Implementing optional SI units in SMOKE emission inputs

[hnqtran]

Potential implementation of metric tons as optional unit in SMOKE emission inputs to accommodate SMOKE applications outside U.S.

[hnqtran]

Found wrong unit conversion issue in src/lib/unitfac.f (lines 158 - 168)

            CASE( 'kg' )
                
                SELECT CASE( BUF2 )
                    CASE( 'g' )
                        UNITFAC = 1000.
                    CASE( 'tons' )
                        UNITFAC = GM2TON / 1000.
                     CASE( 'kg' )
                     CASE DEFAULT
                        CALL CASE_NOT_FOUND( UNITFAC )
                END SELECT

Conversion of 'kg' to 'tons' (US short tons) should be UNITFAC = GM2TON * 1000.

Reason: As defined in src/inc/IOCNST3.EXT:
TON2GM = 907184.74 ! Short tons to grams
GM2TON = 1. / TON2GM ! Grams to short tons (= 1.1023E-6)
Thus, 'kg' to 'tons' = GM2TON * 1000.

[hnqtran]

The overall approach is to create a gate keeper so that when a new env variable SI_UNITS_YN = "Y" specifying SI units in the inputs, the gate keeper will convert emission data from SI back to U.S. unit (or skipping converting stack parameters from U.S. to SI units for point sources) before going through subsequent subroutines as normal.

QA check is implemented to check for indicator of SI units in the header lines of emission input files (e.g., #UNIT_SYS=SI) to make sure that the user is providing data in SI units together with SI_UNITS_YN = "Y" setting in the run script.

Using SI unit in the smoke reporting files are already implemented in SMOKE through UNITS instruction in REPCONFIG files

[hnqtran]

Ongoing list of SMOKE subroutines required update:

• src/inc/EMCNST3.EXT (also see issue Unification of INCLUDE files #109)
• src/lib/unitfac.f
• src/lib/unitmatch.f
• src/smkinven/rdinvdata.f
• SMOKE User Manual

[cjcoats]

Inasmuch as SMOKE emissions units have implicit "per grid cell" and "per hour", implementing SI units is a big deal -- a much bigger deal than the comments above indicate. Note that consequently, SMOKE emissions units are not straightforwardly interpolatable from grid to grid or time step to time step.