FATES-Hydro fails in very dry regions

[jennykowalcz]

I am encountering FATES-Hydro failing in dry regions, I think because it is not killing cohorts that should be dying. Here is an example error message from a run with the main branch (also posted in this discussion):

167:  Could not find a stable solution for hydro 1D solve
167:  
167:  error code:            1
167:  error diag:   0.000000000000000E+000  0.000000000000000E+000
167:   0.000000000000000E+000  0.000000000000000E+000  0.000000000000000E+000
167:  lat:   -4.73684210526324      longitidue:   280.000000000000     
167:  is recruitment:  T
167:  layer:            4
167:  wb_step_err =   6.231370315263201E-005
167:  q_top_eff*dt_step =   1.278546902632006E-006
167:  w_tot_beg =    392231909620.569     
167:  w_tot_end =    392231909620.569     
167:  leaf water:   1.977119271375372E-002  kg/plant
167:  stem_water:   1.307778893556870E-002  kg/plant
167:  troot_water:   1.012960357648249E-002
167:  aroot_water:   2.423778333540367E-002
167:  LWP:   -11.4704920165053     
167:  dbh:   0.764072176312517     
167:  pft:            2
167:  z nodes:    1.25000000000000       0.600000000000000     
167:  -0.222826201934367      -0.227329413955693      -0.227329413955693     
167:  psi_z:   1.224999999999987E-002  5.879999999999441E-003 -2.183696778956801E-003
167:  -2.227828256765818E-003 -2.227828256765818E-003
167:  vol,    theta,   H,  Psi,     kmax-
167:  flux:            1.278546902632006E-006
167:  l:  8.911960606182126E-005  0.221850090989391       -11.4582420165053     
167:   -11.4704920165053     
167:                          7.288414898907708E-004
167:  s:  5.149745376445012E-005  0.253950204905016       -11.4382420068961     
167:   -11.4441220068961     
167:                          1.938993985079932E-003
167:  t:  3.987109367687730E-005  0.254058834166293       -11.4182420060674     
167:                          1.324312610523329E-003
167:  a:  5.650707462205153E-006  0.713138515214688      -0.705970633341333     
167:                     in:  2.645558340631056E-003
167:                    out:  1.268879133417446E-005
167:  r1:   1473234852.71776       0.266238549065649      -0.706057093286554     
167:                          0.000000000000000E+000
167:  r2:  0.000000000000000E+000  0.000000000000000E+000   299.249193326083     
167:                          0.000000000000000E+000
167:  r3:  -8.00830603553817       0.000000000000000E+000  4.743030200075967E-322
167:                          0.000000000000000E+000
167:  r4:  -8.00121554768493       0.000000000000000E+000  8.744961931390064E-322
167:                          0.000000000000000E+000
167:  r5:  -7.99276885173250       0.248867590478379       0.000000000000000E+000
167:  kmax_aroot_radial_out:   1.274987965542320E-005
167:  surf area of root:   0.127498796554232     
167:  aroot_frac_plant:    1191934.06224864        202.920637098726     
167:   1.702448512260032E-004
167:  kmax_upper_shell:    502.128225197214     
167:  kmax_lower_shell:    21.9675255422892     
167:  
167:  tree lai:   0.756667509231287       m2/m2 crown
167:  area and area to volume ratios
167:  
167:  a:  5.650707462205153E-006
167:                          0.127498796554232     
167:  r1:   1473234852.71776     
167:                           1938224.29436465     
167:  r2:  0.000000000000000E+000
167:                        
167:  r3:  -8.00830603553817     
167:                        
167:  r4:  -8.00121554768493     
167:                        
167:  r5:  -7.99276885173250     
167:  inner shell kmaxs:    649.823728610446        488.471638384210     
167:    546.597463408998        502.128225197214        410.895270938480     
167:    323.642125620401        354.118419449814        329.681453045857     
167:    139.079050444771     
167:  ENDRUN:
167:  ERROR in FatesPlantHydraulicsMod.F90 at line 3456  

Using the fates_hydro_solver=2 did not help, and I also got the same error using this branch.

I tried increasing fates_mort_scalar_hydrfailure to 1.0 and replacing the linear function for hmort in EDMortalityFunctionsMod.F90 with an exponential. This allowed the case to run for longer before failing (11 years as opposed to two years), with error message:

 23:  EDPlantHydraulics water balance error exceeds threshold of = 
 23:   2.000000000000000E-005
 23:  transpiration demand:   1.760214942429399E-006  kg/step/plant
 23:  leaf water:   4.838807086771196E-002  kg/plant
 23:  stem_water:   2.758020242712950E-002  kg/plant
 23:  root_water:   4.295458774597534E-002  kg/plant
 23:  LWP:   -1.84694996093702     
 23:  dbh:   0.764072176312517     
 23:  pft:            2
 23:  tree lai:   0.756667509231287       m2/m2 crown
 23:  recruitflag:  F
 23:  lat:  -21.7894736842106      lon:   292.500000000000     
 23:  ENDRUN:
 23:  ERROR in FatesPlantHydraulicsMod.F90 at line 3866 

For now I am just going to adjust my region to avoid dry areas outside the Amazon, but I wanted to post about this in case anyone else encountered this type of error.

ETA these runs are in no-comp mode with three evergreen tropical tree PFTs.

[jennykowalcz]

PR 1164 with fates_hydro_solver=2 resolves this problem, so I am closing this issue. (Note that with fates_hydro_solver=1, it still fails with a "Could not find a stable solution for hydro 1D solve" error in one of the dry gridcells.)