 1:)
+        ks0 => ks(1:, 0, 1:)
+      end if
       gamma3 => gamma(1:, 3, 1:)
       gamma2 => gamma(1:, 2, 1:)
       gamma1 => gamma(1:, 1, 1:)
@@ -834,6 +865,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, jl, 1:)
       rev1 => rev(1:, je, 1:)
       rev0 => rev(1:, jb, 1:)
+      d2wall3 => d2wall(1:, ny, 1:)
+      d2wall2 => d2wall(1:, jl, 1:)
+      d2wall1 => d2wall(1:, je, 1:)
+      d2wall0 => d2wall(1:, jb, 1:)
+      if (useroughsa) then
+        ks3 => ks(1:, ny, 1:)
+        ks2 => ks(1:, jl, 1:)
+        ks1 => ks(1:, je, 1:)
+        ks0 => ks(1:, jb, 1:)
+      end if
       gamma3 => gamma(1:, ny, 1:)
       gamma2 => gamma(1:, jl, 1:)
       gamma1 => gamma(1:, je, 1:)
@@ -857,6 +898,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 1:, 2)
       rev1 => rev(1:, 1:, 1)
       rev0 => rev(1:, 1:, 0)
+      d2wall3 => d2wall(1:, 1:, 3)
+      d2wall2 => d2wall(1:, 1:, 2)
+      d2wall1 => d2wall(1:, 1:, 1)
+      d2wall0 => d2wall(1:, 1:, 0)
+      if (useroughsa) then
+        ks3 => ks(1:, 1:, 3)
+        ks2 => ks(1:, 1:, 2)
+        ks1 => ks(1:, 1:, 1)
+        ks0 => ks(1:, 1:, 0)
+      end if
       gamma3 => gamma(1:, 1:, 3)
       gamma2 => gamma(1:, 1:, 2)
       gamma1 => gamma(1:, 1:, 1)
@@ -880,6 +931,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 1:, kl)
       rev1 => rev(1:, 1:, ke)
       rev0 => rev(1:, 1:, kb)
+      d2wall3 => d2wall(1:, 1:, nz)
+      d2wall2 => d2wall(1:, 1:, kl)
+      d2wall1 => d2wall(1:, 1:, ke)
+      d2wall0 => d2wall(1:, 1:, kb)
+      if (useroughsa) then
+        ks3 => ks(1:, 1:, nz)
+        ks2 => ks(1:, 1:, kl)
+        ks1 => ks(1:, 1:, ke)
+        ks0 => ks(1:, 1:, kb)
+      end if
       gamma3 => gamma(1:, 1:, nz)
       gamma2 => gamma(1:, 1:, kl)
       gamma1 => gamma(1:, 1:, ke)
diff --git a/src/adjoint/outputReverseFast/BCRoutines_fast_b.f90 b/src/adjoint/outputReverseFast/BCRoutines_fast_b.f90
index 37f0e540f..cd24ca801 100644
--- a/src/adjoint/outputReverseFast/BCRoutines_fast_b.f90
+++ b/src/adjoint/outputReverseFast/BCRoutines_fast_b.f90
@@ -170,8 +170,10 @@ subroutine bcsymm1sthalo(nn)
     use constants
     use blockpointers, only : bcdata
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers, only : gamma1, gamma2, ww1, ww2, pp1, pp2, rlv1, &
-&   rlv2, istart, jstart, isize, jsize, rev1, rev2
+&   rlv2, istart, jstart, isize, jsize, rev1, rev2, d2wall1, d2wall2, &
+&   ks1, ks2
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -201,6 +203,8 @@ subroutine bcsymm1sthalo(nn)
 ! laminar and eddy viscosity in the halo.
       gamma1(i, j) = gamma2(i, j)
       pp1(i, j) = pp2(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) rlv1(i, j) = rlv2(i, j)
       if (eddymodel) rev1(i, j) = rev2(i, j)
     end do
@@ -213,8 +217,10 @@ subroutine bcsymm2ndhalo(nn)
     use constants
     use blockpointers, only : bcdata
     use flowvarrefstate, only : viscous, eddymodel
+    use inputphysics, only : useroughsa
     use bcpointers, only : gamma0, gamma3, ww0, ww3, pp0, pp3, rlv0, &
-&   rlv3, rev0, rev3, istart, jstart, isize, jsize
+&   rlv3, d2wall0, d2wall3, ks0, ks3, rev0, rev3, istart, jstart, isize,&
+&   jsize
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -241,6 +247,8 @@ subroutine bcsymm2ndhalo(nn)
 ! laminar and eddy viscosity in the halo.
       gamma0(i, j) = gamma3(i, j)
       pp0(i, j) = pp3(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) rlv0(i, j) = rlv3(i, j)
       if (eddymodel) rev0(i, j) = rev3(i, j)
     end do
@@ -254,8 +262,9 @@ subroutine bcsymmpolar1sthalo(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 1st level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers, only : ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, &
-&   xx, istart, jstart, isize, jsize
+&   xx, istart, jstart, isize, jsize, d2wall1, d2wall2, ks1, ks2
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -307,6 +316,8 @@ subroutine bcsymmpolar1sthalo(nn)
 ! set the pressure and possibly the laminar and
 ! eddy viscosity in the halo.
       pp1(i, j) = pp2(i, j)
+      d2wall1(i, j) = d2wall2(i, j)
+      if (useroughsa) ks1(i, j) = ks2(i, j)
       if (viscous) rlv1(i, j) = rlv2(i, j)
       if (eddymodel) rev1(i, j) = rev2(i, j)
     end do
@@ -320,8 +331,9 @@ subroutine bcsymmpolar2ndhalo(nn)
 ! case of a degenerate line, as this line is the axi-symmetric
 ! centerline. this routine does just the 2nd level halo.
     use constants
+    use inputphysics, only : useroughsa
     use bcpointers, only : ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, &
-&   xx, istart, jstart, isize, jsize
+&   d2wall0, d2wall3, ks0, ks3, xx, istart, jstart, isize, jsize
     use flowvarrefstate, only : viscous, eddymodel
     implicit none
 ! subroutine arguments.
@@ -371,6 +383,8 @@ subroutine bcsymmpolar2ndhalo(nn)
 ! set the pressure and possibly the laminar and
 ! eddy viscosity in the halo.
       pp0(i, j) = pp3(i, j)
+      d2wall0(i, j) = d2wall3(i, j)
+      if (useroughsa) ks0(i, j) = ks3(i, j)
       if (viscous) rlv0(i, j) = rlv3(i, j)
       if (eddymodel) rev0(i, j) = rev3(i, j)
     end do
diff --git a/src/adjoint/outputReverseFast/sa_fast_b.f90 b/src/adjoint/outputReverseFast/sa_fast_b.f90
index a19e6d3de..2e320c0a3 100644
--- a/src/adjoint/outputReverseFast/sa_fast_b.f90
+++ b/src/adjoint/outputReverseFast/sa_fast_b.f90
@@ -37,6 +37,7 @@ subroutine sasource_fast_b()
     real(kind=realtype), parameter :: f23=two*third
 ! local variables.
     integer(kind=inttype) :: i, j, k, nn, ii
+    real(kind=realtype) :: distrough
     real(kind=realtype) :: fv1, fv2, ft2
     real(kind=realtype) :: fv1d, fv2d, ft2d
     real(kind=realtype) :: ss, sst, nu, dist2inv, chi, chi2, chi3
@@ -173,12 +174,26 @@ subroutine sasource_fast_b()
 ! and nu) and the functions fv1 and fv2. the latter corrects
 ! the production term near a viscous wall.
         nu = rlv(i, j, k)/w(i, j, k, irho)
-        dist2inv = one/d2wall(i, j, k)**2
         chi = w(i, j, k, itu1)/nu
+        if (.not.useroughsa) then
+          dist2inv = one/d2wall(i, j, k)**2
+        else
+          distrough = d2wall(i, j, k) + 0.03_realtype*ks(i, j, k)
+          dist2inv = one/distrough**2
+          chi = chi + rsacr1*ks(i, j, k)/distrough
+        end if
         chi2 = chi*chi
         chi3 = chi*chi2
         fv1 = chi3/(chi3+cv13)
-        fv2 = one - chi/(one+chi*fv1)
+        if (.not.useroughsa) then
+          fv2 = one - chi/(one+chi*fv1)
+myIntPtr = myIntPtr + 1
+ myIntStack(myIntPtr) = 0
+        else
+          fv2 = one - w(i, j, k, itu1)/(nu+w(i, j, k, itu1)*fv1)
+myIntPtr = myIntPtr + 1
+ myIntStack(myIntPtr) = 1
+        end if
 ! the function ft2, which is designed to keep a laminar
 ! solution laminar. when running in fully turbulent mode
 ! this function should be set to 0.0.
@@ -313,16 +328,29 @@ subroutine sasource_fast_b()
         else
           chi2d = 0.0_8
         end if
-        tempd = -(fv2d/(one+chi*fv1))
-        chid = tempd
-        tempd0 = -(chi*tempd/(one+chi*fv1))
-        fv1d = chi*tempd0
+branch = myIntStack(myIntPtr)
+ myIntPtr = myIntPtr - 1
+        if (branch .eq. 0) then
+          tempd = -(fv2d/(one+chi*fv1))
+          tempd0 = -(chi*tempd/(one+chi*fv1))
+          chid = tempd + fv1*tempd0
+          fv1d = chi*tempd0
+          nud = 0.0_8
+        else
+          temp0 = w(i, j, k, itu1)
+          temp = nu + temp0*fv1
+          tempd = w(i, j, k, itu1)*fv2d/temp**2
+          wd(i, j, k, itu1) = wd(i, j, k, itu1) + fv1*tempd - fv2d/temp
+          nud = tempd
+          fv1d = temp0*tempd
+          chid = 0.0_8
+        end if
         tempd = fv1d/(cv13+chi3)
         chi3d = (1.0-chi3/(cv13+chi3))*tempd
         chi2d = chi2d + chi*chi3d
-        chid = chid + fv1*tempd0 + chi2*chi3d + 2*chi*chi2d
+        chid = chid + chi2*chi3d + 2*chi*chi2d
         wd(i, j, k, itu1) = wd(i, j, k, itu1) + chid/nu
-        nud = -(w(i, j, k, itu1)*chid/nu**2)
+        nud = nud - w(i, j, k, itu1)*chid/nu**2
         temp = w(i, j, k, irho)
         rlvd(i, j, k) = rlvd(i, j, k) + nud/temp
         wd(i, j, k, irho) = wd(i, j, k, irho) - rlv(i, j, k)*nud/temp**2
@@ -444,6 +472,7 @@ subroutine sasource()
     real(kind=realtype), parameter :: f23=two*third
 ! local variables.
     integer(kind=inttype) :: i, j, k, nn, ii
+    real(kind=realtype) :: distrough
     real(kind=realtype) :: fv1, fv2, ft2
     real(kind=realtype) :: ss, sst, nu, dist2inv, chi, chi2, chi3
     real(kind=realtype) :: rr, gg, gg6, termfw, fwsa, term1, term2
@@ -560,12 +589,22 @@ subroutine sasource()
 ! and nu) and the functions fv1 and fv2. the latter corrects
 ! the production term near a viscous wall.
         nu = rlv(i, j, k)/w(i, j, k, irho)
-        dist2inv = one/d2wall(i, j, k)**2
         chi = w(i, j, k, itu1)/nu
+        if (.not.useroughsa) then
+          dist2inv = one/d2wall(i, j, k)**2
+        else
+          distrough = d2wall(i, j, k) + 0.03_realtype*ks(i, j, k)
+          dist2inv = one/distrough**2
+          chi = chi + rsacr1*ks(i, j, k)/distrough
+        end if
         chi2 = chi*chi
         chi3 = chi*chi2
         fv1 = chi3/(chi3+cv13)
-        fv2 = one - chi/(one+chi*fv1)
+        if (.not.useroughsa) then
+          fv2 = one - chi/(one+chi*fv1)
+        else
+          fv2 = one - w(i, j, k, itu1)/(nu+w(i, j, k, itu1)*fv1)
+        end if
 ! the function ft2, which is designed to keep a laminar
 ! solution laminar. when running in fully turbulent mode
 ! this function should be set to 0.0.
diff --git a/src/adjoint/outputReverseFast/turbBCRoutines_fast_b.f90 b/src/adjoint/outputReverseFast/turbBCRoutines_fast_b.f90
index aacda3694..16d4465f5 100644
--- a/src/adjoint/outputReverseFast/turbBCRoutines_fast_b.f90
+++ b/src/adjoint/outputReverseFast/turbBCRoutines_fast_b.f90
@@ -206,6 +206,7 @@ subroutine bceddywall(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j
+    real(kind=realtype) :: fact
 ! determine the face id on which the subface is located and
 ! loop over the faces of the subface and set the eddy viscosity
 ! in the halo cells.
@@ -213,37 +214,43 @@ subroutine bceddywall(nn)
     case (imin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(1, i, j) = -rev(2, i, j)
+          call saroughfact(2, i, j, fact)
+          rev(1, i, j) = fact*rev(2, i, j)
         end do
       end do
     case (imax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(ie, i, j) = -rev(il, i, j)
+          call saroughfact(il, i, j, fact)
+          rev(ie, i, j) = fact*rev(il, i, j)
         end do
       end do
     case (jmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, 1, j) = -rev(i, 2, j)
+          call saroughfact(i, 2, j, fact)
+          rev(i, 1, j) = fact*rev(i, 2, j)
         end do
       end do
     case (jmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, je, j) = -rev(i, jl, j)
+          call saroughfact(i, jl, j, fact)
+          rev(i, je, j) = fact*rev(i, jl, j)
         end do
       end do
     case (kmin) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, j, 1) = -rev(i, j, 2)
+          call saroughfact(i, j, 2, fact)
+          rev(i, j, 1) = fact*rev(i, j, 2)
         end do
       end do
     case (kmax) 
       do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
         do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-          rev(i, j, ke) = -rev(i, j, kl)
+          call saroughfact(i, j, kl, fact)
+          rev(i, j, ke) = fact*rev(i, j, kl)
         end do
       end do
     end select
@@ -632,7 +639,7 @@ subroutine bcturbwall(nn)
 !      local variables.
 !
     integer(kind=inttype) :: i, j, ii, jj, iimax, jjmax
-    real(kind=realtype) :: tmpd, tmpe, tmpf, nu
+    real(kind=realtype) :: tmpd, tmpe, tmpf, nu, fact
     real(kind=realtype), dimension(:, :, :, :), pointer :: bmt
     real(kind=realtype), dimension(:, :, :), pointer :: bvt, ww2
     real(kind=realtype), dimension(:, :), pointer :: rlv2, dd2wall
@@ -661,37 +668,43 @@ subroutine bcturbwall(nn)
       case (imin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti1(i, j, itu1, itu1) = one
+            call saroughfact(2, i, j, fact)
+            bmti1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (imax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmti2(i, j, itu1, itu1) = one
+            call saroughfact(il, i, j, fact)
+            bmti2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj1(i, j, itu1, itu1) = one
+            call saroughfact(i, 2, j, fact)
+            bmtj1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (jmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtj2(i, j, itu1, itu1) = one
+            call saroughfact(i, jl, j, fact)
+            bmtj2(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmin) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk1(i, j, itu1, itu1) = one
+            call saroughfact(i, j, 2, fact)
+            bmtk1(i, j, itu1, itu1) = -fact
           end do
         end do
       case (kmax) 
         do j=bcdata(nn)%jcbeg,bcdata(nn)%jcend
           do i=bcdata(nn)%icbeg,bcdata(nn)%icend
-            bmtk2(i, j, itu1, itu1) = one
+            call saroughfact(i, j, kl, fact)
+            bmtk2(i, j, itu1, itu1) = -fact
           end do
         end do
       end select
@@ -1170,5 +1183,31 @@ subroutine turbbcnswall(secondhalo)
     end do bocos
   end subroutine turbbcnswall
 
+  subroutine saroughfact(i, j, k, fact)
+! returns either the regular sa-boundary condition
+! or the modified roughness-boundary condition
+    use constants
+    use inputphysics, only : useroughsa
+    use blockpointers, only : ks, d2wall, il, jl, kl
+    implicit none
+! local variablse
+    integer(kind=inttype), intent(in) :: i, j, k
+    real(kind=realtype), intent(out) :: fact
+    if (.not.useroughsa) then
+      fact = -one
+      return
+    else
+      fact = (ks(i, j, k)-d2wall(i, j, k)/0.03_realtype)/(ks(i, j, k)+&
+&       d2wall(i, j, k)/0.03_realtype)
+      if (ks(i, j, k) .eq. 0.01 .or. d2wall(i, j, k) .eq. 0.01) print*, &
+&                                                               i, j, k&
+&                                                               , fact, &
+&                                                               d2wall(i&
+&                                                               , j, k)&
+&                                                               , ks(i, &
+&                                                               j, k)
+    end if
+  end subroutine saroughfact
+
 end module turbbcroutines_fast_b
 
diff --git a/src/adjoint/outputReverseFast/utils_fast_b.f90 b/src/adjoint/outputReverseFast/utils_fast_b.f90
index 09ee3bfa9..744c9bb53 100644
--- a/src/adjoint/outputReverseFast/utils_fast_b.f90
+++ b/src/adjoint/outputReverseFast/utils_fast_b.f90
@@ -671,12 +671,13 @@ subroutine setbcpointers(nn, spatialpointers)
     use blockpointers, only : w, p, rlv, rev, gamma, x, d2wall, &
 &   si, sj, sk, s, globalcell, bcdata, nx, il, ie, ib, ny, jl, je, jb, &
 &   nz, kl, ke, kb, bcfaceid, addgridvelocities, sfacei, sfacej, sfacek,&
-&   addgridvelocities
+&   addgridvelocities, d2wall, ks
     use bcpointers, only : ww0, ww1, ww2, ww3, pp0, pp1, pp2, pp3, rlv0,&
 &   rlv1, rlv2, rlv3, rev0, rev1, rev2, rev3, gamma0, gamma1, gamma2, &
 &   gamma3, gcp, xx, ss, ssi, ssj, ssk, dd2wall, sface, istart, iend, &
-&   jstart, jend, isize, jsize
-    use inputphysics, only : cpmodel, equations
+&   jstart, jend, isize, jsize, d2wall0, d2wall1, d2wall2, d2wall3, ks0,&
+&   ks1, ks2, ks3
+    use inputphysics, only : cpmodel, equations, useroughsa
     implicit none
 ! subroutine arguments.
     integer(kind=inttype), intent(in) :: nn
@@ -711,6 +712,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(2, 1:, 1:)
       rev1 => rev(1, 1:, 1:)
       rev0 => rev(0, 1:, 1:)
+      d2wall3 => d2wall(3, 1:, 1:)
+      d2wall2 => d2wall(2, 1:, 1:)
+      d2wall1 => d2wall(1, 1:, 1:)
+      d2wall0 => d2wall(0, 1:, 1:)
+      if (useroughsa) then
+        ks3 => ks(3, 1:, 1:)
+        ks2 => ks(2, 1:, 1:)
+        ks1 => ks(1, 1:, 1:)
+        ks0 => ks(0, 1:, 1:)
+      end if
       gamma3 => gamma(3, 1:, 1:)
       gamma2 => gamma(2, 1:, 1:)
       gamma1 => gamma(1, 1:, 1:)
@@ -734,6 +745,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(il, 1:, 1:)
       rev1 => rev(ie, 1:, 1:)
       rev0 => rev(ib, 1:, 1:)
+      d2wall3 => d2wall(nx, 1:, 1:)
+      d2wall2 => d2wall(il, 1:, 1:)
+      d2wall1 => d2wall(ie, 1:, 1:)
+      d2wall0 => d2wall(ib, 1:, 1:)
+      if (useroughsa) then
+        ks3 => ks(nx, 1:, 1:)
+        ks2 => ks(il, 1:, 1:)
+        ks1 => ks(ie, 1:, 1:)
+        ks0 => ks(ib, 1:, 1:)
+      end if
       gamma3 => gamma(nx, 1:, 1:)
       gamma2 => gamma(il, 1:, 1:)
       gamma1 => gamma(ie, 1:, 1:)
@@ -757,6 +778,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 2, 1:)
       rev1 => rev(1:, 1, 1:)
       rev0 => rev(1:, 0, 1:)
+      d2wall3 => d2wall(1:, 3, 1:)
+      d2wall2 => d2wall(1:, 2, 1:)
+      d2wall1 => d2wall(1:, 1, 1:)
+      d2wall0 => d2wall(1:, 0, 1:)
+      if (useroughsa) then
+        ks3 => ks(1:, 3, 1:)
+        ks2 => ks(1:, 2, 1:)
+        ks1 => ks(1:, 1, 1:)
+        ks0 => ks(1:, 0, 1:)
+      end if
       gamma3 => gamma(1:, 3, 1:)
       gamma2 => gamma(1:, 2, 1:)
       gamma1 => gamma(1:, 1, 1:)
@@ -780,6 +811,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, jl, 1:)
       rev1 => rev(1:, je, 1:)
       rev0 => rev(1:, jb, 1:)
+      d2wall3 => d2wall(1:, ny, 1:)
+      d2wall2 => d2wall(1:, jl, 1:)
+      d2wall1 => d2wall(1:, je, 1:)
+      d2wall0 => d2wall(1:, jb, 1:)
+      if (useroughsa) then
+        ks3 => ks(1:, ny, 1:)
+        ks2 => ks(1:, jl, 1:)
+        ks1 => ks(1:, je, 1:)
+        ks0 => ks(1:, jb, 1:)
+      end if
       gamma3 => gamma(1:, ny, 1:)
       gamma2 => gamma(1:, jl, 1:)
       gamma1 => gamma(1:, je, 1:)
@@ -803,6 +844,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 1:, 2)
       rev1 => rev(1:, 1:, 1)
       rev0 => rev(1:, 1:, 0)
+      d2wall3 => d2wall(1:, 1:, 3)
+      d2wall2 => d2wall(1:, 1:, 2)
+      d2wall1 => d2wall(1:, 1:, 1)
+      d2wall0 => d2wall(1:, 1:, 0)
+      if (useroughsa) then
+        ks3 => ks(1:, 1:, 3)
+        ks2 => ks(1:, 1:, 2)
+        ks1 => ks(1:, 1:, 1)
+        ks0 => ks(1:, 1:, 0)
+      end if
       gamma3 => gamma(1:, 1:, 3)
       gamma2 => gamma(1:, 1:, 2)
       gamma1 => gamma(1:, 1:, 1)
@@ -826,6 +877,16 @@ subroutine setbcpointers(nn, spatialpointers)
       rev2 => rev(1:, 1:, kl)
       rev1 => rev(1:, 1:, ke)
       rev0 => rev(1:, 1:, kb)
+      d2wall3 => d2wall(1:, 1:, nz)
+      d2wall2 => d2wall(1:, 1:, kl)
+      d2wall1 => d2wall(1:, 1:, ke)
+      d2wall0 => d2wall(1:, 1:, kb)
+      if (useroughsa) then
+        ks3 => ks(1:, 1:, nz)
+        ks2 => ks(1:, 1:, kl)
+        ks1 => ks(1:, 1:, ke)
+        ks0 => ks(1:, 1:, kb)
+      end if
       gamma3 => gamma(1:, 1:, nz)
       gamma2 => gamma(1:, 1:, kl)
       gamma1 => gamma(1:, 1:, ke)
diff --git a/src/bcdata/BCData.F90 b/src/bcdata/BCData.F90
index f8ebb836d..c57e3af9b 100644
--- a/src/bcdata/BCData.F90
+++ b/src/bcdata/BCData.F90
@@ -10,12 +10,32 @@ module BCData
     subroutine setBCVarNamesIsothermalWall
         use cgnsNames
         use constants
+        use inputPhysics, only: useRoughSA
         implicit none
         nbcVar = nbcVarIsothermalWall
         bcVarNames(1) = cgnsTemp
 
+        if (useRoughSA) then
+            nbcVar = nbcVar + 1
+            bcVarNames(2) = cgnsSandGrainRoughness
+        end if
+
     end subroutine setBCVarNamesIsothermalWall
 
+    subroutine setBCVarNamesAdiabaticWall
+        use cgnsNames
+        use constants
+        use inputPhysics, only: useRoughSA
+        implicit none
+        nbcVar = nbcVarAdiabaticWall
+
+        if (useRoughSA) then
+            nbcVar = nbcVar + 1
+            bcVarNames(1) = cgnsSandGrainRoughness
+        end if
+
+    end subroutine setBCVarNamesAdiabaticWall
+
     subroutine setBCVarNamesSubsonicInflow
         use constants
         use cgnsNames
@@ -424,6 +444,7 @@ subroutine BCDataIsothermalWall(boco, bcVarArray, iBeg, iEnd, jBeg, jEnd)
 #ifndef USE_TAPENADE
         use commonFormats, only: strings
 #endif
+        use inputPhysics, only: useRoughSA
         implicit none
         !
         !      Subroutine arguments.
@@ -467,8 +488,62 @@ subroutine BCDataIsothermalWall(boco, bcVarArray, iBeg, iEnd, jBeg, jEnd)
             end do
         end do
 
+        ! Set a value of 0 if it was not possible to determine the
+        ! sand grain roughness
+
+        if (useRoughSA) then
+            ! Set a value of 0 if it was not possible to determine the
+            ! sand grain roughness
+            if (.not. bcVarPresent(1)) then
+                bcVarArray(:, :, 2) = zero
+            end if
+
+            do j = jBeg, jEnd
+                do i = iBeg, iEnd
+                    BCData(boco)%ksNS_Wall(i, j) = bcVarArray(i, j, 2)
+                end do
+            end do
+        end if
+
     end subroutine BCDataIsothermalWall
 
+    subroutine BCDataAdiabaticWall(boco, bcVarArray, iBeg, iEnd, jBeg, jEnd)
+        !
+        !       Tries to extract the equivalent sand grain roughness. It sets
+        !       a default value of 0.0
+        !
+        use constants
+        use cgnsNames
+        use inputPhysics, only: useRoughSA
+        use blockPointers, only: BCFaceID, BCData, nBKGlobal
+        implicit none
+        !
+        !      Subroutine arguments.
+        !
+        integer(kind=intType) :: boco
+        integer(kind=intType) :: iBeg, iEnd, jBeg, jEnd
+        real(kind=realType), dimension(iBeg:iEnd, jBeg:jEnd, nbcVarMax) :: bcVarArray
+        !
+        !      Local variables.
+        !
+        integer(kind=intType) :: i, j
+
+        if (useRoughSA) then
+            ! Set a value of 0 if it was not possible to determine the
+            ! sand grain roughness
+            if (.not. bcVarPresent(1)) then
+                bcVarArray(:, :, 1) = zero
+            end if
+
+            do j = jBeg, jEnd
+                do i = iBeg, iEnd
+                    BCData(boco)%ksNS_Wall(i, j) = bcVarArray(i, j, 1)
+                end do
+            end do
+        end if
+
+    end subroutine BCDataAdiabaticWall
+
     subroutine BCDataSubsonicInflow(boco, bcVarArray, iBeg, iEnd, jBeg, jEnd, allTurbPresent)
         !
         !       BCDataSubsonicInflow tries to extract the prescribed data
@@ -1452,6 +1527,9 @@ subroutine setBCData(bcDataNamesIn, bcDataIn, famLists, sps, &
 
                         select case (BCType(j))
 
+                        case (NSWallAdiabatic)
+                            call setBCVarNamesAdiabaticWall
+                            call errorCheckbcDataNamesIn("NSWallAdiabatic", bcDataNamesIn)
                         case (NSWallIsothermal)
                             call setBCVarNamesSupersonicInflow
                             call errorCheckbcDataNamesIn("NSWallIsothermal", bcDataNamesIn)
@@ -1550,6 +1628,9 @@ subroutine setBCData_d(bcDataNamesIn, bcDataIn, bcDataInd, famLists, sps, &
 
                         select case (BCType(j))
 
+                        case (NSWallAdiabatic)
+                            call setBCVarNamesAdiabaticWall
+                            call errorCheckbcDataNamesIn("NSWallAdiabatic", bcDataNamesIn)
                         case (NSWallIsothermal)
                             call setBCVarNamesSupersonicInflow
                             call errorCheckbcDataNamesIn("NSWallIsothermal", bcDataNamesIn)
@@ -1655,6 +1736,9 @@ subroutine setBCData_b(bcDataNamesIn, bcDataIn, bcDataInd, famLists, sps, &
 
                         select case (BCType(j))
 
+                        case (NSWallAdiabatic)
+                            call setBCVarNamesAdiabaticWall
+                            call errorCheckbcDataNamesIn("NSWallAdiabatic", bcDataNamesIn)
                         case (NSWallIsothermal)
                             call setBCVarNamesSupersonicInflow
                             call errorCheckbcDataNamesIn("NSWallIsothermal", bcDataNamesIn)
@@ -2181,6 +2265,7 @@ subroutine allocMemBCData
         use blockPointers, only: BCData, flowDoms, nBocos, nDom, BCType
         use flowVarRefState, only: nt1, nt2
         use inputTimeSpectral, only: nTimeIntervalsSpectral
+        use inputPhysics, only: useRoughSA
         use iteration, only: nALESteps
         use utils, only: setPointers, terminate
         implicit none
@@ -2250,6 +2335,14 @@ subroutine allocMemBCData
                                       BCData(mm)%area(iNodeBeg + 1:iNodeEnd, jNodeBeg + 1:jNodeEnd), &
                                       BCData(mm)%CpTarget(iNodeBeg:iNodeEnd, jNodeBeg:jNodeEnd), &
                                       stat=ierr)
+                            if (useRoughSA .and. ierr == 0) then
+                                allocate (BCData(mm)%ksNS_Wall(iBeg:iEnd, jBeg:jEnd), stat=ierr)
+                                if (level > 1) then
+                                    ! The extrapolation of the BC for MG does not work
+                                    ! properly. Thus it must be initialized with zero
+                                    BCData(mm)%ksNS_Wall = zero
+                                end if
+                            end if
                             if (ierr /= 0) &
                                  call terminate("allocMemBCData", &
                                  "Memory allocation failure for &
@@ -2273,6 +2366,14 @@ subroutine allocMemBCData
                                       BCData(mm)%area(iNodeBeg + 1:iNodeEnd, jNodeBeg + 1:jNodeEnd), &
                                       BCData(mm)%CpTarget(iNodeBeg:iNodeEnd, jNodeBeg:jNodeEnd), &
                                       stat=ierr)
+                            if (useRoughSA .and. ierr == 0) then
+                                allocate (BCData(mm)%ksNS_Wall(iBeg:iEnd, jBeg:jEnd), stat=ierr)
+                                if (level > 1) then
+                                    ! The extrapolation of the BC for MG does not work
+                                    ! properly. Thus it must be initialized with zero
+                                    BCData(mm)%ksNS_Wall = zero
+                                end if
+                            end if
                             if (ierr /= 0) &
                                  call terminate("allocMemBCData", &
                                  "Memory allocation failure for &
@@ -2632,6 +2733,7 @@ subroutine initBCData
                         nullify (BCData(j)%surfIndex)
                         nullify (BCData(j)%uSlip)
                         nullify (BCData(j)%TNS_Wall)
+                        nullify (BCData(j)%ksNS_Wall)
                         nullify (BCData(j)%CpTarget)
 
                         nullify (BCData(j)%normALE)
@@ -2756,6 +2858,11 @@ subroutine setBCDataFineGrid(initializationPart)
 
                     select case (BCType(j))
 
+                    case (NSWallAdiabatic)
+                        call setBCVarNamesAdiabaticWall ! sets bcVarNames and nbcVar
+                        call extractFromDataSet(bcVarArray)
+                        call BCDataAdiabaticWall(j, bcVarArray, iBeg, iEnd, jBeg, jEnd)
+
                     case (NSWallIsothermal)
                         call setBCVarNamesIsothermalWall ! sets bcVarNames and nbcVar
                         call extractFromDataSet(bcVarArray)
@@ -3010,6 +3117,12 @@ subroutine setBCDataFineGrid_d(initializationPart)
                     ! call the appropriate routine.
                     select case (BCType(j))
 
+                        ! This would be needed if wall roughness is differentiated
+                        ! case (NSWallAdiabatic)
+                        !    call setBCVarNamesAdiabaticWall ! sets bcVarNames and nbcVar
+                        !    call extractFromDataSet_d(bcVarArray, bcVarArrayd)
+                        !    call BCDataAdiabaticWall_d(j, bcVarArray, bcVarArrayd, iBeg, iEnd, jBeg, jEnd)
+
                     case (NSWallIsothermal)
                         call setBCVarNamesIsothermalWall ! sets bcVarNames and nbcVar
                         call extractFromDataSet_d(bcVarArray, bcVarArrayd)
@@ -3106,6 +3219,13 @@ subroutine setBCDataFineGrid_b(initializationPart)
                     ! call the appropriate routine.
                     select case (BCType(j))
 
+                        ! This would be needed if wall roughness is differentiated
+                        ! case (NSWallAdiabatic)
+                        !    call setBCVarNamesAdiabaticWall ! sets bcVarNames and nbcVar
+                        !    call extractFromDataSet(bcVarArray)
+                        !    call BCDataAdiabaticWall_b(j, bcVarArray, bcVarArrayd, iBeg, iEnd, jBeg, jEnd)
+                        !    call extractFromDataSet_b(bcVarArray, bcVarArrayd)
+
                     case (NSWallIsothermal)
                         call setBCVarNamesIsothermalWall ! sets bcVarNames and nbcVar
                         call extractFromDataSet(bcVarArray)
diff --git a/src/f2py/adflow.pyf b/src/f2py/adflow.pyf
index a5a0cc223..e41de1abd 100644
--- a/src/f2py/adflow.pyf
+++ b/src/f2py/adflow.pyf
@@ -260,6 +260,9 @@ python module libadflow
          subroutine updatewalldistancealllevels
          end subroutine updatewalldistancealllevels
 
+         subroutine updatewallroughness
+         end subroutine updatewallroughness
+
        end module walldistance
 
        module initializeflow
@@ -756,6 +759,12 @@ python module libadflow
            integer(kind=inttype), optional,intent(in),check(len(blocks)>=nblocks),depend(blocks) :: nblocks=len(blocks)
          end subroutine updaterotationrate
 
+         subroutine updatesurfaceroughness(ks_in, famlist, nfamlist) ! in :test:updatesurfaceroughness.f90
+           real(kind=realtype) ,intent(in) :: ks_in
+           integer(kind=inttype) dimension(nfamlist),intent(in) :: famlist
+           integer(kind=inttype), optional,intent(in),check(len(famlist)>=nfamlist),depend(famlist) :: nfamlist=len(famlist)
+         end subroutine updatesurfaceroughness
+
          subroutine preprocessingoverset(flag,n, closedfamlist, nfam) ! in :test:preprocessingAPI.F90:preprocessingapi
            integer(kind=inttype) dimension(n) :: flag
            integer(kind=inttype), optional,intent(in),check(len(flag)>=n),depend(flag) :: n=len(flag)
@@ -1202,6 +1211,7 @@ python module libadflow
          real(kind=realtype) :: beta
          integer(kind=inttype) :: liftindex
          real(kind=realtype) :: cavitationnumber
+         logical :: useroughsa
          real(kind=realtype) :: cpmin_rho
        end module inputphysics
 
diff --git a/src/inputParam/inputParamRoutines.F90 b/src/inputParam/inputParamRoutines.F90
index 76dc34e48..a86c877dd 100644
--- a/src/inputParam/inputParamRoutines.F90
+++ b/src/inputParam/inputParamRoutines.F90
@@ -2508,6 +2508,7 @@ subroutine volumeVariables(variables)
         use constants
         use extraOutput
         use utils, only: convertToLowerCase, terminate
+        use inputPhysics, only: useRoughSA
         implicit none
         !
         !      Subroutine arguments.
@@ -2559,6 +2560,8 @@ subroutine volumeVariables(variables)
         volWriteStatus = .false.
         volWriteIntermittency = .false.
 
+        volWriteKs = .false.
+
         ! Initialize nVarSpecified to 0. This serves as a test
         ! later on.
 
@@ -2709,6 +2712,15 @@ subroutine volumeVariables(variables)
                 volWriteIntermittency = .true.
                 nVarSpecified = nVarSpecified + 1
 
+            case ("ks")
+                if (.not. useRoughSA) then
+                    call terminate("volumeVariables", "Can not export surface roughness &
+                    &values ('volumeVariables': ['ks']) when the rough SA variant is not &
+                    &used (useRoughSA = False)")
+                end if
+                volWriteKs = .true.
+                nVarSpecified = nVarSpecified + 1
+
             case default
                 pos = len_trim(keyword)
                 write (errorMessage, "(3a)") "Unknown extra volume output &
diff --git a/src/modules/BCPointers.F90 b/src/modules/BCPointers.F90
index 39b8c6245..f52eda2e8 100644
--- a/src/modules/BCPointers.F90
+++ b/src/modules/BCPointers.F90
@@ -10,6 +10,8 @@ module BCPointers
     real(kind=realType), dimension(:, :, :), pointer :: ww0, ww1, ww2, ww3
     real(kind=realType), dimension(:, :), pointer :: pp0, pp1, pp2, pp3
     real(kind=realType), dimension(:, :), pointer :: rlv0, rlv1, rlv2, rlv3
+    real(kind=realType), dimension(:, :), pointer :: d2wall0, d2wall1, d2wall2, d2wall3
+    real(kind=realType), dimension(:, :), pointer :: ks0, ks1, ks2, ks3
     real(kind=realType), dimension(:, :), pointer :: rev0, rev1, rev2, rev3
     real(kind=realType), dimension(:, :), pointer :: gamma0, gamma1, gamma2, gamma3
     real(kind=realType), dimension(:, :, :), pointer :: ssi, ssj, ssk
@@ -25,6 +27,7 @@ module BCPointers
     real(kind=realType), dimension(:, :), pointer :: pp0d, pp1d, pp2d, pp3d
     real(kind=realType), dimension(:, :), pointer :: rlv0d, rlv1d, rlv2d, rlv3d
     real(kind=realType), dimension(:, :), pointer :: rev0d, rev1d, rev2d, rev3d
+    real(kind=realType), dimension(:, :), pointer :: d2wall0d, d2wall1d, d2wall2d, d2wall3d
     real(kind=realType), dimension(:, :), pointer :: gamma0d, gamma1d, gamma2d, gamma3d
     real(kind=realType), dimension(:, :, :), pointer :: ssid, ssjd, sskd, xxd
     real(kind=realType), dimension(:, :, :), pointer :: ssd
diff --git a/src/modules/block.F90 b/src/modules/block.F90
index 1fdb9b3fe..ab3b66078 100644
--- a/src/modules/block.F90
+++ b/src/modules/block.F90
@@ -96,9 +96,11 @@ module block
         ! uSlip(:,:,3):  the 3 components of the velocity vector on
         !                a viscous wall.
         ! TNS_Wall(:,:): Wall temperature for isothermal walls.
+        ! ksNS_Wall(:,:): Equivalent Sand Grain Roughness on viscous walls.
 
         real(kind=realType), dimension(:, :, :), pointer :: uSlip
         real(kind=realType), dimension(:, :), pointer :: TNS_Wall
+        real(kind=realType), dimension(:, :), pointer :: ksNS_Wall
 
         ! The name of this boundary condition and it's index
         character(maxCGNSNameLen) :: family
@@ -656,12 +658,19 @@ module block
         !
         !         Turbulence model variables.
         !
-        ! d2Wall(2:il,2:jl,2:kl) - Distance from the center of the cell
-        !                          to the nearest viscous wall.
+        ! d2Wall(0:ib, 0:jb, 0:kb) - Distance from the center of the cell
+        !                          to the nearest viscous wall. It is exchanged for non-BC halo cells and extrapolated 
+        !                          for symmetry type BCs. For other BCs, this value is initialize to a large number
         ! intermittency( )       - Function defining the transition location
+        !
+        ! The next two variables are only initialized if roughness is requested (useRoughSA = True)
+        ! nearestWallCellInd(2:il,2:jl,2:kl)   - global cell ID for the nearest wall cell; is needed for rougness
+        ! ks(0:ib, 0:jb, 0:kb)     - Roughness value of the nearest wall - see d2wall description for the values set
 
         real(kind=realType), dimension(:, :, :), pointer :: d2Wall, filterDES
         real(kind=realType), dimension(:, :, :), pointer :: intermittency
+        integer(kind=intType), dimension(:, :, :), pointer :: nearestWallCellInd
+        real(kind=realType), dimension(:, :, :), pointer :: ks
 
         ! bmti1(je,ke,nt1:nt2,nt1:nt2): Matrix used for the implicit
         !                               boundary condition treatment of
diff --git a/src/modules/blockPointers.F90 b/src/modules/blockPointers.F90
index 4705f9fd7..16f2b3c6a 100644
--- a/src/modules/blockPointers.F90
+++ b/src/modules/blockPointers.F90
@@ -152,6 +152,7 @@ module blockPointers
     real(kind=realType), dimension(:, :, :), pointer :: radI, radJ, radK
 
     real(kind=realType), dimension(:, :, :), pointer :: d2Wall
+    real(kind=realType), dimension(:, :, :), pointer :: ks
     real(kind=realType), dimension(:, :, :), pointer :: intermittency
     real(kind=realType), dimension(:, :, :), pointer :: filterDES  ! eran-des
     real(kind=realType), dimension(:, :, :, :), pointer :: bmti1
diff --git a/src/modules/cgnsNames.f90 b/src/modules/cgnsNames.f90
index c3c67549e..02d8c5164 100644
--- a/src/modules/cgnsNames.f90
+++ b/src/modules/cgnsNames.f90
@@ -171,6 +171,9 @@ module cgnsNames
     character(len=maxCGNSNameLen), parameter :: &
         cgnsIntermittency = "Intermittency"
 
+    character(len=maxCGNSNameLen), parameter :: &
+        cgnsSandGrainRoughness = "SandGrainRoughness"
+
 !
 !       Residual names.
 !
diff --git a/src/modules/constants.F90 b/src/modules/constants.F90
index 086240822..730bccf18 100644
--- a/src/modules/constants.F90
+++ b/src/modules/constants.F90
@@ -300,6 +300,7 @@ module constants
     integer(kind=intType), parameter :: nbcVarSubsonicInflow = 17
     integer(kind=intType), parameter :: nbcVarSubsonicOutflow = 1
     integer(kind=intType), parameter :: nbcVarSupersonicInflow = 7
+    integer(kind=intType), parameter :: nbcVarAdiabaticWall = 0
     integer(kind=intType), parameter :: nbcVarIsothermalWall = 1
 
     ! Indices of specific familyExcahnge groups based on BC
diff --git a/src/modules/extraOutput.f90 b/src/modules/extraOutput.f90
index 8a6e33e1e..78fdc9d03 100644
--- a/src/modules/extraOutput.f90
+++ b/src/modules/extraOutput.f90
@@ -36,6 +36,7 @@ module extraOutput
     logical :: volWriteResRhoE, volWriteResTurb, volWriteBlank
     logical :: volWriteShock, volWriteFilteredShock, volWriteGC, volWriteStatus
     logical :: volWriteIntermittency
+    logical :: volWriteKs
 !
 !       The logical variables, which define the isosurface variables
 !       to be written.
diff --git a/src/modules/inputParam.F90 b/src/modules/inputParam.F90
index 9afa9ccaf..0d6a6a7ec 100644
--- a/src/modules/inputParam.F90
+++ b/src/modules/inputParam.F90
@@ -529,6 +529,7 @@ module inputPhysics
     !                      when considering turbulence model effects
     ! useRotationSA:       Determines if we will use rotation correction (SA model only)
     ! useft2SA:            Determines if we will use the ft2 term (SA model only)
+    ! useRoughSA:          Whether or not to use rough version of SA (BC values are set via CGNS-Grid)
     ! wallFunctions:       Whether or not to use wall functions.
     ! wallDistanceNeeded:  Whether or not the wall distance is needed
     !                      for the turbulence model in a RANS problem.
@@ -596,6 +597,7 @@ module inputPhysics
     real(kind=realType), dimension(3, 2) :: momentAxis
     real(kind=realType) :: SSuthDim, muSuthDim, TSuthDim
     real(kind=realType) :: cavitationnumber
+    logical :: useRoughSA
     real(kind=realType) :: cpmin_rho
     real(kind=realType), dimension(:), allocatable :: cpmin_family
 
diff --git a/src/modules/paramTurb.F90 b/src/modules/paramTurb.F90
index db7a162a6..9adf41ae9 100644
--- a/src/modules/paramTurb.F90
+++ b/src/modules/paramTurb.F90
@@ -24,6 +24,7 @@ module paramTurb
     real(kind=realType), parameter :: rsaCt3 = 1.2_realType
     real(kind=realType), parameter :: rsaCt4 = 0.5_realType
     real(kind=realType), parameter :: rsaCrot = 2.0_realType
+    real(kind=realType), parameter :: rsaCr1 = 0.5_realType
 
 !
 !       K-omega constants.
diff --git a/src/modules/wallDistanceData.F90 b/src/modules/wallDistanceData.F90
index cb388b5a5..27b809c8f 100644
--- a/src/modules/wallDistanceData.F90
+++ b/src/modules/wallDistanceData.F90
@@ -43,4 +43,6 @@ module wallDistanceData
     real(kind=realType), dimension(:), pointer :: xVolume
 #endif
 
+    integer(kind=intType), dimension(:, :), allocatable :: nCellBlockOffset
+
 end module wallDistanceData
diff --git a/src/output/outputMod.F90 b/src/output/outputMod.F90
index 1a954888b..016695f48 100644
--- a/src/output/outputMod.F90
+++ b/src/output/outputMod.F90
@@ -232,6 +232,8 @@ subroutine numberOfVolSolVariables(nVolSolvar, nVolDiscrVar)
 
         if (volWriteBlank) nVolDiscrVar = nVolDiscrVar + 1
 
+        if (volwriteKs) nVolDiscrVar = nVolDiscrVar + 1
+
     end subroutine numberOfVolSolVariables
 
     subroutine copyDataBufSinglePrecision(val, buffer, &
@@ -567,6 +569,11 @@ subroutine volSolNames(solNames)
             solNames(nn) = cgnsIntermittency
         end if
 
+        if (volWriteKs) then
+            nn = nn + 1
+            solNames(nn) = cgnsSandGrainRoughness
+        end if
+
     end subroutine volSolNames
 
     subroutine surfSolNames(solNames)
@@ -1346,6 +1353,18 @@ subroutine storeSolInBuffer(buffer, copyInBuffer, solName, &
                 end do
             end do
 
+        case (cgnsSandGrainRoughness)
+            ! It is only possible to write this when it was allocated in the first place
+            ! (useRoughSA = True) but this has been check in 'inputParamRoutines'
+            ! allready
+            do k = kBeg, kEnd
+                do j = jBeg, jEnd
+                    do i = iBeg, iEnd
+                        wIO(i, j, k, 1) = real(ks(i, j, k), realType)
+                    end do
+                end do
+            end do
+
         case default
             call terminate("storeSolInBuffer", &
                            "This should not happen")
diff --git a/src/preprocessing/preprocessingAPI.F90 b/src/preprocessing/preprocessingAPI.F90
index 6516357ec..c6f4098a6 100644
--- a/src/preprocessing/preprocessingAPI.F90
+++ b/src/preprocessing/preprocessingAPI.F90
@@ -19,6 +19,7 @@ subroutine preprocessing
                                  recvBufferSize_1to1, sendBufferSize_1to1, sendBufferSIzeOver, &
                                  recvBufferSizeOver, commPatternOverset, internalOverset, sendBuffer, &
                                  recvBuffer, sendBufferSize, recvBufferSize
+        use haloExchange, only: exchanged2Wall
         use inputPhysics
         use inputTimeSpectral
         use section
@@ -31,6 +32,7 @@ subroutine preprocessing
         use coarseUtils, only: createCoarseBlocks
         use pointMatchedCommPattern, only: determineCommPattern
         use oversetAPI, only: oversetComm, determineClusters, determineViscousDirs
+        use wallDistanceData, only: nCellBlockOffset
         implicit none
         !
         !      Local variables.
@@ -208,6 +210,12 @@ subroutine preprocessing
         ! See the corresponding subroutine header, although the
         ! names are pretty self-explaining
 
+        ! Allocate Block-offset. SA rough might need it. It is filled in
+        ! 'setGlobalCellsAndNodes'. This function is called per level and thus it
+        ! must be allocated before the call
+
+        allocate (nCellBlockOffset(nLevels, nDom))
+
         do level = 1, nLevels
             call xhalo(level)
             call allocateMetric(level)
@@ -232,6 +240,9 @@ subroutine preprocessing
         ! be done and overset connectivity computed
         do level = 1, nLevels
             call computeWallDistance(level, .True.)
+            if (wallDistanceNeeded) then
+                call exchanged2Wall(level)
+            end if
         end do
         call preprocessingADjoint
 
@@ -1918,6 +1929,7 @@ subroutine setGlobalCellsAndNodes(level)
         use inputTimeSpectral
         use utils, only: setPointers, terminate
         use haloExchange, only: whalo1to1intgeneric
+        use wallDistanceData, only: nCellBlockOffset
         implicit none
 
         ! Input variables
@@ -1928,7 +1940,7 @@ subroutine setGlobalCellsAndNodes(level)
         integer(kind=intType) :: ierr, istart
         logical :: commPressure, commLamVis, commEddyVis, commGamma
         integer(kind=intType), dimension(nProc) :: nNodes, nCells, nCellOffset, nNodeOffset
-        integer(kind=intType), dimension(nDom) :: nCellBLockOffset, nNodeBLockOffset
+        integer(kind=intType), dimension(nDom) :: nNodeBLockOffset
         integer(kind=intType) :: npts, nCell, nNode
         integer(kind=intType), dimension(:), allocatable :: nNodesProc, cumNodesProc
         integer(kind=intTYpe), dimension(:), allocatable :: nCellsProc, cumCellsProc
@@ -1995,11 +2007,11 @@ subroutine setGlobalCellsAndNodes(level)
                          adflow_integer, 0, ADflow_comm_world, ierr)
 
         ! Determine the global cell number offset for each local block.
-        nCellBlockOffset(1) = nCellOffsetLocal(level)
+        nCellBlockOffset(level, 1) = nCellOffsetLocal(level)
         do nn = 2, nDom
             call setPointers(nn - 1, level, 1)
-            nCellBlockOffset(nn) = nCellBlockOffset(nn - 1) &
-                                   + nx * ny * nz
+            nCellBlockOffset(level, nn) = nCellBlockOffset(level, nn - 1) &
+                                          + nx * ny * nz
         end do
 
         ! Repeat for nodes.
@@ -2026,7 +2038,7 @@ subroutine setGlobalCellsAndNodes(level)
                             ! instances of a give block adjacent to each other in
                             ! the matrix
                             globalCell(i, j, k) = &
-                                nCellBLockOffset(nn) * nTimeIntervalsSpectral + nx * ny * nz * (sps - 1) + &
+                                nCellBLockOffset(level, nn) * nTimeIntervalsSpectral + nx * ny * nz * (sps - 1) + &
                                 (i - 2) + (j - 2) * nx + (k - 2) * nx * ny
                         end do
                     end do
@@ -4356,4 +4368,41 @@ subroutine updateRotationRate(rotCenter, rotRate, blocks, nblocks)
 
     end subroutine updateRotationRate
 
+    subroutine updateSurfaceRoughness(ks_in, famList, nFamList)
+
+        use constants
+        use blockPointers
+        use inputTimeSpectral, only: nTimeIntervalsSpectral
+        use utils, only: setPointers
+        use sorting, only: famInList
+        implicit none
+
+        real(kind=realType), intent(in) :: ks_in
+        integer(kind=intType), intent(in) :: nFamList, famList(nFamList)
+
+        integer(kind=intType) :: nLevels, level, sps, nn, mm
+
+        nLevels = ubound(flowDoms, 2)
+
+        do level = 1, nLevels
+            do sps = 1, nTimeIntervalsSpectral
+                do nn = 1, nDom
+                    call setPointers(nn, level, sps)
+
+                    ! Loop over the number of boundary subfaces of this block.
+                    do mm = 1, nBocos
+
+                        if (.not. famInList(BCData(mm)%famID, famList)) then
+                            cycle
+                        end if
+
+                        BCData(mm)%ksNS_Wall = ks_in
+
+                    end do
+                end do
+            end do
+        end do
+
+    end subroutine updateSurfaceRoughness
+
 end module preprocessingAPI
diff --git a/src/solver/BCRoutines.F90 b/src/solver/BCRoutines.F90
index 9ce054802..4c48da790 100644
--- a/src/solver/BCRoutines.F90
+++ b/src/solver/BCRoutines.F90
@@ -234,8 +234,9 @@ subroutine bcSymm1stHalo(nn)
         use constants
         use blockPointers, only: BCdata
         use flowVarRefState, only: viscous, eddyModel
+        use inputPhysics, only: useRoughSA
         use BCPointers, only: gamma1, gamma2, ww1, ww2, pp1, pp2, rlv1, rlv2, &
-                              iStart, jStart, iSize, jSize, rev1, rev2
+                              iStart, jStart, iSize, jSize, rev1, rev2, d2wall1, d2wall2, ks1, ks2
         implicit none
 
         ! Subroutine arguments.
@@ -274,6 +275,8 @@ subroutine bcSymm1stHalo(nn)
 
             gamma1(i, j) = gamma2(i, j)
             pp1(i, j) = pp2(i, j)
+            d2wall1(i, j) = d2wall2(i, j)
+            if (useRoughSA) ks1(i, j) = ks2(i, j)
             if (viscous) rlv1(i, j) = rlv2(i, j)
             if (eddyModel) rev1(i, j) = rev2(i, j)
         end do
@@ -287,7 +290,8 @@ subroutine bcSymm2ndHalo(nn)
         use constants
         use blockPointers, only: BCdata
         use flowVarRefState, only: viscous, eddyModel
-        use BCPointers, only: gamma0, gamma3, ww0, ww3, pp0, pp3, rlv0, rlv3, &
+        use inputPhysics, only: useRoughSA
+        use BCPointers, only: gamma0, gamma3, ww0, ww3, pp0, pp3, rlv0, rlv3, d2wall0, d2wall3, ks0, ks3, &
                               rev0, rev3, iStart, jStart, iSize, jSize
         implicit none
 
@@ -323,6 +327,8 @@ subroutine bcSymm2ndHalo(nn)
 
             gamma0(i, j) = gamma3(i, j)
             pp0(i, j) = pp3(i, j)
+            d2wall0(i, j) = d2wall3(i, j)
+            if (useRoughSA) ks0(i, j) = ks3(i, j)
             if (viscous) rlv0(i, j) = rlv3(i, j)
             if (eddyModel) rev0(i, j) = rev3(i, j)
         end do
@@ -339,8 +345,10 @@ subroutine bcSymmPolar1stHalo(nn)
         ! centerline. This routine does just the 1st level halo.
 
         use constants
+        use inputPhysics, only: useRoughSA
         use BCPointers, only: ww1, ww2, pp1, pp2, rlv1, rlv2, rev1, rev2, &
-                              xx, iStart, jStart, iSize, jSize
+                              xx, iStart, jStart, iSize, jSize, &
+                              d2wall1, d2wall2, ks1, ks2
         use flowVarRefState, only: viscous, eddyModel
         implicit none
 
@@ -403,6 +411,8 @@ subroutine bcSymmPolar1stHalo(nn)
             ! eddy viscosity in the halo.
 
             pp1(i, j) = pp2(i, j)
+            d2wall1(i, j) = d2wall2(i, j)
+            if (useRoughSA) ks1(i, j) = ks2(i, j)
             if (viscous) rlv1(i, j) = rlv2(i, j)
             if (eddyModel) rev1(i, j) = rev2(i, j)
         end do
@@ -418,7 +428,8 @@ subroutine bcSymmPolar2ndHalo(nn)
         ! centerline. This routine does just the 2nd level halo.
 
         use constants
-        use BCPointers, only: ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, &
+        use inputPhysics, only: useRoughSA
+        use BCPointers, only: ww0, ww3, pp0, pp3, rlv0, rlv3, rev0, rev3, d2wall0, d2wall3, ks0, ks3, &
                               xx, iStart, jStart, iSize, jSize
         use flowVarRefState, only: viscous, eddyModel
         implicit none
@@ -480,6 +491,8 @@ subroutine bcSymmPolar2ndHalo(nn)
             ! eddy viscosity in the halo.
 
             pp0(i, j) = pp3(i, j)
+            d2wall0(i, j) = d2wall3(i, j)
+            if (useRoughSA) ks0(i, j) = ks3(i, j)
             if (viscous) rlv0(i, j) = rlv3(i, j)
             if (eddyModel) rev0(i, j) = rev3(i, j)
         end do
diff --git a/src/turbulence/sa.F90 b/src/turbulence/sa.F90
index 7459158d4..0651b182b 100644
--- a/src/turbulence/sa.F90
+++ b/src/turbulence/sa.F90
@@ -106,6 +106,7 @@ subroutine saSource
 
         ! Local variables.
         integer(kind=intType) :: i, j, k, nn, ii
+        real(kind=realType) :: distRough
         real(kind=realType) :: fv1, fv2, ft2
         real(kind=realType) :: ss, sst, nu, dist2Inv, chi, chi2, chi3
         real(kind=realType) :: rr, gg, gg6, termFw, fwSa, term1, term2
@@ -243,12 +244,25 @@ subroutine saSource
                         ! the production term near a viscous wall.
 
                         nu = rlv(i, j, k) / w(i, j, k, irho)
-                        dist2Inv = one / (d2Wall(i, j, k)**2)
                         chi = w(i, j, k, itu1) / nu
+
+                        if (.not. useRoughSA) then
+                            dist2Inv = one / (d2Wall(i, j, k)**2)
+                        else
+                            distRough = d2Wall(i, j, k) + 0.03_realType * ks(i, j, k)
+                            dist2Inv = one / (distRough**2)
+                            chi = chi + rsaCr1 * ks(i, j, k) / distRough
+                        end if
+
                         chi2 = chi * chi
                         chi3 = chi * chi2
                         fv1 = chi3 / (chi3 + cv13)
-                        fv2 = one - chi / (one + chi * fv1)
+
+                        if (.not. useRoughSA) then
+                            fv2 = one - chi / (one + chi * fv1)
+                        else
+                            fv2 = one - w(i, j, k, itu1) / (nu + w(i, j, k, itu1) * fv1)
+                        end if
 
                         ! The function ft2, which is designed to keep a laminar
                         ! solution laminar. When running in fully turbulent mode
@@ -307,7 +321,13 @@ subroutine saSource
                         ! treatment.
 
                         dfv1 = three * chi2 * cv13 / ((chi3 + cv13)**2)
-                        dfv2 = (chi2 * dfv1 - one) / (nu * ((one + chi * fv1)**2))
+
+                        if (.not. useRoughSA) then
+                            dfv2 = (chi2 * dfv1 - one) / (nu * ((one + chi * fv1)**2))
+                        else
+                            dfv2 = (w(i, j, k, itu1) * dfv1 - nu) / (nu + w(i, j, k, itu1) * fv1)**2
+                        end if
+
                         dft2 = -two * rsaCt4 * chi * ft2 / nu
 
                         drr = (one - rr * (fv2 + w(i, j, k, itu1) * dfv2)) &
diff --git a/src/turbulence/turbBCRoutines.F90 b/src/turbulence/turbBCRoutines.F90
index 24c7cadc9..4824f4134 100644
--- a/src/turbulence/turbBCRoutines.F90
+++ b/src/turbulence/turbBCRoutines.F90
@@ -320,6 +320,7 @@ subroutine bcEddyWall(nn)
         !      Local variables.
         !
         integer(kind=intType) :: i, j
+        real(kind=realType) :: fact
 
         ! Determine the face id on which the subface is located and
         ! loop over the faces of the subface and set the eddy viscosity
@@ -329,42 +330,48 @@ subroutine bcEddyWall(nn)
         case (iMin)
             do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                 do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                    rev(1, i, j) = -rev(2, i, j)
+                    call saRoughFact(2, i, j, fact)
+                    rev(1, i, j) = fact * rev(2, i, j)
                 end do
             end do
 
         case (iMax)
             do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                 do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                    rev(ie, i, j) = -rev(il, i, j)
+                    call saRoughFact(il, i, j, fact)
+                    rev(ie, i, j) = fact * rev(il, i, j)
                 end do
             end do
 
         case (jMin)
             do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                 do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                    rev(i, 1, j) = -rev(i, 2, j)
+                    call saRoughFact(i, 2, j, fact)
+                    rev(i, 1, j) = fact * rev(i, 2, j)
                 end do
             end do
 
         case (jMax)
             do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                 do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                    rev(i, je, j) = -rev(i, jl, j)
+                    call saRoughFact(i, jl, j, fact)
+                    rev(i, je, j) = fact * rev(i, jl, j)
                 end do
             end do
 
         case (kMin)
             do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                 do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                    rev(i, j, 1) = -rev(i, j, 2)
+                    call saRoughFact(i, j, 2, fact)
+                    rev(i, j, 1) = fact * rev(i, j, 2)
                 end do
             end do
 
         case (kMax)
             do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                 do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                    rev(i, j, ke) = -rev(i, j, kl)
+                    call saRoughFact(i, j, kl, fact)
+                    rev(i, j, ke) = fact * rev(i, j, kl)
                 end do
             end do
         end select
@@ -819,7 +826,7 @@ subroutine bcTurbWall(nn)
         !
         integer(kind=intType) :: i, j, ii, jj, iiMax, jjMax
 
-        real(kind=realType) :: tmpd, tmpe, tmpf, nu
+        real(kind=realType) :: tmpd, tmpe, tmpf, nu, fact
 
         real(kind=realType), dimension(:, :, :, :), pointer :: bmt
         real(kind=realType), dimension(:, :, :), pointer :: bvt, ww2
@@ -839,39 +846,45 @@ subroutine bcTurbWall(nn)
             case (iMin)
                 do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                     do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                        bmti1(i, j, itu1, itu1) = one
+                        call saRoughFact(2, i, j, fact)
+                        bmti1(i, j, itu1, itu1) = -fact
                     end do
                 end do
             case (iMax)
                 do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                     do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                        bmti2(i, j, itu1, itu1) = one
+                        call saRoughFact(il, i, j, fact)
+                        bmti2(i, j, itu1, itu1) = -fact
                     end do
                 end do
             case (jMin)
                 do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                     do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                        bmtj1(i, j, itu1, itu1) = one
+                        call saRoughFact(i, 2, j, fact)
+                        bmtj1(i, j, itu1, itu1) = -fact
                     end do
                 end do
             case (jMax)
                 do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                     do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                        bmtj2(i, j, itu1, itu1) = one
+                        call saRoughFact(i, jl, j, fact)
+                        bmtj2(i, j, itu1, itu1) = -fact
                     end do
                 end do
 
             case (kMin)
                 do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                     do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                        bmtk1(i, j, itu1, itu1) = one
+                        call saRoughFact(i, j, 2, fact)
+                        bmtk1(i, j, itu1, itu1) = -fact
                     end do
                 end do
 
             case (kMax)
                 do j = BCData(nn)%jcBeg, BCData(nn)%jcEnd
                     do i = BCData(nn)%icBeg, BCData(nn)%icEnd
-                        bmtk2(i, j, itu1, itu1) = one
+                        call saRoughFact(i, j, kl, fact)
+                        bmtk2(i, j, itu1, itu1) = -fact
                     end do
                 end do
             end select
@@ -1385,4 +1398,28 @@ subroutine turbBCNSWall(secondHalo)
         end do bocos
     end subroutine turbBCNSWall
 
+    subroutine saRoughFact(i, j, k, fact)
+
+        ! returns either the regular SA-boundary condition
+        ! or the modified Roughness-boundary condition
+
+        use constants
+        use inputPhysics, only: useRoughSA
+        use BlockPointers, only: ks, d2wall, il, jl, kl
+        implicit none
+
+        ! local variablse
+        integer(kind=intType), intent(in) :: i, j, k
+        real(kind=realType), intent(out) :: fact
+
+        if (.not. useRoughSA) then
+            fact = -one
+            return
+        end if
+
+        fact = (ks(i, j, k) - d2wall(i, j, k) / 0.03_realType) / &
+               (ks(i, j, k) + d2wall(i, j, k) / 0.03_realType)
+
+    end subroutine saRoughFact
+
 end module turbBCRoutines
diff --git a/src/utils/haloExchange.F90 b/src/utils/haloExchange.F90
index 11d9c2665..61a5890ec 100644
--- a/src/utils/haloExchange.F90
+++ b/src/utils/haloExchange.F90
@@ -2639,6 +2639,367 @@ subroutine exchangeCoor(level)
 
     end subroutine exchangeCoor
 
+    subroutine exchanged2Wall(level)
+        !
+        !       ExchangeCoor exchanges the d2wall of the given grid
+        !       level.
+        !
+        use block
+        use communication
+        use inputTimeSpectral
+        use utils, only: EChk
+        implicit none
+        !
+        !      Subroutine arguments.
+        !
+        integer(kind=intType), intent(in) :: level
+        !
+        !      Local variables.
+        !
+        integer :: size, procID, ierr, index
+        integer, dimension(mpi_status_size) :: mpiStatus
+
+        integer(kind=intType) :: i, j, ii, jj, mm
+        integer(kind=intType) :: d1, i1, j1, k1, d2, i2, j2, k2
+
+        ! Loop over the number of spectral solutions.
+
+        spectralLoop: do mm = 1, nTimeIntervalsSpectral
+
+            ! Send the coordinates i have to send. The data is first copied
+            ! into the send buffer and this buffer is sent.
+
+            ii = 1
+            sends: do i = 1, commPatternCell_2nd(level)%nProcSend
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternCell_2nd(level)%sendProc(i)
+                size = commPatternCell_2nd(level)%nSend(i)
+
+                ! Copy the data in the correct part of the send buffer.
+
+                jj = ii
+                !DIR$ NOVECTOR
+                do j = 1, commPatternCell_2nd(level)%nSend(i)
+
+                    ! Store the block id and the indices of the donor
+                    ! a bit easier.
+
+                    d1 = commPatternCell_2nd(level)%sendList(i)%block(j)
+                    i1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 1)
+                    j1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 2)
+                    k1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 3)
+
+                    ! Copy the coordinates of this point in the buffer.
+                    ! Update the counter jj accordingly.
+
+                    sendBuffer(jj) = flowDoms(d1, level, mm)%d2Wall(i1, j1, k1)
+                    jj = jj + 1
+
+                end do
+
+                ! Send the data.
+
+                call mpi_isend(sendBuffer(ii), size, adflow_real, procID, &
+                               procID, ADflow_comm_world, sendRequests(i), &
+                               ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Set ii to jj for the next processor.
+
+                ii = jj
+
+            end do sends
+
+            ! Post the nonblocking receives.
+
+            ii = 1
+            receives: do i = 1, commPatternCell_2nd(level)%nProcRecv
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternCell_2nd(level)%recvProc(i)
+                size = commPatternCell_2nd(level)%nRecv(i)
+
+                ! Post the receive.
+
+                call mpi_irecv(recvBuffer(ii), size, adflow_real, procID, &
+                               myID, ADflow_comm_world, recvRequests(i), ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! And update ii.
+
+                ii = ii + size
+
+            end do receives
+
+            ! Copy the local data.
+            !DIR$ NOVECTOR
+            localCopy: do i = 1, internalCell_2nd(level)%nCopy
+
+                ! Store the block and the indices of the donor a bit easier.
+
+                d1 = internalCell_2nd(level)%donorBlock(i)
+                i1 = internalCell_2nd(level)%donorIndices(i, 1)
+                j1 = internalCell_2nd(level)%donorIndices(i, 2)
+                k1 = internalCell_2nd(level)%donorIndices(i, 3)
+                ! Idem for the halo's.
+
+                d2 = internalCell_2nd(level)%haloBlock(i)
+                i2 = internalCell_2nd(level)%haloIndices(i, 1)
+                j2 = internalCell_2nd(level)%haloIndices(i, 2)
+                k2 = internalCell_2nd(level)%haloIndices(i, 3)
+                ! Copy the coordinates.
+                flowDoms(d2, level, mm)%d2Wall(i2, j2, k2) = &
+                    flowDoms(d1, level, mm)%d2Wall(i1, j1, k1)
+
+            end do localCopy
+
+            ! Correct the periodic halos of the internal communication
+            ! pattern
+
+            ! call correctPeriodicCoor(level, mm, &
+            !                          internalCell_2nd(level)%nPeriodic, &
+            !                          internalCell_2nd(level)%periodicData)
+
+            ! Complete the nonblocking receives in an arbitrary sequence and
+            ! copy the coordinates from the buffer into the halo's.
+
+            size = commPatternCell_2nd(level)%nProcRecv
+            completeRecvs: do i = 1, commPatternCell_2nd(level)%nProcRecv
+
+                ! Complete any of the requests.
+
+                call mpi_waitany(size, recvRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Copy the data just arrived in the halo's.
+
+                ii = index
+                jj = commPatternCell_2nd(level)%nRecvCum(ii - 1) + 1
+                !DIR$ NOVECTOR
+                do j = 1, commPatternCell_2nd(level)%nRecv(ii)
+
+                    ! Store the block and the indices of the halo a bit easier.
+
+                    d2 = commPatternCell_2nd(level)%recvList(ii)%block(j)
+                    i2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 1)
+                    j2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 2)
+                    k2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 3)
+
+                    ! Copy the data.
+
+                    flowDoms(d2, level, mm)%d2Wall(i2, j2, k2) = recvBuffer(jj)
+                    jj = jj + 1
+
+                end do
+
+            end do completeRecvs
+
+            ! Correct the periodic halos of the external communication
+            ! pattern.
+
+            ! call correctPeriodicCoor(level, mm, &
+            !                          commPatternCell_2nd(level)%nPeriodic, &
+            !                          commPatternCell_2nd(level)%periodicData)
+
+            ! Complete the nonblocking sends.
+
+            size = commPatternCell_2nd(level)%nProcSend
+            do i = 1, commPatternCell_2nd(level)%nProcSend
+                call mpi_waitany(size, sendRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+            end do
+
+        end do spectralLoop
+
+    end subroutine exchanged2Wall
+
+
+    subroutine exchangeKs(level)
+        !
+        !       ExchangeCoor exchanges the wall roughness of the nearest wall of the given grid
+        !       level.
+        !
+        use block
+        use communication
+        use inputTimeSpectral
+        use utils, only: EChk
+        implicit none
+        !
+        !      Subroutine arguments.
+        !
+        integer(kind=intType), intent(in) :: level
+        !
+        !      Local variables.
+        !
+        integer :: size, procID, ierr, index
+        integer, dimension(mpi_status_size) :: mpiStatus
+
+        integer(kind=intType) :: i, j, ii, jj, mm
+        integer(kind=intType) :: d1, i1, j1, k1, d2, i2, j2, k2
+
+        ! Loop over the number of spectral solutions.
+
+        spectralLoop: do mm = 1, nTimeIntervalsSpectral
+
+            ! Send the coordinates i have to send. The data is first copied
+            ! into the send buffer and this buffer is sent.
+
+            ii = 1
+            sends: do i = 1, commPatternCell_2nd(level)%nProcSend
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternCell_2nd(level)%sendProc(i)
+                size = commPatternCell_2nd(level)%nSend(i)
+
+                ! Copy the data in the correct part of the send buffer.
+
+                jj = ii
+                !DIR$ NOVECTOR
+                do j = 1, commPatternCell_2nd(level)%nSend(i)
+
+                    ! Store the block id and the indices of the donor
+                    ! a bit easier.
+
+                    d1 = commPatternCell_2nd(level)%sendList(i)%block(j)
+                    i1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 1)
+                    j1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 2)
+                    k1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 3)
+
+                    ! Copy the coordinates of this point in the buffer.
+                    ! Update the counter jj accordingly.
+
+                    sendBuffer(jj) = flowDoms(d1, level, mm)%ks(i1, j1, k1)
+                    jj = jj + 1
+
+                end do
+
+                ! Send the data.
+
+                call mpi_isend(sendBuffer(ii), size, adflow_real, procID, &
+                               procID, ADflow_comm_world, sendRequests(i), &
+                               ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Set ii to jj for the next processor.
+
+                ii = jj
+
+            end do sends
+
+            ! Post the nonblocking receives.
+
+            ii = 1
+            receives: do i = 1, commPatternCell_2nd(level)%nProcRecv
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternCell_2nd(level)%recvProc(i)
+                size = commPatternCell_2nd(level)%nRecv(i)
+
+                ! Post the receive.
+
+                call mpi_irecv(recvBuffer(ii), size, adflow_real, procID, &
+                               myID, ADflow_comm_world, recvRequests(i), ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! And update ii.
+
+                ii = ii + size
+
+            end do receives
+
+            ! Copy the local data.
+            !DIR$ NOVECTOR
+            localCopy: do i = 1, internalCell_2nd(level)%nCopy
+
+                ! Store the block and the indices of the donor a bit easier.
+
+                d1 = internalCell_2nd(level)%donorBlock(i)
+                i1 = internalCell_2nd(level)%donorIndices(i, 1)
+                j1 = internalCell_2nd(level)%donorIndices(i, 2)
+                k1 = internalCell_2nd(level)%donorIndices(i, 3)
+                ! Idem for the halo's.
+
+                d2 = internalCell_2nd(level)%haloBlock(i)
+                i2 = internalCell_2nd(level)%haloIndices(i, 1)
+                j2 = internalCell_2nd(level)%haloIndices(i, 2)
+                k2 = internalCell_2nd(level)%haloIndices(i, 3)
+                ! Copy the coordinates.
+                flowDoms(d2, level, mm)%ks(i2, j2, k2) = &
+                    flowDoms(d1, level, mm)%ks(i1, j1, k1)
+
+            end do localCopy
+
+            ! Correct the periodic halos of the internal communication
+            ! pattern
+
+            ! call correctPeriodicCoor(level, mm, &
+            !                          internalCell_2nd(level)%nPeriodic, &
+            !                          internalCell_2nd(level)%periodicData)
+
+            ! Complete the nonblocking receives in an arbitrary sequence and
+            ! copy the coordinates from the buffer into the halo's.
+
+            size = commPatternCell_2nd(level)%nProcRecv
+            completeRecvs: do i = 1, commPatternCell_2nd(level)%nProcRecv
+
+                ! Complete any of the requests.
+
+                call mpi_waitany(size, recvRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Copy the data just arrived in the halo's.
+
+                ii = index
+                jj = commPatternCell_2nd(level)%nRecvCum(ii - 1) + 1
+                !DIR$ NOVECTOR
+                do j = 1, commPatternCell_2nd(level)%nRecv(ii)
+
+                    ! Store the block and the indices of the halo a bit easier.
+
+                    d2 = commPatternCell_2nd(level)%recvList(ii)%block(j)
+                    i2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 1)
+                    j2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 2)
+                    k2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 3)
+
+                    ! Copy the data.
+
+                    flowDoms(d2, level, mm)%ks(i2, j2, k2) = recvBuffer(jj)
+                    jj = jj + 1
+
+                end do
+
+            end do completeRecvs
+
+            ! Correct the periodic halos of the external communication
+            ! pattern.
+
+            ! call correctPeriodicCoor(level, mm, &
+            !                          commPatternCell_2nd(level)%nPeriodic, &
+            !                          commPatternCell_2nd(level)%periodicData)
+
+            ! Complete the nonblocking sends.
+
+            size = commPatternCell_2nd(level)%nProcSend
+            do i = 1, commPatternCell_2nd(level)%nProcSend
+                call mpi_waitany(size, sendRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+            end do
+
+        end do spectralLoop
+
+
+    end subroutine exchangeKs
+
+
     !      ==================================================================
 
     subroutine correctPeriodicCoor(level, sp, nPeriodic, periodicData)
@@ -2665,52 +3026,238 @@ subroutine correctPeriodicCoor(level, sp, nPeriodic, periodicData)
 
         ! Loop over the number of periodic transformations.
 
-        do nn = 1, nPeriodic
+        do nn = 1, nPeriodic
+
+            ! Store the rotation matrix, rotation center and translation
+            ! vector a bit easier.
+
+            rotMatrix = periodicData(nn)%rotMatrix
+            rotCenter = periodicData(nn)%rotCenter
+            translation = periodicData(nn)%translation + rotCenter
+
+            ! Loop over the number of halo nodes for this transformation.
+            !DIR$ NOVECTOR
+            do ii = 1, periodicData(nn)%nHalos
+
+                ! Store the block and the indices a bit easier.
+
+                mm = periodicData(nn)%block(ii)
+                i = periodicData(nn)%indices(ii, 1)
+                j = periodicData(nn)%indices(ii, 2)
+                k = periodicData(nn)%indices(ii, 3)
+
+                ! Determine the vector from the center of rotation to the
+                ! uncorrected halo value.
+
+                dx = flowDoms(mm, level, sp)%x(i, j, k, 1) - rotCenter(1)
+                dy = flowDoms(mm, level, sp)%x(i, j, k, 2) - rotCenter(2)
+                dz = flowDoms(mm, level, sp)%x(i, j, k, 3) - rotCenter(3)
+
+                ! Compute the corrected coordinates.
+
+                flowDoms(mm, level, sp)%x(i, j, k, 1) = rotMatrix(1, 1) * dx &
+                                                        + rotMatrix(1, 2) * dy &
+                                                        + rotMatrix(1, 3) * dz &
+                                                        + translation(1)
+                flowDoms(mm, level, sp)%x(i, j, k, 2) = rotMatrix(2, 1) * dx &
+                                                        + rotMatrix(2, 2) * dy &
+                                                        + rotMatrix(2, 3) * dz &
+                                                        + translation(2)
+                flowDoms(mm, level, sp)%x(i, j, k, 3) = rotMatrix(3, 1) * dx &
+                                                        + rotMatrix(3, 2) * dy &
+                                                        + rotMatrix(3, 3) * dz &
+                                                        + translation(3)
+            end do
+        end do
+
+    end subroutine correctPeriodicCoor
+    subroutine exchangeCoor_b(level)
+        !
+        !       ExchangeCoor_b exchanges the *derivatives* of the given grid
+        !       level IN REVERSE MODE.
+        !
+        use constants
+        use block
+        use communication
+        use inputTimeSpectral
+        use utils, only: EChk
+        implicit none
+        !
+        !      Subroutine arguments.
+        !
+        integer(kind=intType), intent(in) :: level
+        !
+        !      Local variables.
+        !
+        integer :: size, procID, ierr, index
+        integer, dimension(mpi_status_size) :: mpiStatus
+
+        integer(kind=intType) :: i, j, ii, jj, mm, idim
+        integer(kind=intType) :: d1, i1, j1, k1, d2, i2, j2, k2
+
+        ! Loop over the number of spectral solutions.
+
+        spectralLoop: do mm = 1, nTimeIntervalsSpectral
+
+            ! Send the coordinates i have to send. The data is first copied
+            ! into the send buffer and this buffer is sent.
+
+            ii = 1
+            jj = 1
+            recvs: do i = 1, commPatternNode_1st(level)%nProcRecv
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternNode_1st(level)%recvProc(i)
+                size = 3 * commPatternNode_1st(level)%nRecv(i)
+
+                ! Copy the data in the correct part of the send buffer.
+                !DIR$ NOVECTOR
+                do j = 1, commPatternNode_1st(level)%nRecv(i)
+
+                    ! Store the block id and the indices of the donor
+                    ! a bit easier.
+
+                    d1 = commPatternNode_1st(level)%recvList(i)%block(j)
+                    i1 = commPatternNode_1st(level)%recvList(i)%indices(j, 1)
+                    j1 = commPatternNode_1st(level)%recvList(i)%indices(j, 2)
+                    k1 = commPatternNode_1st(level)%recvList(i)%indices(j, 3)
+
+                    ! Copy the coordinates of this point in the buffer.
+                    ! Update the counter jj accordingly.
+
+                    recvBuffer(jj) = flowDomsd(d1, level, mm)%x(i1, j1, k1, 1)
+                    recvBuffer(jj + 1) = flowDomsd(d1, level, mm)%x(i1, j1, k1, 2)
+                    recvBuffer(jj + 2) = flowDomsd(d1, level, mm)%x(i1, j1, k1, 3)
+                    jj = jj + 3
+                    flowDomsd(d1, level, mm)%x(i1, j1, k1, :) = zero
+                end do
+
+                ! Send the data.
+
+                call mpi_isend(recvBuffer(ii), size, adflow_real, procID, &
+                               procID, ADflow_comm_world, recvRequests(i), &
+                               ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Set ii to jj for the next processor.
+
+                ii = jj
+
+            end do recvs
+
+            ! Post the nonblocking receives.
+
+            ii = 1
+            send: do i = 1, commPatternNode_1st(level)%nProcSend
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternNode_1st(level)%sendProc(i)
+                size = 3 * commPatternNode_1st(level)%nSend(i)
+
+                ! Post the receive.
+
+                call mpi_irecv(sendBuffer(ii), size, adflow_real, procID, &
+                               myID, ADflow_comm_world, sendRequests(i), ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! And update ii.
+
+                ii = ii + size
+
+            end do send
+
+            ! Copy the local data.
+            !DIR$ NOVECTOR
+            localCopy: do i = 1, internalNode_1st(level)%nCopy
+
+                ! Store the block and the indices of the donor a bit easier.
+
+                d1 = internalNode_1st(level)%donorBlock(i)
+                i1 = internalNode_1st(level)%donorIndices(i, 1)
+                j1 = internalNode_1st(level)%donorIndices(i, 2)
+                k1 = internalNode_1st(level)%donorIndices(i, 3)
+                ! Idem for the halo's.
+
+                d2 = internalNode_1st(level)%haloBlock(i)
+                i2 = internalNode_1st(level)%haloIndices(i, 1)
+                j2 = internalNode_1st(level)%haloIndices(i, 2)
+                k2 = internalNode_1st(level)%haloIndices(i, 3)
+
+                ! Sum into the '1' values fro the '2' values
+                do idim = 1, 3
+                    flowDomsd(d1, level, mm)%x(i1, j1, k1, idim) = flowDomsd(d1, level, mm)%x(i1, j1, k1, idim) + &
+                                                                   flowDomsd(d2, level, mm)%x(i2, j2, k2, idim)
+                    flowDomsd(d2, level, mm)%x(i2, j2, k2, idim) = zero
+                end do
+            end do localCopy
+
+            ! Correct the periodic halos of the internal communication
+            ! pattern
+
+            ! NOT IMPLEMENTED
+            ! call correctPeriodicCoor(level, mm,                          &
+            !      internalNode_1st(level)%nPeriodic,  &
+            !      internalNode_1st(level)%periodicData)
+
+            ! Complete the nonblocking receives in an arbitrary sequence and
+            ! copy the coordinates from the buffer into the halo's.
+
+            size = commPatternNode_1st(level)%nProcSend
+            completeSends: do i = 1, commPatternNode_1st(level)%nProcSend
+
+                ! Complete any of the requests.
+
+                call mpi_waitany(size, sendRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Copy the data just arrived in the halo's.
 
-            ! Store the rotation matrix, rotation center and translation
-            ! vector a bit easier.
+                ii = index
+                jj = 3 * commPatternNode_1st(level)%nSendCum(ii - 1)
+                !DIR$ NOVECTOR
+                do j = 1, commPatternNode_1st(level)%nSend(ii)
 
-            rotMatrix = periodicData(nn)%rotMatrix
-            rotCenter = periodicData(nn)%rotCenter
-            translation = periodicData(nn)%translation + rotCenter
+                    ! Store the block and the indices of the halo a bit easier.
 
-            ! Loop over the number of halo nodes for this transformation.
-            !DIR$ NOVECTOR
-            do ii = 1, periodicData(nn)%nHalos
+                    d2 = commPatternNode_1st(level)%sendList(ii)%block(j)
+                    i2 = commPatternNode_1st(level)%sendList(ii)%indices(j, 1)
+                    j2 = commPatternNode_1st(level)%sendList(ii)%indices(j, 2)
+                    k2 = commPatternNode_1st(level)%sendList(ii)%indices(j, 3)
 
-                ! Store the block and the indices a bit easier.
+                    ! Sum into the '2' values from the recv buffer
+                    do idim = 1, 3
+                        flowDomsd(d2, level, mm)%x(i2, j2, k2, idim) = flowDomsd(d2, level, mm)%x(i2, j2, k2, idim) + &
+                                                                       sendBuffer(jj + idim)
+                    end do
+                    jj = jj + 3
 
-                mm = periodicData(nn)%block(ii)
-                i = periodicData(nn)%indices(ii, 1)
-                j = periodicData(nn)%indices(ii, 2)
-                k = periodicData(nn)%indices(ii, 3)
+                end do
 
-                ! Determine the vector from the center of rotation to the
-                ! uncorrected halo value.
+            end do completeSends
 
-                dx = flowDoms(mm, level, sp)%x(i, j, k, 1) - rotCenter(1)
-                dy = flowDoms(mm, level, sp)%x(i, j, k, 2) - rotCenter(2)
-                dz = flowDoms(mm, level, sp)%x(i, j, k, 3) - rotCenter(3)
+            ! Correct the periodic halos of the external communication
+            ! pattern.
+            ! NOT IMLEMENTED
+            ! call correctPeriodicCoor(level, mm,                            &
+            !      commPatternNode_1st(level)%nPeriodic, &
+            !      commPatternNode_1st(level)%periodicData)
 
-                ! Compute the corrected coordinates.
+            ! Complete the nonblocking sends.
 
-                flowDoms(mm, level, sp)%x(i, j, k, 1) = rotMatrix(1, 1) * dx &
-                                                        + rotMatrix(1, 2) * dy &
-                                                        + rotMatrix(1, 3) * dz &
-                                                        + translation(1)
-                flowDoms(mm, level, sp)%x(i, j, k, 2) = rotMatrix(2, 1) * dx &
-                                                        + rotMatrix(2, 2) * dy &
-                                                        + rotMatrix(2, 3) * dz &
-                                                        + translation(2)
-                flowDoms(mm, level, sp)%x(i, j, k, 3) = rotMatrix(3, 1) * dx &
-                                                        + rotMatrix(3, 2) * dy &
-                                                        + rotMatrix(3, 3) * dz &
-                                                        + translation(3)
+            size = commPatternNode_1st(level)%nProcRecv
+            do i = 1, commPatternNode_1st(level)%nProcRecv
+                call mpi_waitany(size, recvRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
             end do
-        end do
 
-    end subroutine correctPeriodicCoor
-    subroutine exchangeCoor_b(level)
+        end do spectralLoop
+
+    end subroutine exchangeCoor_b
+    subroutine exchanged2Wall_b(level)
         !
         !       ExchangeCoor_b exchanges the *derivatives* of the given grid
         !       level IN REVERSE MODE.
@@ -2743,34 +3290,32 @@ subroutine exchangeCoor_b(level)
 
             ii = 1
             jj = 1
-            recvs: do i = 1, commPatternNode_1st(level)%nProcRecv
+            recvs: do i = 1, commPatternCell_2nd(level)%nProcRecv
 
                 ! Store the processor id and the size of the message
                 ! a bit easier.
 
-                procID = commPatternNode_1st(level)%recvProc(i)
-                size = 3 * commPatternNode_1st(level)%nRecv(i)
+                procID = commPatternCell_2nd(level)%recvProc(i)
+                size = commPatternCell_2nd(level)%nRecv(i)
 
                 ! Copy the data in the correct part of the send buffer.
                 !DIR$ NOVECTOR
-                do j = 1, commPatternNode_1st(level)%nRecv(i)
+                do j = 1, commPatternCell_2nd(level)%nRecv(i)
 
                     ! Store the block id and the indices of the donor
                     ! a bit easier.
 
-                    d1 = commPatternNode_1st(level)%recvList(i)%block(j)
-                    i1 = commPatternNode_1st(level)%recvList(i)%indices(j, 1)
-                    j1 = commPatternNode_1st(level)%recvList(i)%indices(j, 2)
-                    k1 = commPatternNode_1st(level)%recvList(i)%indices(j, 3)
+                    d2 = commPatternCell_2nd(level)%recvList(i)%block(j)
+                    i2 = commPatternCell_2nd(level)%recvList(i)%indices(j, 1)
+                    j2 = commPatternCell_2nd(level)%recvList(i)%indices(j, 2)
+                    k2 = commPatternCell_2nd(level)%recvList(i)%indices(j, 3)
 
                     ! Copy the coordinates of this point in the buffer.
                     ! Update the counter jj accordingly.
 
-                    recvBuffer(jj) = flowDomsd(d1, level, mm)%x(i1, j1, k1, 1)
-                    recvBuffer(jj + 1) = flowDomsd(d1, level, mm)%x(i1, j1, k1, 2)
-                    recvBuffer(jj + 2) = flowDomsd(d1, level, mm)%x(i1, j1, k1, 3)
-                    jj = jj + 3
-                    flowDomsd(d1, level, mm)%x(i1, j1, k1, :) = zero
+                    recvBuffer(jj) = flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2)
+                    flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2) = zero
+                    jj = jj + 1
                 end do
 
                 ! Send the data.
@@ -2789,13 +3334,13 @@ subroutine exchangeCoor_b(level)
             ! Post the nonblocking receives.
 
             ii = 1
-            send: do i = 1, commPatternNode_1st(level)%nProcSend
+            send: do i = 1, commPatternCell_2nd(level)%nProcSend
 
                 ! Store the processor id and the size of the message
                 ! a bit easier.
 
-                procID = commPatternNode_1st(level)%sendProc(i)
-                size = 3 * commPatternNode_1st(level)%nSend(i)
+                procID = commPatternCell_2nd(level)%sendProc(i)
+                size = commPatternCell_2nd(level)%nSend(i)
 
                 ! Post the receive.
 
@@ -2811,27 +3356,27 @@ subroutine exchangeCoor_b(level)
 
             ! Copy the local data.
             !DIR$ NOVECTOR
-            localCopy: do i = 1, internalNode_1st(level)%nCopy
+            localCopy: do i = 1, internalCell_2nd(level)%nCopy
 
                 ! Store the block and the indices of the donor a bit easier.
 
-                d1 = internalNode_1st(level)%donorBlock(i)
-                i1 = internalNode_1st(level)%donorIndices(i, 1)
-                j1 = internalNode_1st(level)%donorIndices(i, 2)
-                k1 = internalNode_1st(level)%donorIndices(i, 3)
+                d1 = internalCell_2nd(level)%donorBlock(i)
+                i1 = internalCell_2nd(level)%donorIndices(i, 1)
+                j1 = internalCell_2nd(level)%donorIndices(i, 2)
+                k1 = internalCell_2nd(level)%donorIndices(i, 3)
+
                 ! Idem for the halo's.
 
-                d2 = internalNode_1st(level)%haloBlock(i)
-                i2 = internalNode_1st(level)%haloIndices(i, 1)
-                j2 = internalNode_1st(level)%haloIndices(i, 2)
-                k2 = internalNode_1st(level)%haloIndices(i, 3)
+                d2 = internalCell_2nd(level)%haloBlock(i)
+                i2 = internalCell_2nd(level)%haloIndices(i, 1)
+                j2 = internalCell_2nd(level)%haloIndices(i, 2)
+                k2 = internalCell_2nd(level)%haloIndices(i, 3)
 
                 ! Sum into the '1' values fro the '2' values
-                do idim = 1, 3
-                    flowDomsd(d1, level, mm)%x(i1, j1, k1, idim) = flowDomsd(d1, level, mm)%x(i1, j1, k1, idim) + &
-                                                                   flowDomsd(d2, level, mm)%x(i2, j2, k2, idim)
-                    flowDomsd(d2, level, mm)%x(i2, j2, k2, idim) = zero
-                end do
+                flowDomsd(d1, level, mm)%d2Wall(i1, j1, k1) = flowDomsd(d1, level, mm)%d2Wall(i1, j1, k1) + &
+                                                              flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2)
+                flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2) = zero
+
             end do localCopy
 
             ! Correct the periodic halos of the internal communication
@@ -2845,8 +3390,8 @@ subroutine exchangeCoor_b(level)
             ! Complete the nonblocking receives in an arbitrary sequence and
             ! copy the coordinates from the buffer into the halo's.
 
-            size = commPatternNode_1st(level)%nProcSend
-            completeSends: do i = 1, commPatternNode_1st(level)%nProcSend
+            size = commPatternCell_2nd(level)%nProcSend
+            completeSends: do i = 1, commPatternCell_2nd(level)%nProcSend
 
                 ! Complete any of the requests.
 
@@ -2856,23 +3401,21 @@ subroutine exchangeCoor_b(level)
                 ! Copy the data just arrived in the halo's.
 
                 ii = index
-                jj = 3 * commPatternNode_1st(level)%nSendCum(ii - 1)
+                jj = commPatternCell_2nd(level)%nSendCum(ii - 1)
                 !DIR$ NOVECTOR
-                do j = 1, commPatternNode_1st(level)%nSend(ii)
+                do j = 1, commPatternCell_2nd(level)%nSend(ii)
 
                     ! Store the block and the indices of the halo a bit easier.
 
-                    d2 = commPatternNode_1st(level)%sendList(ii)%block(j)
-                    i2 = commPatternNode_1st(level)%sendList(ii)%indices(j, 1)
-                    j2 = commPatternNode_1st(level)%sendList(ii)%indices(j, 2)
-                    k2 = commPatternNode_1st(level)%sendList(ii)%indices(j, 3)
+                    d2 = commPatternCell_2nd(level)%sendList(ii)%block(j)
+                    i2 = commPatternCell_2nd(level)%sendList(ii)%indices(j, 1)
+                    j2 = commPatternCell_2nd(level)%sendList(ii)%indices(j, 2)
+                    k2 = commPatternCell_2nd(level)%sendList(ii)%indices(j, 3)
 
                     ! Sum into the '2' values from the recv buffer
-                    do idim = 1, 3
-                        flowDomsd(d2, level, mm)%x(i2, j2, k2, idim) = flowDomsd(d2, level, mm)%x(i2, j2, k2, idim) + &
-                                                                       sendBuffer(jj + idim)
-                    end do
-                    jj = jj + 3
+                    jj = jj + 1
+                    flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2) = flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2) + &
+                                                                  sendBuffer(jj)
 
                 end do
 
@@ -2887,15 +3430,15 @@ subroutine exchangeCoor_b(level)
 
             ! Complete the nonblocking sends.
 
-            size = commPatternNode_1st(level)%nProcRecv
-            do i = 1, commPatternNode_1st(level)%nProcRecv
+            size = commPatternCell_2nd(level)%nProcRecv
+            do i = 1, commPatternCell_2nd(level)%nProcRecv
                 call mpi_waitany(size, recvRequests, index, mpiStatus, ierr)
                 call EChk(ierr, __FILE__, __LINE__)
             end do
 
         end do spectralLoop
 
-    end subroutine exchangeCoor_b
+    end subroutine exchanged2Wall_b
     subroutine exchangeCoor_d(level)
         !
         !       ExchangeCoor_d exchanges the *derivatives* of the given grid
@@ -3083,6 +3626,185 @@ subroutine exchangeCoor_d(level)
         end do spectralLoop
 
     end subroutine exchangeCoor_d
+    subroutine exchanged2Wall_d(level)
+        !
+        !       ExchangeCoor_d exchanges the *derivatives* of the given grid
+        !       level.
+        !
+        use block
+        use communication
+        use inputTimeSpectral
+        use utils, only: EChk
+        implicit none
+        !
+        !      Subroutine arguments.
+        !
+        integer(kind=intType), intent(in) :: level
+        !
+        !      Local variables.
+        !
+        integer :: size, procID, ierr, index
+        integer, dimension(mpi_status_size) :: mpiStatus
+
+        integer(kind=intType) :: i, j, ii, jj, mm
+        integer(kind=intType) :: d1, i1, j1, k1, d2, i2, j2, k2
+
+        ! Loop over the number of spectral solutions.
+
+        spectralLoop: do mm = 1, nTimeIntervalsSpectral
+
+            ! Send the coordinates i have to send. The data is first copied
+            ! into the send buffer and this buffer is sent.
+
+            ii = 1
+            sends: do i = 1, commPatternCell_2nd(level)%nProcSend
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternCell_2nd(level)%sendProc(i)
+                size = commPatternCell_2nd(level)%nSend(i)
+
+                ! Copy the data in the correct part of the send buffer.
+
+                jj = ii
+                !DIR$ NOVECTOR
+                do j = 1, commPatternCell_2nd(level)%nSend(i)
+
+                    ! Store the block id and the indices of the donor
+                    ! a bit easier.
+
+                    d1 = commPatternCell_2nd(level)%sendList(i)%block(j)
+                    i1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 1)
+                    j1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 2)
+                    k1 = commPatternCell_2nd(level)%sendList(i)%indices(j, 3)
+
+                    ! Copy the coordinates of this point in the buffer.
+                    ! Update the counter jj accordingly.
+
+                    sendBuffer(jj) = flowDomsd(d1, level, mm)%d2Wall(i1, j1, k1)
+                    jj = jj + 1
+
+                end do
+
+                ! Send the data.
+
+                call mpi_isend(sendBuffer(ii), size, adflow_real, procID, &
+                               procID, ADflow_comm_world, sendRequests(i), &
+                               ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Set ii to jj for the next processor.
+
+                ii = jj
+
+            end do sends
+
+            ! Post the nonblocking receives.
+
+            ii = 1
+            receives: do i = 1, commPatternCell_2nd(level)%nProcRecv
+
+                ! Store the processor id and the size of the message
+                ! a bit easier.
+
+                procID = commPatternCell_2nd(level)%recvProc(i)
+                size = commPatternCell_2nd(level)%nRecv(i)
+
+                ! Post the receive.
+
+                call mpi_irecv(recvBuffer(ii), size, adflow_real, procID, &
+                               myID, ADflow_comm_world, recvRequests(i), ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! And update ii.
+
+                ii = ii + size
+
+            end do receives
+
+            ! Copy the local data.
+            !DIR$ NOVECTOR
+            localCopy: do i = 1, internalCell_2nd(level)%nCopy
+
+                ! Store the block and the indices of the donor a bit easier.
+
+                d1 = internalCell_2nd(level)%donorBlock(i)
+                i1 = internalCell_2nd(level)%donorIndices(i, 1)
+                j1 = internalCell_2nd(level)%donorIndices(i, 2)
+                k1 = internalCell_2nd(level)%donorIndices(i, 3)
+                ! Idem for the halo's.
+
+                d2 = internalCell_2nd(level)%haloBlock(i)
+                i2 = internalCell_2nd(level)%haloIndices(i, 1)
+                j2 = internalCell_2nd(level)%haloIndices(i, 2)
+                k2 = internalCell_2nd(level)%haloIndices(i, 3)
+                ! Copy the coordinates.
+                flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2) = &
+                    flowDomsd(d1, level, mm)%d2Wall(i1, j1, k1)
+
+            end do localCopy
+
+            ! Correct the periodic halos of the internal communication
+            ! pattern
+
+            ! NOT IMPLEMENTED
+            ! call correctPeriodicCoor(level, mm,                          &
+            !      internalCell_2nd(level)%nPeriodic,  &
+            !      internalCell_2nd(level)%periodicData)
+
+            ! Complete the nonblocking receives in an arbitrary sequence and
+            ! copy the coordinates from the buffer into the halo's.
+
+            size = commPatternCell_2nd(level)%nProcRecv
+            completeRecvs: do i = 1, commPatternCell_2nd(level)%nProcRecv
+
+                ! Complete any of the requests.
+
+                call mpi_waitany(size, recvRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Copy the data just arrived in the halo's.
+
+                ii = index
+                jj = commPatternCell_2nd(level)%nRecvCum(ii - 1) + 1
+                !DIR$ NOVECTOR
+                do j = 1, commPatternCell_2nd(level)%nRecv(ii)
+
+                    ! Store the block and the indices of the halo a bit easier.
+
+                    d2 = commPatternCell_2nd(level)%recvList(ii)%block(j)
+                    i2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 1)
+                    j2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 2)
+                    k2 = commPatternCell_2nd(level)%recvList(ii)%indices(j, 3)
+
+                    ! Copy the data.
+
+                    flowDomsd(d2, level, mm)%d2Wall(i2, j2, k2) = recvBuffer(jj)
+                    jj = jj + 1
+
+                end do
+
+            end do completeRecvs
+
+            ! Correct the periodic halos of the external communication
+            ! pattern.
+            ! NOT IMLEMENTED
+            ! call correctPeriodicCoor(level, mm,                            &
+            !      commPatternCell_2nd(level)%nPeriodic, &
+            !      commPatternCell_2nd(level)%periodicData)
+
+            ! Complete the nonblocking sends.
+
+            size = commPatternCell_2nd(level)%nProcSend
+            do i = 1, commPatternCell_2nd(level)%nProcSend
+                call mpi_waitany(size, sendRequests, index, mpiStatus, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+            end do
+
+        end do spectralLoop
+
+    end subroutine exchanged2Wall_d
 
     ! -----------------------------------------------------------------
     !              Comm routines for zippers
diff --git a/src/utils/utils.F90 b/src/utils/utils.F90
index b50cc8c17..a9a8b5a78 100644
--- a/src/utils/utils.F90
+++ b/src/utils/utils.F90
@@ -888,12 +888,14 @@ subroutine setBCPointers(nn, spatialPointers)
         use blockPointers, only: w, p, rlv, rev, gamma, x, d2wall, &
                                  si, sj, sk, s, globalCell, BCData, nx, il, ie, ib, &
                                  ny, jl, je, jb, nz, kl, ke, kb, BCFaceID, &
-                                 addgridvelocities, sFaceI, sFaceJ, sFaceK, addGridVelocities
+                                 addgridvelocities, sFaceI, sFaceJ, sFaceK, addGridVelocities, &
+                                 d2wall, ks
         use BCPointers, only: ww0, ww1, ww2, ww3, pp0, pp1, pp2, pp3, &
                               rlv0, rlv1, rlv2, rlv3, rev0, rev1, rev2, rev3, &
                               gamma0, gamma1, gamma2, gamma3, gcp, xx, ss, ssi, ssj, ssk, dd2wall, &
-                              sFace, iStart, iEnd, jStart, jEnd, iSize, jSize
-        use inputPhysics, only: cpModel, equations
+                              sFace, iStart, iEnd, jStart, jEnd, iSize, jSize, &
+                              d2wall0, d2wall1, d2wall2, d2wall3, ks0, ks1, ks2, ks3
+        use inputPhysics, only: cpModel, equations, useRoughSA
         implicit none
 
         ! Subroutine arguments.
@@ -939,6 +941,18 @@ subroutine setBCPointers(nn, spatialPointers)
             rev1 => rev(1, 1:, 1:)
             rev0 => rev(0, 1:, 1:)
 
+            d2wall3 => d2wall(3, 1:, 1:)
+            d2wall2 => d2wall(2, 1:, 1:)
+            d2wall1 => d2wall(1, 1:, 1:)
+            d2wall0 => d2wall(0, 1:, 1:)
+
+            if (useRoughSA) then
+                ks3 => ks(3, 1:, 1:)
+                ks2 => ks(2, 1:, 1:)
+                ks1 => ks(1, 1:, 1:)
+                ks0 => ks(0, 1:, 1:)
+            end if
+
             gamma3 => gamma(3, 1:, 1:)
             gamma2 => gamma(2, 1:, 1:)
             gamma1 => gamma(1, 1:, 1:)
@@ -969,6 +983,18 @@ subroutine setBCPointers(nn, spatialPointers)
             rev1 => rev(ie, 1:, 1:)
             rev0 => rev(ib, 1:, 1:)
 
+            d2wall3 => d2wall(nx, 1:, 1:)
+            d2wall2 => d2wall(il, 1:, 1:)
+            d2wall1 => d2wall(ie, 1:, 1:)
+            d2wall0 => d2wall(ib, 1:, 1:)
+
+            if (useRoughSA) then
+                ks3 => ks(nx, 1:, 1:)
+                ks2 => ks(il, 1:, 1:)
+                ks1 => ks(ie, 1:, 1:)
+                ks0 => ks(ib, 1:, 1:)
+            end if
+
             gamma3 => gamma(nx, 1:, 1:)
             gamma2 => gamma(il, 1:, 1:)
             gamma1 => gamma(ie, 1:, 1:)
@@ -999,6 +1025,18 @@ subroutine setBCPointers(nn, spatialPointers)
             rev1 => rev(1:, 1, 1:)
             rev0 => rev(1:, 0, 1:)
 
+            d2wall3 => d2wall(1:, 3, 1:)
+            d2wall2 => d2wall(1:, 2, 1:)
+            d2wall1 => d2wall(1:, 1, 1:)
+            d2wall0 => d2wall(1:, 0, 1:)
+
+            if (useRoughSA) then
+                ks3 => ks(1:, 3, 1:)
+                ks2 => ks(1:, 2, 1:)
+                ks1 => ks(1:, 1, 1:)
+                ks0 => ks(1:, 0, 1:)
+            end if
+
             gamma3 => gamma(1:, 3, 1:)
             gamma2 => gamma(1:, 2, 1:)
             gamma1 => gamma(1:, 1, 1:)
@@ -1029,6 +1067,18 @@ subroutine setBCPointers(nn, spatialPointers)
             rev1 => rev(1:, je, 1:)
             rev0 => rev(1:, jb, 1:)
 
+            d2wall3 => d2wall(1:, ny, 1:)
+            d2wall2 => d2wall(1:, jl, 1:)
+            d2wall1 => d2wall(1:, je, 1:)
+            d2wall0 => d2wall(1:, jb, 1:)
+
+            if (useRoughSA) then
+                ks3 => ks(1:, ny, 1:)
+                ks2 => ks(1:, jl, 1:)
+                ks1 => ks(1:, je, 1:)
+                ks0 => ks(1:, jb, 1:)
+            end if
+
             gamma3 => gamma(1:, ny, 1:)
             gamma2 => gamma(1:, jl, 1:)
             gamma1 => gamma(1:, je, 1:)
@@ -1059,6 +1109,18 @@ subroutine setBCPointers(nn, spatialPointers)
             rev1 => rev(1:, 1:, 1)
             rev0 => rev(1:, 1:, 0)
 
+            d2wall3 => d2wall(1:, 1:, 3)
+            d2wall2 => d2wall(1:, 1:, 2)
+            d2wall1 => d2wall(1:, 1:, 1)
+            d2wall0 => d2wall(1:, 1:, 0)
+
+            if (useRoughSA) then
+                ks3 => ks(1:, 1:, 3)
+                ks2 => ks(1:, 1:, 2)
+                ks1 => ks(1:, 1:, 1)
+                ks0 => ks(1:, 1:, 0)
+            end if
+
             gamma3 => gamma(1:, 1:, 3)
             gamma2 => gamma(1:, 1:, 2)
             gamma1 => gamma(1:, 1:, 1)
@@ -1089,6 +1151,18 @@ subroutine setBCPointers(nn, spatialPointers)
             rev1 => rev(1:, 1:, ke)
             rev0 => rev(1:, 1:, kb)
 
+            d2wall3 => d2wall(1:, 1:, nz)
+            d2wall2 => d2wall(1:, 1:, kl)
+            d2wall1 => d2wall(1:, 1:, ke)
+            d2wall0 => d2wall(1:, 1:, kb)
+
+            if (useRoughSA) then
+                ks3 => ks(1:, 1:, nz)
+                ks2 => ks(1:, 1:, kl)
+                ks1 => ks(1:, 1:, ke)
+                ks0 => ks(1:, 1:, kb)
+            end if
+
             gamma3 => gamma(1:, 1:, nz)
             gamma2 => gamma(1:, 1:, kl)
             gamma1 => gamma(1:, 1:, ke)
@@ -2056,14 +2130,16 @@ subroutine setbcpointers_d(nn, spatialpointers)
     &   gamma, x, xd, d2wall, d2walld, si, sid, sj, sjd, sk, skd, s, sd, &
     &   globalcell, bcdata, bcdatad, nx, il, ie, ib, ny, jl, je, jb, nz, kl,&
     &   ke, kb, bcfaceid, addgridvelocities, sfacei, sfaceid, sfacej, &
-    &   sfacejd, sfacek, sfacekd, addgridvelocities
+    &   sfacejd, sfacek, sfacekd, addgridvelocities, d2wall, d2walld, ks
         use bcpointers_d, only: ww0, ww0d, ww1, ww1d, ww2, ww2d, ww3, ww3d,&
     &   pp0, pp0d, pp1, pp1d, pp2, pp2d, pp3, pp3d, rlv0, rlv0d, rlv1, rlv1d&
     &   , rlv2, rlv2d, rlv3, rlv3d, rev0, rev0d, rev1, rev1d, rev2, rev2d, &
     &   rev3, rev3d, gamma0, gamma1, gamma2, gamma3, gcp, xx, xxd, ss, ssd, &
     &   ssi, ssid, ssj, ssjd, ssk, sskd, dd2wall, sface, istart, iend, &
-    &   jstart, jend, isize, jsize
-        use inputphysics, only: cpmodel, equations
+    &   jstart, jend, isize, jsize, &
+        d2wall0, d2wall0d, d2wall1, d2wall1d, d2wall2, d2wall2d, d2wall3, d2wall3d, &
+        ks0, ks1, ks2, ks3
+        use inputphysics, only: cpmodel, equations, useRoughSA
         implicit none
 ! subroutine arguments.
         integer(kind=inttype), intent(in) :: nn
@@ -2114,11 +2190,25 @@ subroutine setbcpointers_d(nn, spatialpointers)
             rev1 => rev(1, 1:, 1:)
             rev0d => revd(0, 1:, 1:)
             rev0 => rev(0, 1:, 1:)
+            d2wall3d => d2walld(3, 1:, 1:)
+            d2wall3 =>  d2wall(3, 1:, 1:)
+            d2wall2d => d2walld(2, 1:, 1:)
+            d2wall2 =>  d2wall(2, 1:, 1:)
+            d2wall1d => d2walld(1, 1:, 1:)
+            d2wall1 =>  d2wall(1, 1:, 1:)
+            d2wall0d => d2walld(0, 1:, 1:)
+            d2wall0 =>  d2wall(0, 1:, 1:)
             gamma3 => gamma(3, 1:, 1:)
             gamma2 => gamma(2, 1:, 1:)
             gamma1 => gamma(1, 1:, 1:)
             gamma0 => gamma(0, 1:, 1:)
             gcp => globalcell(2, 1:, 1:)
+            if (useRoughSA) then
+                ks3 => ks(3, 1:, 1:)
+                ks2 => ks(2, 1:, 1:)
+                ks1 => ks(1, 1:, 1:)
+                ks0 => ks(0, 1:, 1:)
+            end if
         case (imax)
 !---------------------------------------------------------------------------
             ww3d => wd(nx, 1:, 1:, :)
@@ -2153,11 +2243,25 @@ subroutine setbcpointers_d(nn, spatialpointers)
             rev1 => rev(ie, 1:, 1:)
             rev0d => revd(ib, 1:, 1:)
             rev0 => rev(ib, 1:, 1:)
+            d2wall3d => d2walld(nx, 1:, 1:)
+            d2wall3 =>  d2wall(nx, 1:, 1:)
+            d2wall2d => d2walld(il, 1:, 1:)
+            d2wall2 =>  d2wall(il, 1:, 1:)
+            d2wall1d => d2walld(ie, 1:, 1:)
+            d2wall1 =>  d2wall(ie, 1:, 1:)
+            d2wall0d => d2walld(ib, 1:, 1:)
+            d2wall0 =>  d2wall(ib, 1:, 1:)
             gamma3 => gamma(nx, 1:, 1:)
             gamma2 => gamma(il, 1:, 1:)
             gamma1 => gamma(ie, 1:, 1:)
             gamma0 => gamma(ib, 1:, 1:)
             gcp => globalcell(il, 1:, 1:)
+            if (useRoughSA) then
+                ks3 => ks(nx, 1:, 1:)
+                ks2 => ks(il, 1:, 1:)
+                ks1 => ks(ie, 1:, 1:)
+                ks0 => ks(ib, 1:, 1:)
+            end if
         case (jmin)
 !---------------------------------------------------------------------------
             ww3d => wd(1:, 3, 1:, :)
@@ -2192,11 +2296,25 @@ subroutine setbcpointers_d(nn, spatialpointers)
             rev1 => rev(1:, 1, 1:)
             rev0d => revd(1:, 0, 1:)
             rev0 => rev(1:, 0, 1:)
+            d2wall3d => d2walld(1:, 3, 1:)
+            d2wall3 =>  d2wall(1:, 3, 1:)
+            d2wall2d => d2walld(1:, 2, 1:)
+            d2wall2 =>  d2wall(1:, 2, 1:)
+            d2wall1d => d2walld(1:, 1, 1:)
+            d2wall1 =>  d2wall(1:, 1, 1:)
+            d2wall0d => d2walld(1:, 0, 1:)
+            d2wall0 =>  d2wall(1:, 0, 1:)
             gamma3 => gamma(1:, 3, 1:)
             gamma2 => gamma(1:, 2, 1:)
             gamma1 => gamma(1:, 1, 1:)
             gamma0 => gamma(1:, 0, 1:)
             gcp => globalcell(1:, 2, 1:)
+            if (useRoughSA) then
+                ks3 => ks(1:, 3, 1:)
+                ks2 => ks(1:, 2, 1:)
+                ks1 => ks(1:, 1, 1:)
+                ks0 => ks(1:, 0, 1:)
+            end if
         case (jmax)
 !---------------------------------------------------------------------------
             ww3d => wd(1:, ny, 1:, :)
@@ -2231,11 +2349,25 @@ subroutine setbcpointers_d(nn, spatialpointers)
             rev1 => rev(1:, je, 1:)
             rev0d => revd(1:, jb, 1:)
             rev0 => rev(1:, jb, 1:)
+            d2wall3d => d2walld(1:, ny, 1:)
+            d2wall3 =>  d2wall(1:, ny, 1:)
+            d2wall2d => d2walld(1:, jl, 1:)
+            d2wall2 =>  d2wall(1:, jl, 1:)
+            d2wall1d => d2walld(1:, je, 1:)
+            d2wall1 =>  d2wall(1:, je, 1:)
+            d2wall0d => d2walld(1:, jb, 1:)
+            d2wall0 =>  d2wall(1:, jb, 1:)
             gamma3 => gamma(1:, ny, 1:)
             gamma2 => gamma(1:, jl, 1:)
             gamma1 => gamma(1:, je, 1:)
             gamma0 => gamma(1:, jb, 1:)
             gcp => globalcell(1:, jl, 1:)
+            if (useRoughSA) then
+                ks3 => ks(1:, ny, 1:)
+                ks2 => ks(1:, jl, 1:)
+                ks1 => ks(1:, je, 1:)
+                ks0 => ks(1:, jb, 1:)
+            end if
         case (kmin)
 !---------------------------------------------------------------------------
             ww3d => wd(1:, 1:, 3, :)
@@ -2270,11 +2402,25 @@ subroutine setbcpointers_d(nn, spatialpointers)
             rev1 => rev(1:, 1:, 1)
             rev0d => revd(1:, 1:, 0)
             rev0 => rev(1:, 1:, 0)
+            d2wall3d => d2walld(1:, 1:, 3)
+            d2wall3 =>  d2wall(1:, 1:, 3)
+            d2wall2d => d2walld(1:, 1:, 2)
+            d2wall2 =>  d2wall(1:, 1:, 2)
+            d2wall1d => d2walld(1:, 1:, 1)
+            d2wall1 =>  d2wall(1:, 1:, 1)
+            d2wall0d => d2walld(1:, 1:, 0)
+            d2wall0 =>  d2wall(1:, 1:, 0)
             gamma3 => gamma(1:, 1:, 3)
             gamma2 => gamma(1:, 1:, 2)
             gamma1 => gamma(1:, 1:, 1)
             gamma0 => gamma(1:, 1:, 0)
             gcp => globalcell(1:, 1:, 2)
+            if (useRoughSA) then
+                ks3 => ks(1:, 1:, 3)
+                ks2 => ks(1:, 1:, 2)
+                ks1 => ks(1:, 1:, 1)
+                ks0 => ks(1:, 1:, 0)
+            end if
         case (kmax)
 !---------------------------------------------------------------------------
             ww3d => wd(1:, 1:, nz, :)
@@ -2309,11 +2455,25 @@ subroutine setbcpointers_d(nn, spatialpointers)
             rev1 => rev(1:, 1:, ke)
             rev0d => revd(1:, 1:, kb)
             rev0 => rev(1:, 1:, kb)
+            d2wall3d => d2walld(1:, 1:, nz)
+            d2wall3 =>  d2wall(1:, 1:, nz)
+            d2wall2d => d2walld(1:, 1:, kl)
+            d2wall2 =>  d2wall(1:, 1:, kl)
+            d2wall1d => d2walld(1:, 1:, ke)
+            d2wall1 =>  d2wall(1:, 1:, ke)
+            d2wall0d => d2walld(1:, 1:, kb)
+            d2wall0 =>  d2wall(1:, 1:, kb)
             gamma3 => gamma(1:, 1:, nz)
             gamma2 => gamma(1:, 1:, kl)
             gamma1 => gamma(1:, 1:, ke)
             gamma0 => gamma(1:, 1:, kb)
             gcp => globalcell(1:, 1:, kl)
+            if (useRoughSA) then
+                ks3 => ks(1:, 1:, nz)
+                ks2 => ks(1:, 1:, kl)
+                ks1 => ks(1:, 1:, ke)
+                ks0 => ks(1:, 1:, kb)
+            end if
         end select
         if (spatialpointers) then
             select case (bcfaceid(nn))
@@ -3242,6 +3402,7 @@ subroutine setPointers(nn, mm, ll)
         ! from blockPointers so use a bare use.
         use constants
         use blockPointers
+        use inputPhysics, only: useRoughSA
         implicit none
         !
         !      Subroutine arguments
@@ -3480,6 +3641,9 @@ subroutine setPointers(nn, mm, ll)
 
         d2Wall => flowDoms(nn, mm, ll)%d2Wall
         filterDES => flowDoms(nn, mm, ll)%filterDES  ! eran-des
+        if (useRoughSA) then
+            ks => flowDoms(nn, mm, ll)%ks
+        end if
 
         ! Arrays used for the implicit treatment of the turbulent wall
         ! boundary conditions. As these variables are only allocated for
@@ -4600,7 +4764,7 @@ subroutine deallocDerivativeValues(level)
         use inputtimespectral, only: nTimeIntervalsSpectral
         use wallDistanceData, only: xSurfVec, xSurfVecd
         use flowVarRefState, only: winfd
-        use inputPhysics, only: wallDistanceNeeded
+        use inputPhysics, only: wallDistanceNeeded, useRoughSA
         use adjointVars, only: derivVarsAllocated
         use BCPointers_b
 
@@ -4917,6 +5081,10 @@ subroutine deallocateBlock(nn, level, sps)
                 deallocate (BCData(i)%TNS_Wall, stat=ierr)
             if (ierr /= 0) deallocationFailure = .true.
 
+            if (associated(BCData(i)%ksNS_Wall)) &
+                deallocate (BCData(i)%ksNS_Wall, stat=ierr)
+            if (ierr /= 0) deallocationFailure = .true.
+
             if (associated(BCData(i)%ptInlet)) &
                 deallocate (BCData(i)%ptInlet, stat=ierr)
             if (ierr /= 0) deallocationFailure = .true.
@@ -4996,6 +5164,7 @@ subroutine deallocateBlock(nn, level, sps)
 
             nullify (BCData(i)%uSlip)
             nullify (BCData(i)%TNS_Wall)
+            nullify (BCData(i)%ksNS_Wall)
 
             nullify (BCData(i)%normALE)
             nullify (BCData(i)%rfaceALE)
diff --git a/src/wallDistance/wallDistance.F90 b/src/wallDistance/wallDistance.F90
index 105cd5e6d..1a9845a4d 100644
--- a/src/wallDistance/wallDistance.F90
+++ b/src/wallDistance/wallDistance.F90
@@ -126,6 +126,216 @@ end subroutine updateWallDistancesQuickly
     ! ----------------------------------------------------------------------
 
 #ifndef USE_TAPENADE
+
+    subroutine updateWallRoughness()
+
+        ! Sets the roughness-value (ks) of the nearest wall-cell in the volume cells.
+        !
+        ! At first, it creates two lists: (1) ks values on the surface; (2) global
+        ! cellIndex corresponding to this ks-value.
+        !
+        ! Then it gathers the full list on each proc *THIS DOES NOT SCALE IN MEMORY*
+        !
+        ! After that, it iterate through every volume cell and finds the index in
+        ! list (1) that corresponds to the cellIndex of the nearest surface-cell.
+        ! Then it uses this index to set the ks value listed in (2).
+        !
+        !
+        ! A more memory efficient approach would be to create a 'PETSc Scatter'.
+        ! This should be straight forward using the cellIndex-list mentioned above.
+        ! You might take a look at 'wallScatter' further down this file for
+        ! inspiration.
+
+        use constants
+        use blockPointers
+        use inputTimeSpectral, only: nTimeIntervalsSpectral
+        use inputPhysics, only: useRoughSA
+        use utils, only: setPointers, EChk, terminate
+        use surfaceFamilies, only: BCFamGroups
+        use communication, only: adflow_comm_world, nProc, myID
+        use sorting, only: famInList
+        use wallDistanceData, only: nCellBlockOffset
+        use haloExchange, only : exchangeKs
+        implicit none
+
+        ! Local Variables
+        integer(kind=intType) :: i, j, k, ii, jj, ierr, iCell
+        integer(kind=intType) :: iBeg, jBeg, iEnd, jEnd, ni, nj
+        integer(kind=intType) :: nn, sps, level, nLevels, mm
+        integer(kind=intType), dimension(:), allocatable :: nCellProc, cumCellProc
+        integer(kind=intType), dimension(:), pointer :: wallFamList
+        integer(kind=intType), dimension(:), allocatable :: cellIdLocal, cellIdGlobal
+        integer(kind=intType) :: nCellsLocal, nCellsGlobal
+        real(kind=realType), dimension(:), allocatable :: ksLocal, ksGlobal
+
+        character(len=maxStringLen) :: errorMessage
+
+        ! exit if not in use
+        if (.not. useRoughSA) then
+            return
+        end if
+
+        wallFamList => BCFamGroups(iBCGroupWalls)%famList
+        nLevels = ubound(flowDoms, 2)
+
+        do level = 1, nLevels
+            do sps = 1, nTimeIntervalsSpectral
+
+                ! figure out the local space needed
+                nCellsLocal = 0
+                do nn = 1, nDom
+                    call setPointers(nn, level, sps)
+
+                    do mm = 1, nBocos
+                        if (.not. famInList(BCdata(mm)%famID, wallFamlist)) then
+                            cycle
+                        end if
+                        nCellsLocal = nCellsLocal + &
+                                      (bcData(mm)%inEnd - bcData(mm)%inBeg) * (bcData(mm)%jnEnd - bcData(mm)%jnBeg)
+                    end do
+                end do
+
+                ! Now communicate these sizes with everyone
+                allocate (nCellProc(nProc), cumCellProc(0:nProc))
+
+                call mpi_allgather(nCellsLocal, 1, adflow_integer, nCellProc, 1, adflow_integer, &
+                                   adflow_comm_world, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! Now make cumulative versions of these
+                cumCellProc(0) = 0
+                do i = 1, nProc
+                    cumCellProc(i) = cumCellProc(i - 1) + nCellProc(i)
+                end do
+
+                ! And save the total number of nodes and cells for reference
+                nCellsGlobal = cumCellProc(nProc)
+
+                ! Allocate the space for the local ks values and cellId's
+                allocate (ksLocal(nCellsLocal), cellIdLocal(nCellsLocal))
+
+                ! Move all the local ks-values in a list
+                ! Create a second list with the global cell ID corresponding to the ks-values
+                iCell = 0
+                do nn = 1, nDom
+                    call setPointers(nn, level, sps)
+                    do mm = 1, nBocos
+                        if (.not. famInList(BCdata(mm)%famID, wallFamlist)) then
+                            cycle
+                        end if
+
+                        jBeg = BCData(mm)%jnBeg; jEnd = BCData(mm)%jnEnd
+                        iBeg = BCData(mm)%inBeg; iEnd = BCData(mm)%inEnd
+                        ni = iEnd - iBeg
+                        nj = jEnd - jBeg
+
+                        do jj = 1, nj
+                            do ii = 1, ni
+                                iCell = iCell + 1
+
+                                ! saving local ks-value is easy
+                                ksLocal(iCell) = BCData(mm)%ksNS_Wall(ii, jj)
+
+                                ! to calculate the global cellID, we must associate the
+                                ! BC-cell to the volume cell first. We basically have to
+                                ! set surface i-j values to global i,j,k values.
+
+                                select case (BCFaceID(mm))
+                                case (iMin)
+                                    i = 2
+                                    j = ii + 1
+                                    k = jj + 1
+                                case (iMax)
+                                    i = il
+                                    j = ii + 1
+                                    k = jj + 1
+                                case (jMin)
+                                    i = ii + 1
+                                    j = 2
+                                    k = jj + 1
+                                case (jMax)
+                                    i = ii + 1
+                                    j = jl
+                                    k = jj + 1
+                                case (kMin)
+                                    i = ii + 1
+                                    j = jj + 1
+                                    k = 2
+                                case (kMax)
+                                    i = ii + 1
+                                    j = jj + 1
+                                    k = kl
+                                end select
+
+                                cellIdLocal(iCell) = nCellBLockOffset(level, nn) * nTimeIntervalsSpectral + &
+                                                     nx * ny * nz * (sps - 1) + (i - 2) + (j - 2) * nx + &
+                                                     (k - 2) * nx * ny
+                            end do
+                        end do
+                    end do
+                end do
+
+                ! allocate global arrays
+                allocate (ksGlobal(nCellsGlobal), cellIdGlobal(nCellsGlobal))
+
+                ! gather all the surface-ks values on each proc
+                call mpi_allgatherv(ksLocal, nCellsLocal, adflow_real, &
+                                    ksGlobal, nCellProc, cumCellProc, adflow_real, &
+                                    adflow_comm_world, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! gather all the cellId's on each proc
+                call mpi_allgatherv(cellIdLocal, nCellsLocal, adflow_integer, &
+                                    cellIdGlobal, nCellProc, cumCellProc, adflow_integer, &
+                                    adflow_comm_world, ierr)
+                call EChk(ierr, __FILE__, __LINE__)
+
+                ! free local memory
+                deallocate (cumCellProc, nCellProc, ksLocal, cellIdLocal)
+
+                ! set the ks-values in the volume
+                do nn = 1, nDom
+                    call setPointers(nn, level, sps)
+                    do k = 2, kl
+                        do j = 2, jl
+                            do i = 2, il
+                                if (flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) == -1) then
+                                    ! This cell is too far away and has no
+                                    ! association. Set the roughness to zero.
+                                    ks(i, j, k) = zero
+                                    cycle
+                                end if
+
+                                ! find the index of the surface cell (Requires gfortran > 9.0 )
+                                iCell = findloc( &
+                                cellIdGlobal, &
+                                    flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k), &
+                                    DIM = 1)
+
+                                if (iCell == 0) then
+                                    write (errorMessage, 100) &
+                                        flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k)
+100                                 format("Could not find surface cell with id ", I10.1)
+                                    call terminate("updateWallRoughness", errorMessage)
+                                end if
+
+                                ! set the ks value
+                                ks(i, j, k) = ksGlobal(iCell)
+                            end do
+                        end do
+                    end do
+                end do
+
+                ! free global memory
+                deallocate (ksGlobal, cellIdGlobal)
+            end do
+        end do
+
+
+        call exchangeKs(1)
+
+    end subroutine updateWallRoughness
+
     subroutine computeWallDistance(level, allocMem)
         !
         !       wallDistance computes the distances of the cell centers to
@@ -474,6 +684,7 @@ subroutine initWallDistance(level, sps, allocMem)
         !
         use constants
         use blockPointers, only: nDom, flowDoms
+        use inputPhysics, only: useRoughSA
         use utils, only: terminate
         implicit none
         !
@@ -486,7 +697,7 @@ subroutine initWallDistance(level, sps, allocMem)
         !
         integer :: ierr
 
-        integer(kind=intType) :: nn, il, jl, kl
+        integer(kind=intType) :: nn, ib, jb, kb
 
         ! Loop over the domains.
 
@@ -496,15 +707,26 @@ subroutine initWallDistance(level, sps, allocMem)
 
             if (allocMem) then
 
-                il = flowDoms(nn, level, sps)%il
-                jl = flowDoms(nn, level, sps)%jl
-                kl = flowDoms(nn, level, sps)%kl
+                ib = flowDoms(nn, level, sps)%ib
+                jb = flowDoms(nn, level, sps)%jb
+                kb = flowDoms(nn, level, sps)%kb
 
-                allocate (flowDoms(nn, level, sps)%d2Wall(2:il, 2:jl, 2:kl), &
+                allocate (flowDoms(nn, level, sps)%d2Wall(0:ib, 0:jb, 0:kb), &
                           stat=ierr)
                 if (ierr /= 0) &
                     call terminate("initWallDistance", &
                                    "Memory allocation failure for d2Wall")
+
+                if (useRoughSA) then
+                    allocate (flowDoms(nn, level, sps)%ks(0:ib, 0:jb, 0:kb), &
+                              stat=ierr)
+                    if (ierr /= 0) &
+                        call terminate("initWallDistance", &
+                                       "Memory allocation failure for ks")
+
+                    ! initalize to zero
+                    flowDoms(nn, level, sps)%ks = zero
+                end if
             end if
 
             ! Initialize the wall distances to a large value.
@@ -1531,6 +1753,8 @@ subroutine updateWallDistanceAllLevels
         use block, only: flowDoms
         use inputPhysics, only: equations
         use iteration, only: groundLevel
+        use haloExchange, only: exchanged2Wall
+        use inputPhysics, only: wallDistanceNeeded
         implicit none
         !
         !      Local variables.
@@ -1546,6 +1770,9 @@ subroutine updateWallDistanceAllLevels
         nLevels = ubound(flowDoms, 2)
         do nn = groundLevel, nLevels
             call computeWallDistance(nn, .false.)
+            if (wallDistanceNeeded) then
+                call exchanged2Wall(nn)
+            end if
         end do
 
     end subroutine updateWallDistanceAllLevels
@@ -1737,6 +1964,10 @@ subroutine determineWallAssociation(level, sps)
             allocate (fullWall%conn(4, nCells))
             allocate (fullWall%ind(nNodes))
 
+            if (useRoughSA) then
+                allocate (fullWall%indCell(nCells))
+            end if
+
             nNodes = 0
             nCells = 0
             ii = 0
@@ -1753,6 +1984,9 @@ subroutine determineWallAssociation(level, sps)
                 do j = 1, walls(i)%nCells
                     nCells = nCells + 1
                     fullWall%conn(:, nCells) = walls(i)%conn(:, j) + ii
+                    if (useRoughSA) then
+                        fullWall%indCell(nCells) = walls(i)%indCell(j)
+                    end if
                 end do
 
                 ! Increment the node offset
@@ -1781,6 +2015,9 @@ subroutine determineWallAssociation(level, sps)
             if (.not. associated(flowDoms(nn, level, sps)%surfNodeIndices)) then
                 allocate (flowDoms(nn, level, sps)%surfNodeIndices(4, 2:il, 2:jl, 2:kl))
                 allocate (flowDoms(nn, level, sps)%uv(2, 2:il, 2:jl, 2:kl))
+                if (useRoughSA) then
+                    allocate (flowDoms(nn, level, sps)%nearestWallCellInd(2:il, 2:jl, 2:kl))
+                end if
             end if
 
             ! Set the cluster for this block
@@ -1823,10 +2060,16 @@ subroutine determineWallAssociation(level, sps)
                                         walls(c)%ind(walls(c)%conn(kk, cellID))
                                 end do
                                 flowDoms(nn, level, sps)%uv(:, i, j, k) = uvw(1:2)
+                                if (useRoughSA) then
+                                    flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) = walls(c)%indCell(cellID)
+                                end if
                             else
                                 ! Just set dummy values. These will never be used.
                                 flowDoms(nn, level, sps)%surfNodeIndices(:, i, j, k) = 0
                                 flowDoms(nn, level, sps)%uv(:, i, j, k) = 0
+                                if (useRoughSA) then
+                                    flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) = -1
+                                end if
                             end if
 
                             ! We are done with this point.
@@ -1856,7 +2099,9 @@ subroutine determineWallAssociation(level, sps)
                                         fullWall%ind(fullWall%conn(kk, cellID))
                                 end do
                                 flowDoms(nn, level, sps)%uv(:, i, j, k) = uvw(1:2)
-
+                                if (useRoughSA) then
+                                    flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) = fullWall%indCell(cellID)
+                                end if
                             else
 
                                 ! This point is *closer* than the nearWallDist AND
@@ -1876,6 +2121,9 @@ subroutine determineWallAssociation(level, sps)
                                             walls(c)%ind(walls(c)%conn(kk, cellID2))
                                     end do
                                     flowDoms(nn, level, sps)%uv(:, i, j, k) = uvw2(1:2)
+                                    if (useRoughSA) then
+                                        flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) = fullWall%indCell(cellID2)
+                                    end if
                                 else
                                     ! The full wall distance is better. Take that.
 
@@ -1884,7 +2132,9 @@ subroutine determineWallAssociation(level, sps)
                                             fullWall%ind(fullWall%conn(kk, cellID))
                                     end do
                                     flowDoms(nn, level, sps)%uv(:, i, j, k) = uvw(1:2)
-
+                                    if (useRoughSA) then
+                                        flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) = fullWall%indCell(cellID)
+                                    end if
                                 end if
                             end if
                         else
@@ -1895,7 +2145,9 @@ subroutine determineWallAssociation(level, sps)
 
                             flowDoms(nn, level, sps)%surfNodeIndices(:, i, j, k) = 0
                             flowDoms(nn, level, sps)%uv(:, i, j, k) = 0
-
+                            if (useRoughSA) then
+                                flowDoms(nn, level, sps)%nearestWallCellInd(i, j, k) = -1
+                            end if
                         end if
                     end do
                 end do
diff --git a/tests/reg_tests/refs/adjoint_rans_rough_sa.json b/tests/reg_tests/refs/adjoint_rans_rough_sa.json
new file mode 100644
index 000000000..11a894b86
--- /dev/null
+++ b/tests/reg_tests/refs/adjoint_rans_rough_sa.json
@@ -0,0 +1,824 @@
+{
+    "Eval Functions Sens:": {
+        "mdo_tutorial_cd": {
+            "P_mdo_tutorial": 1.1906967920303641e-08,
+            "T_mdo_tutorial": 5.7032101376918994e-08,
+            "alpha_mdo_tutorial": 0.005064588210334737,
+            "beta_mdo_tutorial": -0.0008845056168120727,
+            "mach_mdo_tutorial": 0.09103183719324591,
+            "shape": {
+                "__ndarray__": [
+                    [
+                        0.0014588561633413414,
+                        -0.00014952202413208286,
+                        -0.015675497912023488,
+                        -0.014075987294571204,
+                        0.0013475090668360465,
+                        0.0018275783162532001,
+                        -0.012563970825243631,
+                        -0.010466961719321863,
+                        0.001309403504721567,
+                        -0.003255740658843033,
+                        0.0022778281149183153,
+                        0.006688705333327987,
+                        -0.0018114416783712132,
+                        0.005977863590813565,
+                        0.00983532858473589,
+                        0.007136309610539732,
+                        0.0005148684493146728,
+                        -0.0019678518803612076,
+                        0.002182198480909613,
+                        0.004428502895915052,
+                        0.004347966506713412,
+                        0.0036422607708835367,
+                        0.003191112901777175,
+                        0.004015687155590336,
+                        0.002041987254186593,
+                        0.00216664730487107,
+                        0.0019894942777849856,
+                        0.0016303801990091183,
+                        -0.019657992669769538,
+                        -0.020829881820670266,
+                        -0.02078458749231964,
+                        -0.018884803905128002,
+                        -0.0003694888783061555,
+                        -0.0003130644455712657,
+                        -0.0003057568816089407,
+                        0.0002687965388743136,
+                        -0.017918307322412758,
+                        -0.018865739227695775,
+                        -0.018591839838973466,
+                        -0.016488737753057772,
+                        0.0009126579204761291,
+                        0.0008121165240817337,
+                        0.0008901093226245515,
+                        0.0010405934847586217,
+                        -0.004064662192245191,
+                        -0.0037963248994502177,
+                        -0.0032041778422513857,
+                        -0.002629339527700363,
+                        0.0033331749562904543,
+                        0.0034921467579424295,
+                        0.003357522945849655,
+                        0.0030107497004535763,
+                        0.007440267908310526,
+                        0.007319659848397562,
+                        0.006915771850659987,
+                        0.00616224817217152,
+                        0.0004202215889943551,
+                        0.003131548800067598,
+                        0.005383622398366069,
+                        0.0068249401797487366,
+                        0.01037563525198086,
+                        0.011987962339168481,
+                        0.011950338215280503,
+                        0.009432492888063287,
+                        0.011250802433469834,
+                        0.009854871187475862,
+                        0.008009282452942326,
+                        0.005465380393532938,
+                        0.00617765891204428,
+                        0.005047041663475567,
+                        0.004344579308030662,
+                        0.004048996264051067
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    72
+                ]
+            },
+            "span": 0.00229664892833175,
+            "twist": {
+                "__ndarray__": [
+                    [
+                        -0.0009085025652585955,
+                        -0.0011352774844096654,
+                        -0.001113026381374366,
+                        -0.0009513310772895932,
+                        -0.0006868004819044367,
+                        -0.0003151052366628259
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    6
+                ]
+            },
+            "xRef_mdo_tutorial": 0.0,
+            "yRef_mdo_tutorial": 0.0,
+            "zRef_mdo_tutorial": 0.0
+        },
+        "mdo_tutorial_cl": {
+            "P_mdo_tutorial": -2.2527598084617437e-07,
+            "T_mdo_tutorial": -1.3233798184113253e-07,
+            "alpha_mdo_tutorial": 0.11733097471392283,
+            "beta_mdo_tutorial": -0.005417244776871343,
+            "mach_mdo_tutorial": 0.44617528313282584,
+            "shape": {
+                "__ndarray__": [
+                    [
+                        0.0006642837325210593,
+                        -0.009763975859965732,
+                        -0.3801481999887656,
+                        -0.3170998426235122,
+                        0.012629573615866977,
+                        0.02315397958563808,
+                        -0.19944009087270742,
+                        -0.1691912552790567,
+                        0.0351498387083226,
+                        0.06625716321778724,
+                        0.09120797369196265,
+                        0.15959099100025592,
+                        0.07840389483505791,
+                        0.06061702563900386,
+                        0.0471183163334587,
+                        0.1618610610277068,
+                        0.027871493172952767,
+                        0.03253649812508121,
+                        0.04177297503503383,
+                        0.05856781186588396,
+                        0.0313022523470325,
+                        0.02349066959756217,
+                        0.04666961628569403,
+                        0.07533994488661211,
+                        0.008047029629268385,
+                        0.013538816921931512,
+                        0.01504317535187119,
+                        0.01571547469209155,
+                        -0.46385021365505386,
+                        -0.4739910307711922,
+                        -0.44823629888730077,
+                        -0.3689000683369753,
+                        0.008718298792151594,
+                        0.029550012094386215,
+                        0.043126338752077574,
+                        0.04434158317192526,
+                        -0.3956395141009607,
+                        -0.4079461176825463,
+                        -0.3862046757086361,
+                        -0.3163656910038524,
+                        0.06398787247549687,
+                        0.07479431304247744,
+                        0.07286149409348636,
+                        0.05846138573422667,
+                        0.08736651118290667,
+                        0.08697245700465966,
+                        0.0800864819250691,
+                        0.06426740266179096,
+                        0.10936456656828654,
+                        0.10829983086081271,
+                        0.10157510658835196,
+                        0.08255237277265481,
+                        0.174394811331467,
+                        0.16923186958607747,
+                        0.15458484011444382,
+                        0.1203484559615018,
+                        0.11961152090585189,
+                        0.12724633074407918,
+                        0.11338965533117872,
+                        0.07770697893018691,
+                        0.05670971831375479,
+                        0.03836441180009568,
+                        0.024469827398488818,
+                        0.019976171681793435,
+                        0.04934729534821406,
+                        0.05111766067208243,
+                        0.056306951989697436,
+                        0.060501561915658614,
+                        0.18336415728743627,
+                        0.1839642086404485,
+                        0.172950832038159,
+                        0.14031382776067444
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    72
+                ]
+            },
+            "span": 0.04345766409999452,
+            "twist": {
+                "__ndarray__": [
+                    [
+                        -0.024654336084479405,
+                        -0.028750831788918046,
+                        -0.025904206045517835,
+                        -0.020158455062478375,
+                        -0.012724745076232755,
+                        -0.004816128687393934
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    6
+                ]
+            },
+            "xRef_mdo_tutorial": 0.0,
+            "yRef_mdo_tutorial": 0.0,
+            "zRef_mdo_tutorial": 0.0
+        },
+        "mdo_tutorial_cmz": {
+            "P_mdo_tutorial": -3.656206038639332e-07,
+            "T_mdo_tutorial": -2.3220688985427506e-07,
+            "alpha_mdo_tutorial": 0.15587229388605317,
+            "beta_mdo_tutorial": -0.009601346793082532,
+            "mach_mdo_tutorial": 0.8704444540175915,
+            "shape": {
+                "__ndarray__": [
+                    [
+                        -0.029047420922817188,
+                        -0.04708038357322772,
+                        -0.49032488213844316,
+                        -0.4013165545103022,
+                        0.009544996586770077,
+                        0.026725344461077258,
+                        -0.44647176923194504,
+                        -0.3753146224937229,
+                        0.01782744019989366,
+                        0.08962447709312292,
+                        0.13469816620993494,
+                        0.2308823656803091,
+                        0.06582854869951546,
+                        0.07763779529085339,
+                        0.09150764410109205,
+                        0.2507771488956023,
+                        0.05119051469137706,
+                        0.07611290577460351,
+                        0.10139817895149289,
+                        0.14288701914486024,
+                        0.05984782406821292,
+                        0.060871228049129186,
+                        0.11793992001578524,
+                        0.18543399479046915,
+                        -0.0357932204896516,
+                        -0.02906455119109569,
+                        -0.01801481334783786,
+                        -0.0024114995232683567,
+                        -0.6512116044666574,
+                        -0.7548564706156609,
+                        -0.8101131292991757,
+                        -0.7525335106763098,
+                        -0.044525501572274084,
+                        -0.017099525064875748,
+                        0.01681056318887561,
+                        0.041434083335101235,
+                        -0.545809688007461,
+                        -0.640587162239928,
+                        -0.6906120551468395,
+                        -0.6398075609220193,
+                        0.054791469450099516,
+                        0.08331789087620665,
+                        0.09984195609069375,
+                        0.09540815288288124,
+                        0.1306150031496364,
+                        0.14676641870076226,
+                        0.15217947592334216,
+                        0.13677505061708325,
+                        0.17617158370280409,
+                        0.1955737376811594,
+                        0.20395944236578603,
+                        0.18339267529484207,
+                        0.2786572043773582,
+                        0.3033890890341621,
+                        0.30782564735585616,
+                        0.26624818061430483,
+                        0.12544638422745383,
+                        0.1608101384189547,
+                        0.1687095486591529,
+                        0.13384106597714177,
+                        0.08919774441517907,
+                        0.07838533748537196,
+                        0.06617923921964464,
+                        0.06122622884183163,
+                        0.11227959483655524,
+                        0.1280157086416137,
+                        0.1452414371823498,
+                        0.15534712133864104,
+                        0.3110117072275599,
+                        0.34532456199659967,
+                        0.3561773650682832,
+                        0.31731939747434174
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    72
+                ]
+            },
+            "span": 0.06413597231156332,
+            "twist": {
+                "__ndarray__": [
+                    [
+                        -0.02496719304046418,
+                        -0.03216019106103158,
+                        -0.03408737451797081,
+                        -0.03126987441866044,
+                        -0.02296784989696189,
+                        -0.009658514402793021
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    6
+                ]
+            },
+            "xRef_mdo_tutorial": -0.12640436393489274,
+            "yRef_mdo_tutorial": 0.0035902070779581328,
+            "zRef_mdo_tutorial": 0.0
+        },
+        "mdo_tutorial_drag": {
+            "P_mdo_tutorial": 0.5056157462992771,
+            "T_mdo_tutorial": 0.023250847089342996,
+            "alpha_mdo_tutorial": 2064.731321589265,
+            "beta_mdo_tutorial": -360.59524986194486,
+            "mach_mdo_tutorial": 62149.935067818835,
+            "shape": {
+                "__ndarray__": [
+                    [
+                        594.7464806710151,
+                        -60.95713879814559,
+                        -6390.586988773744,
+                        -5738.498500250781,
+                        549.3524963677182,
+                        745.067127970098,
+                        -5122.079626035299,
+                        -4267.170953733126,
+                        533.8176208049622,
+                        -1327.3003517970392,
+                        928.624965889924,
+                        2726.851390291158,
+                        -738.4885434383391,
+                        2437.055428702888,
+                        4009.6667574251287,
+                        2909.330702024834,
+                        209.90156941660013,
+                        -802.2538545856731,
+                        889.6386766972295,
+                        1805.4120606066551,
+                        1772.5789854569516,
+                        1484.8768710738282,
+                        1300.952907796529,
+                        1637.1153395910706,
+                        832.4773637868072,
+                        883.298773249864,
+                        811.077027167412,
+                        664.6733995320438,
+                        -8014.1704516117015,
+                        -8491.92622065088,
+                        -8473.460628868917,
+                        -7698.956856042612,
+                        -150.63322590781638,
+                        -127.63011317048972,
+                        -124.65096549433201,
+                        109.58297296828539,
+                        -7304.935529201246,
+                        -7691.184568347064,
+                        -7579.521265552777,
+                        -6722.1286071665745,
+                        372.07238101978254,
+                        331.0836645378183,
+                        362.8797686476965,
+                        424.22915186639136,
+                        -1657.0814825344248,
+                        -1547.6857350076039,
+                        -1306.279222728841,
+                        -1071.9291386528516,
+                        1358.8687661805395,
+                        1423.6783902780644,
+                        1368.7949545640595,
+                        1227.4224378809442,
+                        3033.248420860041,
+                        2984.0789269947195,
+                        2819.421868077051,
+                        2512.225334830884,
+                        171.315937401259,
+                        1276.6698148115597,
+                        2194.795179365885,
+                        2782.3916124800135,
+                        4229.93897952753,
+                        4887.252486432245,
+                        4871.91388360555,
+                        3845.438700605667,
+                        4586.72713607695,
+                        4017.633885710158,
+                        3265.2242704154896,
+                        2228.1262788355048,
+                        2518.507985262193,
+                        2057.5779453657105,
+                        1771.1980922979335,
+                        1650.6947969283333
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    72
+                ]
+            },
+            "span": 936.2978351023139,
+            "twist": {
+                "__ndarray__": [
+                    [
+                        -370.378325804621,
+                        -462.8299248441359,
+                        -453.758595158707,
+                        -387.83865358942626,
+                        -279.9948204628006,
+                        -128.4621028827006
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    6
+                ]
+            },
+            "xRef_mdo_tutorial": 0.0,
+            "yRef_mdo_tutorial": 0.0,
+            "zRef_mdo_tutorial": 0.0
+        },
+        "mdo_tutorial_lift": {
+            "P_mdo_tutorial": 8.270592846885688,
+            "T_mdo_tutorial": -0.053951548436925734,
+            "alpha_mdo_tutorial": 47833.491771372086,
+            "beta_mdo_tutorial": -2208.5023506349153,
+            "mach_mdo_tutorial": 600018.4073654441,
+            "shape": {
+                "__ndarray__": [
+                    [
+                        270.81519207461315,
+                        -3980.5776785900707,
+                        -154978.81817142142,
+                        -129275.2638407544,
+                        5148.8245717165655,
+                        9439.414397472698,
+                        -81307.73624698544,
+                        -68975.89095216605,
+                        14329.886244609828,
+                        27011.72030062839,
+                        37183.66671473962,
+                        65062.05521098424,
+                        31963.699846357027,
+                        24712.349012509727,
+                        19209.195202824365,
+                        65987.51735977479,
+                        11362.650336749326,
+                        13264.479555633185,
+                        17030.006462282712,
+                        23876.925541483586,
+                        12761.302236837797,
+                        9576.676181534183,
+                        19026.26916735209,
+                        30714.588731374057,
+                        3280.6130392607665,
+                        5519.504882733236,
+                        6132.801727450802,
+                        6406.884722471794,
+                        -189102.45510289547,
+                        -193236.6634248001,
+                        -182736.9743303752,
+                        -150393.17985961973,
+                        3554.2760515851987,
+                        12046.94893063933,
+                        17581.74578244701,
+                        18077.176627530247,
+                        -161294.31710868227,
+                        -166311.47325681994,
+                        -157447.92219289733,
+                        -128975.96490845073,
+                        26086.575850811285,
+                        30492.145541157413,
+                        29704.17391203255,
+                        23833.53773612961,
+                        35617.57927904792,
+                        35456.931271659865,
+                        32649.656951211997,
+                        26200.53471715887,
+                        44585.74649855906,
+                        44151.67504533622,
+                        41410.1394539392,
+                        33654.95133195592,
+                        71097.27668361232,
+                        68992.44859285219,
+                        63021.147617856484,
+                        49063.65852638495,
+                        48763.224842896896,
+                        51875.78411774403,
+                        46226.694685412665,
+                        31679.581170257843,
+                        23119.417962151383,
+                        15640.40340266184,
+                        9975.859233811978,
+                        8143.885671231952,
+                        20117.905367559186,
+                        20839.647902794906,
+                        22955.2181871585,
+                        24665.276761774807,
+                        74753.89964294108,
+                        74998.52857853827,
+                        70508.59520531689,
+                        57203.14130147164
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    72
+                ]
+            },
+            "span": 17716.820500284084,
+            "twist": {
+                "__ndarray__": [
+                    [
+                        -10051.079734920826,
+                        -11721.139103706326,
+                        -10560.626720636898,
+                        -8218.19895987124,
+                        -5187.624072678598,
+                        -1963.439343276776
+                    ]
+                ],
+                "dtype": "float64",
+                "shape": [
+                    1,
+                    6
+                ]
+            },
+            "xRef_mdo_tutorial": 0.0,
+            "yRef_mdo_tutorial": 0.0,
+            "zRef_mdo_tutorial": 0.0
+        }
+    },
+    "Norm of residual": 3.237191827215383e-14,
+    "metadata": {
+        "ADPC": false,
+        "ANKADPC": false,
+        "ANKAMGLevels": 2,
+        "ANKAMGNSmooth": 1,
+        "ANKASMOverlap": 1,
+        "ANKASMOverlapCoarse": 0,
+        "ANKCFL0": 5.0,
+        "ANKCFLCutback": 0.5,
+        "ANKCFLExponent": 0.5,
+        "ANKCFLFactor": 10.0,
+        "ANKCFLLimit": 100000.0,
+        "ANKCFLMin": 1.0,
+        "ANKCFLReset": true,
+        "ANKCharTimeStepType": "None",
+        "ANKConstCFLStep": 0.4,
+        "ANKCoupledSwitchTol": 1e-16,
+        "ANKGlobalPreconditioner": "additive Schwarz",
+        "ANKInnerPreconIts": 1,
+        "ANKInnerPreconItsCoarse": 1,
+        "ANKJacobianLag": 10,
+        "ANKLinResMax": 0.1,
+        "ANKLinearSolveBuffer": 0.01,
+        "ANKLinearSolveTol": 0.05,
+        "ANKMaxIter": 40,
+        "ANKNSubiterTurb": 1,
+        "ANKOuterPreconIts": 1,
+        "ANKPCILUFill": 2,
+        "ANKPCILUFillCoarse": 0,
+        "ANKPCUpdateCutoff": 1e-16,
+        "ANKPCUpdateTol": 0.5,
+        "ANKPCUpdateTolAfterCutoff": 0.0001,
+        "ANKPhysicalLSTol": 0.2,
+        "ANKPhysicalLSTolTurb": 0.99,
+        "ANKSecondOrdSwitchTol": 0.01,
+        "ANKStepFactor": 1.0,
+        "ANKStepMin": 0.01,
+        "ANKSubspaceSize": -1,
+        "ANKSwitchTol": 0.01,
+        "ANKTurbCFLScale": 1.0,
+        "ANKTurbKSPDebug": false,
+        "ANKUnsteadyLSTol": 1.0,
+        "ANKUseApproxSA": false,
+        "ANKUseFullVisc": true,
+        "ANKUseMatrixFree": true,
+        "ANKUseTurbDADI": true,
+        "ASMOverlap": 1,
+        "ASMOverlapCoarse": 0,
+        "CFL": 1.5,
+        "CFLCoarse": 1.25,
+        "CFLLimit": 1.5,
+        "GMRESOrthogonalizationType": "modified Gram-Schmidt",
+        "ILUFill": 2,
+        "ILUFillCoarse": 0,
+        "L2Convergence": 1e-15,
+        "L2ConvergenceCoarse": 0.01,
+        "L2ConvergenceRel": 1e-16,
+        "MGCycle": "2w",
+        "MGStartLevel": -1,
+        "NKADPC": false,
+        "NKAMGLevels": 2,
+        "NKAMGNSmooth": 1,
+        "NKASMOverlap": 1,
+        "NKASMOverlapCoarse": 0,
+        "NKFixedStep": 0.25,
+        "NKGlobalPreconditioner": "additive Schwarz",
+        "NKInnerPreconIts": 1,
+        "NKInnerPreconItsCoarse": 1,
+        "NKJacobianLag": 2,
+        "NKLS": "cubic",
+        "NKLinearSolveTol": 1e-06,
+        "NKOuterPreconIts": 1,
+        "NKPCILUFill": 2,
+        "NKPCILUFillCoarse": 0,
+        "NKSubspaceSize": 60,
+        "NKSwitchTol": 1e-05,
+        "NKUseEW": false,
+        "NKViscPC": false,
+        "RKReset": false,
+        "TSStability": false,
+        "acousticScaleFactor": 1.0,
+        "adjointAMGLevels": 2,
+        "adjointAMGNSmooth": 1,
+        "adjointDivTol": 100000.0,
+        "adjointL2Convergence": 1e-16,
+        "adjointL2ConvergenceAbs": 1e-16,
+        "adjointL2ConvergenceRel": 1e-16,
+        "adjointMaxIter": 500,
+        "adjointMaxL2DeviationFactor": 1.0,
+        "adjointMonitorStep": 10,
+        "adjointSolver": "GMRES",
+        "adjointSubspaceSize": 100,
+        "alphaFollowing": true,
+        "alphaMode": false,
+        "altitudeMode": false,
+        "applyAdjointPCSubspaceSize": 20,
+        "applyPCSubspaceSize": 10,
+        "approxPC": true,
+        "backgroundVolScale": 1.0,
+        "betaMode": false,
+        "blockSplitting": true,
+        "cavExponent": 0,
+        "cavSensorOffset": 0.0,
+        "cavSensorSharpness": 10.0,
+        "cavitationNumber": 1.4,
+        "closedSurfaceFamilies": null,
+        "coarseDiscretization": "central plus scalar dissipation",
+        "computeCavitation": false,
+        "coupledSolution": false,
+        "cpMinRho": 100.0,
+        "cutCallback": null,
+        "debugZipper": false,
+        "deltaT": 0.01,
+        "designSurfaceFamily": null,
+        "discretization": "central plus scalar dissipation",
+        "dissContMagnitude": 1.0,
+        "dissContMidpoint": 3.0,
+        "dissContSharpness": 3.0,
+        "dissipationLumpingParameter": 6.0,
+        "dissipationScalingExponent": 0.67,
+        "eddyVisInfRatio": 0.009,
+        "equationMode": "steady",
+        "equationType": "RANS",
+        "eulerWallTreatment": "linear pressure extrapolation",
+        "explicitSurfaceCallback": null,
+        "firstRun": true,
+        "flowType": "external",
+        "forcesAsTractions": true,
+        "frozenTurbulence": false,
+        "globalPreconditioner": "additive Schwarz",
+        "gridFile": "input_files/mdo_tutorial_rough.cgns",
+        "gridPrecision": "double",
+        "gridPrecisionSurface": "single",
+        "infChangeCorrection": true,
+        "infChangeCorrectionTol": 1e-12,
+        "infChangeCorrectionType": "offset",
+        "innerPreconIts": 1,
+        "innerPreconItsCoarse": 1,
+        "isoVariables": [],
+        "isosurface": {},
+        "liftIndex": 2,
+        "limiter": "van Albada",
+        "loadBalanceIter": 10,
+        "loadImbalance": 0.1,
+        "localPreconditioner": "ILU",
+        "lowSpeedPreconditioner": false,
+        "machMode": false,
+        "matrixOrdering": "RCM",
+        "maxL2DeviationFactor": 1.0,
+        "meshMaxSkewness": 1.0,
+        "meshSurfaceFamily": null,
+        "monitorVariables": [
+            "resrho",
+            "resturb",
+            "cd"
+        ],
+        "nCycles": 1000,
+        "nCyclesCoarse": 100,
+        "nFloodIter": -1,
+        "nRKReset": 5,
+        "nRefine": 10,
+        "nSaveSurface": 1,
+        "nSaveVolume": 1,
+        "nSubiter": 3,
+        "nSubiterTurb": 3,
+        "nTimeStepsCoarse": 48,
+        "nTimeStepsFine": 400,
+        "nearWallDist": 0.1,
+        "numberSolutions": true,
+        "outerPreconIts": 3,
+        "outputDirectory": "tests/output_files",
+        "outputSurfaceFamily": "allSurfaces",
+        "overlapFactor": 0.9,
+        "oversetDebugPrint": false,
+        "oversetLoadBalance": true,
+        "oversetPriority": {},
+        "oversetProjTol": 1e-12,
+        "oversetUpdateMode": "frozen",
+        "pMode": false,
+        "partitionLikeNProc": -1,
+        "partitionOnly": false,
+        "preconditionerSide": "right",
+        "printAllOptions": true,
+        "printBadlySkewedCells": false,
+        "printIntro": true,
+        "printIterations": true,
+        "printNegativeVolumes": false,
+        "printTiming": true,
+        "printWarnings": true,
+        "qMode": false,
+        "rMode": false,
+        "recomputeOverlapMatrix": true,
+        "resAveraging": "never",
+        "restartAdjoint": true,
+        "restartFile": "input_files/mdo_tutorial_rough.cgns",
+        "restrictionRelaxation": 0.8,
+        "selfZipCutoff": 120.0,
+        "sepSensorOffset": 0.0,
+        "sepSensorSharpness": 10.0,
+        "setMonitor": true,
+        "skipAfterFailedAdjoint": true,
+        "smoothParameter": 1.5,
+        "smoother": "DADI",
+        "solutionPrecision": "single",
+        "solutionPrecisionSurface": "single",
+        "storeConvHist": true,
+        "storeRindLayer": true,
+        "surfaceVariables": [
+            "cp",
+            "vx",
+            "vy",
+            "vz",
+            "mach"
+        ],
+        "timeAccuracy": 2,
+        "timeIntegrationScheme": "BDF",
+        "timeIntervals": 1,
+        "timeLimit": -1.0,
+        "turbResScale": 10000.0,
+        "turbulenceModel": "SA",
+        "turbulenceOrder": "first order",
+        "turbulenceProduction": "strain",
+        "updateWallAssociations": false,
+        "useALE": true,
+        "useANKSolver": true,
+        "useApproxWallDistance": true,
+        "useBlockettes": false,
+        "useDiagTSPC": true,
+        "useDissContinuation": false,
+        "useExternalDynamicMesh": false,
+        "useGridMotion": false,
+        "useLinResMonitor": false,
+        "useMatrixFreedrdw": true,
+        "useNKSolver": true,
+        "useOversetWallScaling": false,
+        "useQCR": false,
+        "useRotationSA": false,
+        "useRoughSA": true,
+        "useSkewnessCheck": false,
+        "useTSInterpolatedGridVelocity": false,
+        "useWallFunctions": false,
+        "useZipperMesh": true,
+        "useft2SA": true,
+        "verifyExtra": true,
+        "verifySpatial": true,
+        "verifyState": true,
+        "vis2": 0.25,
+        "vis2Coarse": 0.5,
+        "vis4": 0.0156,
+        "viscPC": false,
+        "viscWallTreatment": "constant pressure extrapolation",
+        "viscousSurfaceVelocities": true,
+        "volumeVariables": [
+            "resrho"
+        ],
+        "wallDistCutoff": 1e+20,
+        "windAxis": false,
+        "writeSolutionDigits": 3,
+        "writeSolutionEachIter": false,
+        "writeSurfaceSolution": true,
+        "writeTecplotSurfaceSolution": false,
+        "writeVolumeSolution": true,
+        "zipperSurfaceFamily": null
+    }
+}
\ No newline at end of file
diff --git a/tests/reg_tests/refs/funcs_rans_rough_sa.json b/tests/reg_tests/refs/funcs_rans_rough_sa.json
new file mode 100644
index 000000000..895d67e8b
--- /dev/null
+++ b/tests/reg_tests/refs/funcs_rans_rough_sa.json
@@ -0,0 +1,1101 @@
+{
+    "Dot product test for (w, xV) -> (dw, F)": 25051748893.868687,
+    "Dot product test for Xv -> R": -8183826.0616009245,
+    "Dot product test for w -> F": 48678.90552195639,
+    "Dot product test for w -> R": 25061252292.450912,
+    "Dot product test for xV -> F": -1368251.4261434944,
+    "Eval Functions:": {
+        "mdo_tutorial_cd": 0.024566400785789796,
+        "mdo_tutorial_cfx": 0.011668173003364507,
+        "mdo_tutorial_cfy": 0.4108141827884018,
+        "mdo_tutorial_cfz": 0.0075654034410337035,
+        "mdo_tutorial_cl": 0.4102449646172038,
+        "mdo_tutorial_cmx": -0.7922014856117325,
+        "mdo_tutorial_cmy": -0.008981003314113325,
+        "mdo_tutorial_cmz": 0.6295255266022379,
+        "mdo_tutorial_cofxx": 3.332310132439291,
+        "mdo_tutorial_cofxy": -0.1348493497186727,
+        "mdo_tutorial_cofxz": 1.6012560760241168,
+        "mdo_tutorial_cofyx": 4.97642146149654,
+        "mdo_tutorial_cofyy": 0.2601267269394276,
+        "mdo_tutorial_cofyz": 6.271918158078769,
+        "mdo_tutorial_cofzx": 6.327751594707444,
+        "mdo_tutorial_cofzy": 0.25617991212816216,
+        "mdo_tutorial_cofzz": 10.43278001294453,
+        "mdo_tutorial_colx": 4.977890283785735,
+        "mdo_tutorial_coly": 0.2604795921346493,
+        "mdo_tutorial_colz": 6.276090851527301,
+        "mdo_tutorial_drag": 10015.230272350786,
+        "mdo_tutorial_fx": 4756.880770011643,
+        "mdo_tutorial_fy": 167480.72603917567,
+        "mdo_tutorial_fz": 3084.2636748406208,
+        "mdo_tutorial_lift": 167248.6671751417,
+        "mdo_tutorial_mx": -1049635.2803761212,
+        "mdo_tutorial_my": -11899.470151067593,
+        "mdo_tutorial_mz": 834096.1417269012,
+        "mdo_tutorial_sepsensor": 0.016301194542570726,
+        "mdo_tutorial_sepsensoravgx": 0.09540342694812158,
+        "mdo_tutorial_sepsensoravgy": 9.994583497283502e-05,
+        "mdo_tutorial_sepsensoravgz": 0.05019620000103362
+    },
+    "Norm of residual": 2.2925508532857043e-16,
+    "Norm of state vector": 472.57778771100806,
+    "Sum of Forces x": 4756.880770011646,
+    "Sum of Forces y": 167480.72603917564,
+    "Sum of Forces z": 3084.263674840624,
+    "Sum of Tractions x": 336771.1069549067,
+    "Sum of Tractions y": 1650799.5101746526,
+    "Sum of Tractions z": 279940.1407894979,
+    "Total number of adjoint state DOF": {
+        "__ndarray__": 145152,
+        "dtype": "int64",
+        "shape": []
+    },
+    "Total number of spatial DOF": {
+        "__ndarray__": 91125,
+        "dtype": "int64",
+        "shape": []
+    },
+    "Total number of state DOF": {
+        "__ndarray__": 145152,
+        "dtype": "int64",
+        "shape": []
+    },
+    "dFuncs/dP": {
+        "cd": 1.025586307836628e-07,
+        "cfx": 1.0254457111038074e-07,
+        "cfy": 2.0585057746026076e-09,
+        "cfz": 3.080957025224556e-09,
+        "cl": -1.1635127916521157e-09,
+        "cmx": -5.02580368870114e-09,
+        "cmy": 1.836634687981284e-07,
+        "cmz": 2.0700897671058624e-09,
+        "cofxx": 1.037501293959403e-05,
+        "cofxy": 1.4128496198936057e-06,
+        "cofxz": 3.848381803761694e-05,
+        "cofyx": -2.090715333502599e-09,
+        "cofyy": 4.5109549389085536e-12,
+        "cofyz": 7.602063203362935e-09,
+        "cofzx": -4.184628224366399e-07,
+        "cofzy": -1.4398784783816656e-07,
+        "cofzz": -2.272928808272515e-06,
+        "colx": 1.551312768394871e-09,
+        "coly": 1.8444193827696344e-09,
+        "colz": 9.911171655450564e-09,
+        "drag": 0.5425726162154229,
+        "fx": 0.2796494092508621,
+        "fy": 8.374875513592974,
+        "fz": 0.1554692283020746,
+        "lift": 8.361959017862183,
+        "mx": -52.48842300766143,
+        "my": -0.35162675793461173,
+        "mz": 41.70754987248287,
+        "sepsensor": 0.0,
+        "sepsensoravgx": 0.0,
+        "sepsensoravgy": 0.0,
+        "sepsensoravgz": 0.0
+    },
+    "dFuncs/dT": {
+        "cd": -7.97407560708349e-21,
+        "cfx": -7.972667277561772e-21,
+        "cfy": -1.7008056169588572e-22,
+        "cfz": -2.2482267665849947e-22,
+        "cl": 8.043089385984346e-23,
+        "cmx": 4.087206383658274e-22,
+        "cmy": -1.4671395918848986e-20,
+        "cmz": -1.4513050628161639e-22,
+        "cofxx": -8.46577777879809e-19,
+        "cofxy": -1.1621669191703017e-19,
+        "cofxz": -3.0893539917499644e-18,
+        "cofyx": 2.2830502610902396e-22,
+        "cofyy": 4.3709209984172894e-24,
+        "cofyz": -5.698972668736165e-22,
+        "cofzx": 3.849894121230114e-20,
+        "cofzy": 1.1246443854871941e-20,
+        "cofzz": 1.811270017266413e-19,
+        "colx": -1.9081860443013808e-23,
+        "coly": -1.329751358440995e-22,
+        "colz": -6.623775130485751e-22,
+        "drag": -3.250871143495798e-15,
+        "fx": -3.250296995716384e-15,
+        "fy": -6.93384433921787e-17,
+        "fz": -9.165570882013707e-17,
+        "lift": 3.2790066808780993e-17,
+        "mx": 5.415384970091867e-16,
+        "my": -1.9439012736638154e-14,
+        "mz": -1.9229211560289045e-16,
+        "sepsensor": 0.0,
+        "sepsensoravgx": 0.0,
+        "sepsensoravgy": 0.0,
+        "sepsensoravgz": 0.0
+    },
+    "dFuncs/dXv * xVDot": {
+        "cd": 4.2529682732075225,
+        "cfx": 4.249035399663727,
+        "cfy": 0.19195702983176197,
+        "cfz": 0.3506482601505133,
+        "cl": 0.05839688337189065,
+        "cmx": -1.0927832721306239,
+        "cmy": -1.1999451606004583,
+        "cmz": 1.3595113636997622,
+        "cofxx": -488.09187989891825,
+        "cofxy": -76.51059615419034,
+        "cofxz": -739.7568018827052,
+        "cofyx": 4.862135728695632,
+        "cofyy": 0.37025124340482335,
+        "cofyz": 5.697393456600473,
+        "cofzx": -19.405430308047734,
+        "cofzy": -12.803792657294428,
+        "cofzz": -140.92641161911095,
+        "colx": 5.837204211267569,
+        "coly": 0.567383095306283,
+        "colz": 7.882287552102762,
+        "drag": 1733850.1056212431,
+        "fx": 1732246.7517349087,
+        "fy": 78257.04192181274,
+        "fz": 142952.2826981613,
+        "lift": 23807.241413052383,
+        "mx": -1447894.1242421914,
+        "my": -1589879.3399891835,
+        "mz": 1801298.1764476371,
+        "sepsensor": -0.11694957310130713,
+        "sepsensoravgx": -0.6029863600829567,
+        "sepsensoravgy": 0.007262037328436888,
+        "sepsensoravgz": -0.09612647922522868
+    },
+    "dFuncs/dalpha": {
+        "cd": 0.007160125372297845,
+        "cfx": 0.0,
+        "cfy": 0.0,
+        "cfz": 0.0,
+        "cl": -0.00042876457907655406,
+        "cmx": 0.0,
+        "cmy": 0.0,
+        "cmz": 0.0,
+        "cofxx": 0.0,
+        "cofxy": 0.0,
+        "cofxz": 0.0,
+        "cofyx": 0.0,
+        "cofyy": 0.0,
+        "cofyz": 0.0,
+        "cofzx": 0.0,
+        "cofzy": 0.0,
+        "cofzz": 0.0,
+        "colx": 0.0008172790369756601,
+        "coly": 0.0001963405165176115,
+        "colz": 0.0023217613923074195,
+        "drag": 2919.039911778386,
+        "fx": 0.0,
+        "fy": 0.0,
+        "fz": 0.0,
+        "lift": -174.79874359792956,
+        "mx": 0.0,
+        "my": 0.0,
+        "mz": 0.0,
+        "sepsensor": 0.0008890133278288019,
+        "sepsensoravgx": 0.006211849207427457,
+        "sepsensoravgy": 1.1629154598405563e-06,
+        "sepsensoravgz": 0.006150040814983931
+    },
+    "dFuncs/dbeta": {
+        "cd": -0.007561670371126668,
+        "cfx": 0.0,
+        "cfy": 0.0,
+        "cfz": 0.0,
+        "cl": 0.00023763506481515244,
+        "cmx": 0.0,
+        "cmy": 0.0,
+        "cmz": 0.0,
+        "cofxx": 0.0,
+        "cofxy": 0.0,
+        "cofxz": 0.0,
+        "cofyx": 0.0,
+        "cofyy": 0.0,
+        "cofyz": 0.0,
+        "cofzx": 0.0,
+        "cofzy": 0.0,
+        "cofzz": 0.0,
+        "colx": 0.0007819093658142975,
+        "coly": -2.490596656023271e-06,
+        "colz": 0.0024077689758154846,
+        "drag": -3082.7417769009207,
+        "fx": 0.0,
+        "fy": 0.0,
+        "fz": 0.0,
+        "lift": 96.87906322384137,
+        "mx": 0.0,
+        "my": 0.0,
+        "mz": 0.0,
+        "sepsensor": 0.181360493959623,
+        "sepsensoravgx": 1.1060754915788906,
+        "sepsensoravgy": 0.001164630699853899,
+        "sepsensoravgz": 0.7192972767550523
+    },
+    "dFuncs/dmach": {
+        "cd": -0.06141600196447446,
+        "cfx": -0.029170432508411257,
+        "cfy": -1.0270354569710043,
+        "cfz": -0.01891350860258426,
+        "cl": -1.0256124115430092,
+        "cmx": 1.9805037140293307,
+        "cmy": 0.02245250828528332,
+        "cmz": -1.5738138165055942,
+        "cofxx": 0.0,
+        "cofxy": 0.0,
+        "cofxz": 0.0,
+        "cofyx": 0.0,
+        "cofyy": 0.0,
+        "cofyz": 0.0,
+        "cofzx": 0.0,
+        "cofzy": 0.0,
+        "cofzz": 0.0,
+        "colx": 0.0,
+        "coly": 0.0,
+        "colz": 0.0,
+        "drag": 0.0,
+        "fx": 0.0,
+        "fy": 0.0,
+        "fz": 0.0,
+        "lift": 0.0,
+        "mx": 0.0,
+        "my": 0.0,
+        "mz": 0.0,
+        "sepsensor": 0.0,
+        "sepsensoravgx": 0.0,
+        "sepsensoravgy": 0.0,
+        "sepsensoravgz": 0.0
+    },
+    "dFuncs/dw * wDot": {
+        "cd": 0.031101767260431196,
+        "cfx": 0.028800395325415777,
+        "cfy": 0.07371942797663703,
+        "cfz": 0.021716000987843997,
+        "cl": 0.07277840961013626,
+        "cmx": -0.15842697054164165,
+        "cmy": 0.0050247716470549885,
+        "cmz": 0.12014382033187077,
+        "cofxx": -5.883673815216708,
+        "cofxy": 0.4009727991396247,
+        "cofxz": 10.11585037814496,
+        "cofyx": 0.05940247196308444,
+        "cofyy": -0.3350265708291385,
+        "cofyz": 0.127914595757047,
+        "cofzx": 1.3755158785313275,
+        "cofzy": -0.7322158660616279,
+        "cofzz": -15.884837396918154,
+        "colx": 0.06807682146337765,
+        "coly": -0.33487572734605553,
+        "colz": 0.12855069447564793,
+        "drag": 12679.568476732591,
+        "fx": 11741.345166265506,
+        "fy": 30053.936397515386,
+        "fz": 8853.179282724243,
+        "lift": 29670.30202986036,
+        "mx": -209909.39888885352,
+        "my": 6657.621441481978,
+        "mz": 159185.7561869155,
+        "sepsensor": -12.044419253289893,
+        "sepsensoravgx": -68.10234813562751,
+        "sepsensoravgy": -0.08084780944322405,
+        "sepsensoravgz": -28.399498948465336
+    },
+    "dFuncs/dxRef": {
+        "cd": 0.0,
+        "cfx": 0.0,
+        "cfy": 0.0,
+        "cfz": 0.0,
+        "cl": 0.0,
+        "cmx": 0.0,
+        "cmy": 0.0023278164433949864,
+        "cmz": -0.12640436393489288,
+        "cofxx": 0.0,
+        "cofxy": 0.0,
+        "cofxz": 0.0,
+        "cofyx": 0.0,
+        "cofyy": 0.0,
+        "cofyz": 0.0,
+        "cofzx": 0.0,
+        "cofzy": 0.0,
+        "cofzz": 0.0,
+        "colx": 0.0,
+        "coly": 0.0,
+        "colz": 0.0,
+        "drag": 0.0,
+        "fx": 0.0,
+        "fy": 0.0,
+        "fz": 0.0,
+        "lift": 0.0,
+        "mx": 0.0,
+        "my": 3084.2636748406217,
+        "mz": -167480.72603917567,
+        "sepsensor": 0.0,
+        "sepsensoravgx": 0.0,
+        "sepsensoravgy": 0.0,
+        "sepsensoravgz": 0.0
+    },
+    "dFuncs/dyRef": {
+        "cd": 0.0,
+        "cfx": 0.0,
+        "cfy": 0.0,
+        "cfz": 0.0,
+        "cl": 0.0,
+        "cmx": -0.0023278164433949864,
+        "cmy": 0.0,
+        "cmz": 0.003590207077958309,
+        "cofxx": 0.0,
+        "cofxy": 0.0,
+        "cofxz": 0.0,
+        "cofyx": 0.0,
+        "cofyy": 0.0,
+        "cofyz": 0.0,
+        "cofzx": 0.0,
+        "cofzy": 0.0,
+        "cofzz": 0.0,
+        "colx": 0.0,
+        "coly": 0.0,
+        "colz": 0.0,
+        "drag": 0.0,
+        "fx": 0.0,
+        "fy": 0.0,
+        "fz": 0.0,
+        "lift": 0.0,
+        "mx": -3084.2636748406217,
+        "my": 0.0,
+        "mz": 4756.880770011641,
+        "sepsensor": 0.0,
+        "sepsensoravgx": 0.0,
+        "sepsensoravgy": 0.0,
+        "sepsensoravgz": 0.0
+    },
+    "dFuncs/dzRef": {
+        "cd": 0.0,
+        "cfx": 0.0,
+        "cfy": 0.0,
+        "cfz": 0.0,
+        "cl": 0.0,
+        "cmx": 0.12640436393489285,
+        "cmy": -0.003590207077958309,
+        "cmz": 0.0,
+        "cofxx": 0.0,
+        "cofxy": 0.0,
+        "cofxz": 0.0,
+        "cofyx": 0.0,
+        "cofyy": 0.0,
+        "cofyz": 0.0,
+        "cofzx": 0.0,
+        "cofzy": 0.0,
+        "cofzz": 0.0,
+        "colx": 0.0,
+        "coly": 0.0,
+        "colz": 0.0,
+        "drag": 0.0,
+        "fx": 0.0,
+        "fy": 0.0,
+        "fz": 0.0,
+        "lift": 0.0,
+        "mx": 167480.72603917564,
+        "my": -4756.880770011641,
+        "mz": 0.0,
+        "sepsensor": 0.0,
+        "sepsensoravgx": 0.0,
+        "sepsensoravgy": 0.0,
+        "sepsensoravgz": 0.0
+    },
+    "metadata": {
+        "ADPC": false,
+        "ANKADPC": false,
+        "ANKAMGLevels": 2,
+        "ANKAMGNSmooth": 1,
+        "ANKASMOverlap": 1,
+        "ANKASMOverlapCoarse": 0,
+        "ANKCFL0": 5.0,
+        "ANKCFLCutback": 0.5,
+        "ANKCFLExponent": 0.5,
+        "ANKCFLFactor": 10.0,
+        "ANKCFLLimit": 100000.0,
+        "ANKCFLMin": 1.0,
+        "ANKCFLReset": true,
+        "ANKCharTimeStepType": "None",
+        "ANKConstCFLStep": 0.4,
+        "ANKCoupledSwitchTol": 1e-16,
+        "ANKGlobalPreconditioner": "additive Schwarz",
+        "ANKInnerPreconIts": 1,
+        "ANKInnerPreconItsCoarse": 1,
+        "ANKJacobianLag": 10,
+        "ANKLinResMax": 0.1,
+        "ANKLinearSolveBuffer": 0.01,
+        "ANKLinearSolveTol": 0.05,
+        "ANKMaxIter": 40,
+        "ANKNSubiterTurb": 1,
+        "ANKOuterPreconIts": 1,
+        "ANKPCILUFill": 2,
+        "ANKPCILUFillCoarse": 0,
+        "ANKPCUpdateCutoff": 1e-16,
+        "ANKPCUpdateTol": 0.5,
+        "ANKPCUpdateTolAfterCutoff": 0.0001,
+        "ANKPhysicalLSTol": 0.2,
+        "ANKPhysicalLSTolTurb": 0.99,
+        "ANKSecondOrdSwitchTol": 0.01,
+        "ANKStepFactor": 1.0,
+        "ANKStepMin": 0.01,
+        "ANKSubspaceSize": -1,
+        "ANKSwitchTol": 0.01,
+        "ANKTurbCFLScale": 1.0,
+        "ANKTurbKSPDebug": false,
+        "ANKUnsteadyLSTol": 1.0,
+        "ANKUseApproxSA": false,
+        "ANKUseFullVisc": true,
+        "ANKUseMatrixFree": true,
+        "ANKUseTurbDADI": true,
+        "ASMOverlap": 1,
+        "ASMOverlapCoarse": 0,
+        "CFL": 1.5,
+        "CFLCoarse": 1.25,
+        "CFLLimit": 1.5,
+        "GMRESOrthogonalizationType": "modified Gram-Schmidt",
+        "ILUFill": 2,
+        "ILUFillCoarse": 0,
+        "L2Convergence": 1e-15,
+        "L2ConvergenceCoarse": 0.01,
+        "L2ConvergenceRel": 1e-16,
+        "MGCycle": "2w",
+        "MGStartLevel": -1,
+        "NKADPC": false,
+        "NKAMGLevels": 2,
+        "NKAMGNSmooth": 1,
+        "NKASMOverlap": 1,
+        "NKASMOverlapCoarse": 0,
+        "NKFixedStep": 0.25,
+        "NKGlobalPreconditioner": "additive Schwarz",
+        "NKInnerPreconIts": 1,
+        "NKInnerPreconItsCoarse": 1,
+        "NKJacobianLag": 2,
+        "NKLS": "cubic",
+        "NKLinearSolveTol": 0.3,
+        "NKOuterPreconIts": 1,
+        "NKPCILUFill": 2,
+        "NKPCILUFillCoarse": 0,
+        "NKSubspaceSize": 60,
+        "NKSwitchTol": 1e-05,
+        "NKUseEW": true,
+        "NKViscPC": false,
+        "RKReset": false,
+        "TSStability": false,
+        "acousticScaleFactor": 1.0,
+        "adjointAMGLevels": 2,
+        "adjointAMGNSmooth": 1,
+        "adjointDivTol": 100000.0,
+        "adjointL2Convergence": 1e-16,
+        "adjointL2ConvergenceAbs": 1e-16,
+        "adjointL2ConvergenceRel": 1e-16,
+        "adjointMaxIter": 500,
+        "adjointMaxL2DeviationFactor": 1.0,
+        "adjointMonitorStep": 10,
+        "adjointSolver": "GMRES",
+        "adjointSubspaceSize": 100,
+        "alphaFollowing": true,
+        "alphaMode": false,
+        "altitudeMode": false,
+        "applyAdjointPCSubspaceSize": 20,
+        "applyPCSubspaceSize": 10,
+        "approxPC": true,
+        "backgroundVolScale": 1.0,
+        "betaMode": false,
+        "blockSplitting": true,
+        "cavExponent": 0,
+        "cavSensorOffset": 0.0,
+        "cavSensorSharpness": 10.0,
+        "cavitationNumber": 1.4,
+        "closedSurfaceFamilies": null,
+        "coarseDiscretization": "central plus scalar dissipation",
+        "computeCavitation": false,
+        "coupledSolution": false,
+        "cpMinRho": 100.0,
+        "cutCallback": null,
+        "debugZipper": false,
+        "deltaT": 0.01,
+        "designSurfaceFamily": null,
+        "discretization": "central plus scalar dissipation",
+        "dissContMagnitude": 1.0,
+        "dissContMidpoint": 3.0,
+        "dissContSharpness": 3.0,
+        "dissipationLumpingParameter": 6.0,
+        "dissipationScalingExponent": 0.67,
+        "eddyVisInfRatio": 0.009,
+        "equationMode": "steady",
+        "equationType": "RANS",
+        "eulerWallTreatment": "linear pressure extrapolation",
+        "explicitSurfaceCallback": null,
+        "firstRun": true,
+        "flowType": "external",
+        "forcesAsTractions": true,
+        "frozenTurbulence": false,
+        "globalPreconditioner": "additive Schwarz",
+        "gridFile": "input_files/mdo_tutorial_rough.cgns",
+        "gridPrecision": "double",
+        "gridPrecisionSurface": "single",
+        "infChangeCorrection": true,
+        "infChangeCorrectionTol": 1e-12,
+        "infChangeCorrectionType": "offset",
+        "innerPreconIts": 1,
+        "innerPreconItsCoarse": 1,
+        "isoVariables": [],
+        "isosurface": {},
+        "liftIndex": 2,
+        "limiter": "van Albada",
+        "loadBalanceIter": 10,
+        "loadImbalance": 0.1,
+        "localPreconditioner": "ILU",
+        "lowSpeedPreconditioner": false,
+        "machMode": false,
+        "matrixOrdering": "RCM",
+        "maxL2DeviationFactor": 1.0,
+        "meshMaxSkewness": 1.0,
+        "meshSurfaceFamily": null,
+        "monitorVariables": [
+            "cpu",
+            "resrho",
+            "resturb",
+            "cl",
+            "cd",
+            "cmz",
+            "yplus",
+            "totalr"
+        ],
+        "nCycles": 1000,
+        "nCyclesCoarse": 100,
+        "nFloodIter": -1,
+        "nRKReset": 5,
+        "nRefine": 10,
+        "nSaveSurface": 1,
+        "nSaveVolume": 1,
+        "nSubiter": 3,
+        "nSubiterTurb": 3,
+        "nTimeStepsCoarse": 48,
+        "nTimeStepsFine": 400,
+        "nearWallDist": 0.1,
+        "numberSolutions": true,
+        "outerPreconIts": 3,
+        "outputDirectory": "tests/output_files",
+        "outputSurfaceFamily": "allSurfaces",
+        "overlapFactor": 0.9,
+        "oversetDebugPrint": false,
+        "oversetLoadBalance": true,
+        "oversetPriority": {},
+        "oversetProjTol": 1e-12,
+        "oversetUpdateMode": "frozen",
+        "pMode": false,
+        "partitionLikeNProc": -1,
+        "partitionOnly": false,
+        "preconditionerSide": "right",
+        "printAllOptions": true,
+        "printBadlySkewedCells": false,
+        "printIntro": true,
+        "printIterations": true,
+        "printNegativeVolumes": false,
+        "printTiming": true,
+        "printWarnings": true,
+        "qMode": false,
+        "rMode": false,
+        "recomputeOverlapMatrix": true,
+        "resAveraging": "never",
+        "restartAdjoint": true,
+        "restartFile": "input_files/mdo_tutorial_rough.cgns",
+        "restrictionRelaxation": 0.8,
+        "selfZipCutoff": 120.0,
+        "sepSensorOffset": 0.0,
+        "sepSensorSharpness": 10.0,
+        "setMonitor": true,
+        "skipAfterFailedAdjoint": true,
+        "smoothParameter": 1.5,
+        "smoother": "DADI",
+        "solutionPrecision": "single",
+        "solutionPrecisionSurface": "single",
+        "storeConvHist": true,
+        "storeRindLayer": true,
+        "surfaceVariables": [
+            "cp",
+            "vx",
+            "vy",
+            "vz",
+            "mach"
+        ],
+        "timeAccuracy": 2,
+        "timeIntegrationScheme": "BDF",
+        "timeIntervals": 1,
+        "timeLimit": -1.0,
+        "turbResScale": 10000.0,
+        "turbulenceModel": "SA",
+        "turbulenceOrder": "first order",
+        "turbulenceProduction": "strain",
+        "updateWallAssociations": false,
+        "useALE": true,
+        "useANKSolver": true,
+        "useApproxWallDistance": true,
+        "useBlockettes": false,
+        "useDiagTSPC": true,
+        "useDissContinuation": false,
+        "useExternalDynamicMesh": false,
+        "useGridMotion": false,
+        "useLinResMonitor": false,
+        "useMatrixFreedrdw": true,
+        "useNKSolver": true,
+        "useOversetWallScaling": false,
+        "useQCR": false,
+        "useRotationSA": false,
+        "useRoughSA": true,
+        "useSkewnessCheck": false,
+        "useTSInterpolatedGridVelocity": false,
+        "useWallFunctions": false,
+        "useZipperMesh": true,
+        "useft2SA": true,
+        "verifyExtra": true,
+        "verifySpatial": true,
+        "verifyState": true,
+        "vis2": 0.25,
+        "vis2Coarse": 0.5,
+        "vis4": 0.0156,
+        "viscPC": false,
+        "viscWallTreatment": "constant pressure extrapolation",
+        "viscousSurfaceVelocities": true,
+        "volumeVariables": [
+            "resrho"
+        ],
+        "wallDistCutoff": 1e+20,
+        "windAxis": false,
+        "writeSolutionDigits": 3,
+        "writeSolutionEachIter": false,
+        "writeSurfaceSolution": true,
+        "writeTecplotSurfaceSolution": false,
+        "writeVolumeSolution": true,
+        "zipperSurfaceFamily": null
+    },
+    "||FBar^T * dF/dXv||": 47500739.33793807,
+    "||FBar^T * dF/dw||": 159520.82502357475,
+    "||FBar^T * dF/xDv||": {
+        "P_mdo_tutorial": 57.28609964209288,
+        "T_mdo_tutorial": 3.552713678800501e-14,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dF/dP||": 5.5922286389459215,
+    "||dF/dT||": 1.4357072986878935e-15,
+    "||dF/dXv * xVDot||": 7581135.361357854,
+    "||dF/dalpha||": 0.0,
+    "||dF/dbeta||": 0.0,
+    "||dF/dmach||": 0.0,
+    "||dF/dw * wDot||": 131479.67592004602,
+    "||dF/dxRef||": 0.0,
+    "||dF/dyRef||": 0.0,
+    "||dF/dzRef||": 0.0,
+    "||dR/dP||": 0.07356605182246514,
+    "||dR/dT||": 9.220630057589264e-07,
+    "||dR/dXv * xVDot||": 26726959.34145183,
+    "||dR/dalpha||": 0.010271084828244875,
+    "||dR/dbeta||": 0.6070879776942455,
+    "||dR/dmach||": 0.9283860217641269,
+    "||dR/dw * wDot||": 2695502679.787602,
+    "||dR/dxRef||": 0.0,
+    "||dR/dyRef||": 0.0,
+    "||dR/dzRef||": 0.0,
+    "||dcd/dXdv||": {
+        "P_mdo_tutorial": 1.0255863078366487e-07,
+        "T_mdo_tutorial": -1.6940658945086007e-21,
+        "alpha_mdo_tutorial": 0.007160125372297845,
+        "beta_mdo_tutorial": -0.00013197604492666198,
+        "mach_mdo_tutorial": -0.06141600196447446,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcd/dXv||": 27.91337076794424,
+    "||dcd/dw||": 0.014161257652762544,
+    "||dcfx/dXdv||": {
+        "P_mdo_tutorial": 1.0254457111038457e-07,
+        "T_mdo_tutorial": 1.8634724839594607e-20,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": -0.029170432508411247,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcfx/dXv||": 27.91271619511463,
+    "||dcfx/dw||": 0.014090362780761695,
+    "||dcfy/dXdv||": {
+        "P_mdo_tutorial": 2.058505774629507e-09,
+        "T_mdo_tutorial": 1.164670302474663e-21,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": -1.027035456971004,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcfy/dXv||": 2.7931029821306073,
+    "||dcfy/dw||": 0.11354907277670545,
+    "||dcfz/dXdv||": {
+        "P_mdo_tutorial": 3.08095702522557e-09,
+        "T_mdo_tutorial": 1.0587911840678754e-22,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": -0.018913508602584266,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcfz/dXv||": 2.519605387991788,
+    "||dcfz/dw||": 0.007655345743081827,
+    "||dcl/dXdv||": {
+        "P_mdo_tutorial": -1.163512791604249e-09,
+        "T_mdo_tutorial": 5.293955920339377e-22,
+        "alpha_mdo_tutorial": -0.000428764579076554,
+        "beta_mdo_tutorial": 4.147514299214542e-06,
+        "mach_mdo_tutorial": -1.0256124115430092,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcl/dXv||": 2.7865537908099736,
+    "||dcl/dw||": 0.11354025292131027,
+    "||dcmx/dXdv||": {
+        "P_mdo_tutorial": -5.025803688770831e-09,
+        "T_mdo_tutorial": 5.293955920339377e-22,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 1.9805037140293307,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": -0.0023278164433949877,
+        "zRef_mdo_tutorial": 0.12640436393489274
+    },
+    "||dcmx/dXv||": 5.614216937125317,
+    "||dcmx/dw||": 0.21272941416694607,
+    "||dcmy/dXdv||": {
+        "P_mdo_tutorial": 1.8366346879813508e-07,
+        "T_mdo_tutorial": -6.437450399132683e-20,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.022452508285283317,
+        "xRef_mdo_tutorial": 0.0023278164433949877,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": -0.003590207077958317
+    },
+    "||dcmy/dXv||": 54.53814156559551,
+    "||dcmy/dw||": 0.033839731077361684,
+    "||dcmz/dXdv||": {
+        "P_mdo_tutorial": 2.0700897671831215e-09,
+        "T_mdo_tutorial": 2.4980854499101436e-22,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": -1.5738138165055942,
+        "xRef_mdo_tutorial": -0.12640436393489274,
+        "yRef_mdo_tutorial": 0.003590207077958317,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcmz/dXv||": 3.6697227583365826,
+    "||dcmz/dw||": 0.1701587174303123,
+    "||dcofxx/dXdv||": {
+        "P_mdo_tutorial": 1.0375012939594107e-05,
+        "T_mdo_tutorial": -2.168404344971009e-19,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofxx/dXv||": 3932.1540349438337,
+    "||dcofxx/dw||": 2.443894963885116,
+    "||dcofxy/dXdv||": {
+        "P_mdo_tutorial": 1.4128496198935765e-06,
+        "T_mdo_tutorial": 9.063252535621014e-20,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofxy/dXv||": 623.1593133276406,
+    "||dcofxy/dw||": 0.23025338834961076,
+    "||dcofxz/dXdv||": {
+        "P_mdo_tutorial": 3.848381803761748e-05,
+        "T_mdo_tutorial": -4.336808689942018e-18,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofxz/dXv||": 11552.29885023033,
+    "||dcofxz/dw||": 5.698262110691424,
+    "||dcofyx/dXdv||": {
+        "P_mdo_tutorial": -2.0907153332327543e-09,
+        "T_mdo_tutorial": 3.8116482626443515e-21,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofyx/dXv||": 12.79568795413275,
+    "||dcofyx/dw||": 0.4192809957752443,
+    "||dcofyy/dXdv||": {
+        "P_mdo_tutorial": 4.510954947963962e-12,
+        "T_mdo_tutorial": 3.705769144237564e-22,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofyy/dXv||": 1.9032269476611576,
+    "||dcofyy/dw||": 0.0825146038902352,
+    "||dcofyz/dXdv||": {
+        "P_mdo_tutorial": 7.602063203682738e-09,
+        "T_mdo_tutorial": 1.5881867761018131e-21,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofyz/dXv||": 20.030844678339538,
+    "||dcofyz/dw||": 0.8509322500970162,
+    "||dcofzx/dXdv||": {
+        "P_mdo_tutorial": -4.184628224365662e-07,
+        "T_mdo_tutorial": 1.2197274440461925e-19,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofzx/dXv||": 827.0135715836623,
+    "||dcofzx/dw||": 2.862119939722487,
+    "||dcofzy/dXdv||": {
+        "P_mdo_tutorial": -1.4398784783813178e-07,
+        "T_mdo_tutorial": 8.470329472543003e-22,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofzy/dXv||": 83.78696877492811,
+    "||dcofzy/dw||": 0.29477884488655265,
+    "||dcofzz/dXdv||": {
+        "P_mdo_tutorial": -2.272928808272855e-06,
+        "T_mdo_tutorial": -9.486769009248164e-20,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcofzz/dXv||": 1682.3717949452966,
+    "||dcofzz/dw||": 5.947434235310889,
+    "||dcolx/dXdv||": {
+        "P_mdo_tutorial": 1.5513127686803932e-09,
+        "T_mdo_tutorial": -8.470329472543003e-22,
+        "alpha_mdo_tutorial": 0.0008172790369756611,
+        "beta_mdo_tutorial": 1.3646892885640295e-05,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcolx/dXv||": 12.775631153550293,
+    "||dcolx/dw||": 0.41986360397311606,
+    "||dcoly/dXdv||": {
+        "P_mdo_tutorial": 1.844419382777588e-09,
+        "T_mdo_tutorial": -3.970466940254533e-22,
+        "alpha_mdo_tutorial": 0.00019634051651761149,
+        "beta_mdo_tutorial": -4.346911198676798e-08,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcoly/dXv||": 2.121373186987298,
+    "||dcoly/dw||": 0.08274333483032113,
+    "||dcolz/dXdv||": {
+        "P_mdo_tutorial": 9.911171655798822e-09,
+        "T_mdo_tutorial": 4.235164736271502e-22,
+        "alpha_mdo_tutorial": 0.0023217613923074195,
+        "beta_mdo_tutorial": 4.202349625535192e-05,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dcolz/dXv||": 20.13173905846084,
+    "||dcolz/dw||": 0.852397192595701,
+    "||ddrag/dXdv||": {
+        "P_mdo_tutorial": 0.5425726162154232,
+        "T_mdo_tutorial": -3.552713678800501e-15,
+        "alpha_mdo_tutorial": 2919.0399117783863,
+        "beta_mdo_tutorial": -53.80399399570157,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||ddrag/dXv||": 11379722.99467551,
+    "||ddrag/dw||": 5773.261519878235,
+    "||dfx/dXdv||": {
+        "P_mdo_tutorial": 0.2796494092508623,
+        "T_mdo_tutorial": -2.6645352591003757e-15,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dfx/dXv||": 11379456.138424333,
+    "||dfx/dw||": 5744.359098460929,
+    "||dfy/dXdv||": {
+        "P_mdo_tutorial": 8.374875513592984,
+        "T_mdo_tutorial": 1.249000902703301e-16,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dfy/dXv||": 1138692.223755006,
+    "||dfy/dw||": 46291.685989607286,
+    "||dfz/dXdv||": {
+        "P_mdo_tutorial": 0.15546922830207505,
+        "T_mdo_tutorial": -8.326672684688674e-17,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dfz/dXv||": 1027192.7245764921,
+    "||dfz/dw||": 3120.9313525395996,
+    "||dlift/dXdv||": {
+        "P_mdo_tutorial": 8.361959017862175,
+        "T_mdo_tutorial": -8.326672684688674e-17,
+        "alpha_mdo_tutorial": -174.7987435979296,
+        "beta_mdo_tutorial": 1.6908586295037848,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dlift/dXv||": 1136022.2494374104,
+    "||dlift/dw||": 46288.09031095977,
+    "||dmx/dXdv||": {
+        "P_mdo_tutorial": -52.48842300766146,
+        "T_mdo_tutorial": -4.163336342344337e-15,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": -3084.263674840623,
+        "zRef_mdo_tutorial": 167480.72603917567
+    },
+    "||dmx/dXv||": 7438612.87301356,
+    "||dmx/dw||": 281857.96459463687,
+    "||dmy/dXdv||": {
+        "P_mdo_tutorial": -0.3516267579346164,
+        "T_mdo_tutorial": 6.750155989720952e-14,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 3084.263674840623,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": -4756.880770011646
+    },
+    "||dmy/dXv||": 72260856.04875143,
+    "||dmy/dw||": 44836.290088261136,
+    "||dmz/dXdv||": {
+        "P_mdo_tutorial": 41.70754987248288,
+        "T_mdo_tutorial": -7.181755190543981e-16,
+        "alpha_mdo_tutorial": 0.0,
+        "beta_mdo_tutorial": 0.0,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": -167480.72603917567,
+        "yRef_mdo_tutorial": 4756.880770011646,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dmz/dXv||": 4862235.86588564,
+    "||dmz/dw||": 225453.4942464666,
+    "||dsepsensor/dXdv||": {
+        "P_mdo_tutorial": 0.0,
+        "T_mdo_tutorial": 0.0,
+        "alpha_mdo_tutorial": 0.0008890133278288006,
+        "beta_mdo_tutorial": 0.00316533775263871,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dsepsensor/dXv||": 0.5244726837411959,
+    "||dsepsensor/dw||": 8.307879565539196,
+    "||dsepsensoravgx/dXdv||": {
+        "P_mdo_tutorial": 0.0,
+        "T_mdo_tutorial": 0.0,
+        "alpha_mdo_tutorial": 0.0062118492074274565,
+        "beta_mdo_tutorial": 0.01930465910366646,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dsepsensoravgx/dXv||": 3.3555033669837817,
+    "||dsepsensoravgx/dw||": 42.297878810572435,
+    "||dsepsensoravgy/dXdv||": {
+        "P_mdo_tutorial": 0.0,
+        "T_mdo_tutorial": 0.0,
+        "alpha_mdo_tutorial": 1.1629154598405582e-06,
+        "beta_mdo_tutorial": 2.0326640282256375e-05,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dsepsensoravgy/dXv||": 0.003795589801898465,
+    "||dsepsensoravgy/dw||": 0.07333646978426117,
+    "||dsepsensoravgz/dXdv||": {
+        "P_mdo_tutorial": 0.0,
+        "T_mdo_tutorial": 0.0,
+        "alpha_mdo_tutorial": 0.006150040814983934,
+        "beta_mdo_tutorial": 0.01255410578000454,
+        "mach_mdo_tutorial": 0.0,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    },
+    "||dsepsensoravgz/dXv||": 3.3049236735056238,
+    "||dsepsensoravgz/dw||": 10.598482329462021,
+    "||dwBar^T * dR/dXv||": 108834126.81700659,
+    "||dwBar^T * dR/dw||": 2217397593.2762737,
+    "||dwBar^T * dR/xDv||": {
+        "P_mdo_tutorial": 0.8893628179225483,
+        "T_mdo_tutorial": -7.760276830026669e-06,
+        "alpha_mdo_tutorial": -0.002000483758618282,
+        "beta_mdo_tutorial": -0.025156969826716998,
+        "mach_mdo_tutorial": -3.283371024149603,
+        "xRef_mdo_tutorial": 0.0,
+        "yRef_mdo_tutorial": 0.0,
+        "zRef_mdo_tutorial": 0.0
+    }
+}
\ No newline at end of file
diff --git a/tests/reg_tests/test_adjoint.py b/tests/reg_tests/test_adjoint.py
index 10eecfa2a..3b24fdb34 100644
--- a/tests/reg_tests/test_adjoint.py
+++ b/tests/reg_tests/test_adjoint.py
@@ -208,6 +208,44 @@ def span(val, geo):
         "evalFuncs": ["fy", "my"],
         "N_PROCS": 2,
     },
+    # Rough SA test
+    {
+        "name": "Rough_SA_wing",
+        "options": {
+            "gridFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+            "restartFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+            "equationType": "RANS",
+            "useBlockettes": False,
+            "useRoughSA": True,
+            "MGCycle": "2w",
+            "equationType": "RANS",
+            "smoother": "DADI",
+            "CFL": 1.5,
+            "CFLCoarse": 1.25,
+            "resAveraging": "never",
+            "nSubiter": 3,
+            "nSubiterTurb": 3,
+            "nCyclesCoarse": 100,
+            "nCycles": 1000,
+            "monitorVariables": ["resrho", "resturb", "cd"],
+            "volumeVariables": ["resrho"],
+            "useNKsolver": True,
+            "ANKSwitchTol": 1e-2,
+            "ANKSecondordSwitchTol": 1e-2,
+            "NKSwitchTol": 1e-5,
+            "NKjacobianlag": 2,
+            "L2Convergence": 1e-15,
+            "adjointL2Convergence": 1e-16,
+            # to get slightly better complex convergence
+            "NKUseEW": False,
+            "NKLinearSolveTol": 1e-6,
+        },
+        "ref_file": "adjoint_rans_rough_sa.json",
+        "aero_prob": ap_tutorial_wing,
+        "evalFuncs": ["cl", "cd", "cmz", "lift", "drag"],
+        # "evalFuncs": ["cd"],
+        "N_PROCS": 2,
+    },
 ]
 
 
diff --git a/tests/reg_tests/test_functionals.py b/tests/reg_tests/test_functionals.py
index 4631ee626..d18a6e26b 100644
--- a/tests/reg_tests/test_functionals.py
+++ b/tests/reg_tests/test_functionals.py
@@ -211,6 +211,73 @@
             "aero_prob": ap_CRM,
             "dot_prod_tol": 1e-10,
         },
+        # Rough SA test
+        {
+            "name": "Rough_SA_wing",
+            "options": {
+                "gridFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+                "restartFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+                "equationType": "RANS",
+                "useBlockettes": False,
+                "useRoughSA": True,
+                "MGCycle": "2w",
+                "equationType": "RANS",
+                "smoother": "DADI",
+                "CFL": 1.5,
+                "CFLCoarse": 1.25,
+                "resAveraging": "never",
+                "nSubiter": 3,
+                "nSubiterTurb": 3,
+                "nCyclesCoarse": 100,
+                "nCycles": 1000,
+                "monitorVariables": ["cpu", "resrho", "resturb", "cl", "cd", "cmz", "yplus", "totalr"],
+                "useNKsolver": True,
+                "ANKSwitchTol": 1e-2,
+                "ANKSecondordSwitchTol": 1e-2,
+                "L2Convergence": 1e-15,
+                "NKSwitchTol": 1e-5,
+                "adjointL2Convergence": 1e-16,
+                "blockSplitting": True,
+                "NKjacobianlag": 2,
+            },
+            "ref_file": "funcs_rans_rough_sa.json",
+            "aero_prob": ap_tutorial_wing,
+            "dot_prod_tol": 1e-10,
+        },
+        # Rough Tutorial wing RANS
+        # This test makes sure a roughness value of 0 equals the standard SA model
+        # It checks if the outcome is consistent with the test 'rans_tut_wing'
+        {
+            "name": "Rough_SA_rans_tut_wing",
+            "options": {
+                "gridfile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rans_scalar_jst.cgns"),
+                "restartfile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rans_scalar_jst.cgns"),
+                "useBlockettes": False,
+                "useRoughSA": True,
+                "mgcycle": "sg",
+                "equationtype": "RANS",
+                "smoother": "DADI",
+                "cfl": 1.5,
+                "cflcoarse": 1.25,
+                "resaveraging": "never",
+                "nsubiter": 3,
+                "nsubiterturb": 3,
+                "ncyclescoarse": 100,
+                "ncycles": 1000,
+                "monitorvariables": ["cpu", "resrho", "resturb", "cl", "cd", "cmz", "yplus", "totalr"],
+                "usenksolver": True,
+                "l2convergence": 1e-14,
+                "l2convergencecoarse": 1e-4,
+                "nkswitchtol": 1e-3,
+                "adjointl2convergence": 1e-14,
+                "frozenturbulence": False,
+            },
+            "ref_file": "funcs_rans_tut_wing.json",
+            "aero_prob": ap_tutorial_wing,
+            "no_train": True,  # This test should not be able to over-write
+            # the training file as it is coming from a different test
+            "dot_prod_tol": 1e-10,
+        },
     ]
 )
 class TestFunctionals(reg_test_classes.RegTest):
diff --git a/tests/reg_tests/test_jacVecProdBWDFast.py b/tests/reg_tests/test_jacVecProdBWDFast.py
index 67d4c4a71..f56091c57 100644
--- a/tests/reg_tests/test_jacVecProdBWDFast.py
+++ b/tests/reg_tests/test_jacVecProdBWDFast.py
@@ -33,6 +33,40 @@
         "aero_prob": copy.deepcopy(ap_tutorial_wing),
         "N_PROCS": 1,
     },
+
+    {
+        "name": "rough_sa_tut_wing_1core",
+        "options": {
+            "gridFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+            "restartFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+                "equationType": "RANS",
+                "useBlockettes": False,
+                "useRoughSA": True,
+                "MGCycle": "2w",
+                "equationType": "RANS",
+                "smoother": "DADI",
+                "CFL": 1.5,
+                "CFLCoarse": 1.25,
+                "resAveraging": "never",
+                "nSubiter": 3,
+                "nSubiterTurb": 3,
+                "nCyclesCoarse": 100,
+                "nCycles": 1000,
+                "monitorVariables": ["cpu", "resrho", "resturb", "cl", "cd", "cmz", "yplus", "totalr"],
+                "useNKsolver": True,
+                "ANKSwitchTol": 1e-2,
+                "ANKSecondordSwitchTol": 1e-2,
+                "L2Convergence": 1e-15,
+                "NKSwitchTol": 1e-5,
+                "adjointL2Convergence": 1e-16,
+                "blockSplitting": True,
+                "NKjacobianlag": 2,
+
+        },
+        "ref_file": "funcs_rans_rough_sa.json",
+        "aero_prob": copy.deepcopy(ap_tutorial_wing),
+        "N_PROCS": 1,
+    },
 ]
 
 
diff --git a/tests/reg_tests/test_jacVecProdFWD.py b/tests/reg_tests/test_jacVecProdFWD.py
index 5b8ac34a8..bf5b1b9e4 100644
--- a/tests/reg_tests/test_jacVecProdFWD.py
+++ b/tests/reg_tests/test_jacVecProdFWD.py
@@ -35,6 +35,41 @@
         "aero_prob": copy.deepcopy(ap_tutorial_wing),
         "N_PROCS": 1,
     },
+
+    {
+        "name": "rough_sa_tut_wing_1core",
+        "options": {
+            "gridFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+            "restartFile": os.path.join(baseDir, "../../input_files/mdo_tutorial_rough.cgns"),
+                "equationType": "RANS",
+                "useBlockettes": False,
+                "useRoughSA": True,
+                "MGCycle": "2w",
+                "equationType": "RANS",
+                "smoother": "DADI",
+                "CFL": 1.5,
+                "CFLCoarse": 1.25,
+                "resAveraging": "never",
+                "nSubiter": 3,
+                "nSubiterTurb": 3,
+                "nCyclesCoarse": 100,
+                "nCycles": 1000,
+                "monitorVariables": ["cpu", "resrho", "resturb", "cl", "cd", "cmz", "yplus", "totalr"],
+                "useNKsolver": True,
+                "ANKSwitchTol": 1e-2,
+                "ANKSecondordSwitchTol": 1e-2,
+                "L2Convergence": 1e-15,
+                "NKSwitchTol": 1e-5,
+                "adjointL2Convergence": 1e-16,
+                "blockSplitting": True,
+                "NKjacobianlag": 2,
+
+        },
+        "ref_file": "funcs_rans_rough_sa.json",
+        "aero_prob": copy.deepcopy(ap_tutorial_wing),
+        "N_PROCS": 1,
+    },
+
 ]