Have been trying to understand how the climate models work. Looking at NASA Model E, it seem a strange beast, because it has lots of data files for input but some parameters are hard-coded into the Fortran code.
http://www.giss.nasa.gov/tools/modelE/modelEsrc/
Land Ice initialisation appears to have a fixed base ice growth rate
Code:
C**** The net accumulation from IPCC2 report is 2016x10**12 kg/year
C**** for Antarctica and for Greenland it is 316x10**12 kg/year
C****
!@var ACCPDA total accumulation per year for Antarctica (kg/yr)
!@var ACCPDG total accumulation per year for Greenland (kg/yr)
Fluxes are added in a predetermined order:
Code:
C****
C**** SURFACE INTERACTION AND GROUND CALCULATION
C****
C**** NOTE THAT FLUXES ARE APPLIED IN TOP-DOWN ORDER SO THAT THE
C**** FLUXES FROM ONE MODULE CAN BE SUBSEQUENTLY APPLIED TO THAT BELOW
C****
IF (MODD5S.EQ.0) CALL DIAGCA (4)
C**** APPLY PRECIPITATION TO SEA/LAKE/LAND ICE
CALL PRECIP_SI
CALL PRECIP_LI
C**** APPLY PRECIPITATION AND RUNOFF TO LAKES/OCEANS
CALL PRECIP_LK
CALL PRECIP_OC
CALL TIMER (MNOW,MSURF)
CALL CHECKT ('PRECIP')
Changing model behaviour seems a regular event:
Code:
c**** SLE001 E001M12 SOMTQ SLB211M9
c**** (same as frank''s soils64+2bare_soils+old runoff)
c**** change to evap calculation to prevent negative runoff
c**** soils45 but with snowmelt subroutine snmlt changed
c**** to melt snow before 1st layer ground ice.
ccc comments from new soils
c**** 8/11/97 - modified to include snow model
c**** 10/29/96 - aroot and broot back to original values; added
c**** call to cpars to change vegetation parameters for pilps.
c**** 9/7/96 - back to heat capacities/10
c**** 5/14/96 - added soils64 surface runoff changes/corrections
c**** 11/13/95 - changed aroot and broot for rainf for 1.5m root depth
c**** 10/11/95 - back to full heat capacity, to avoid cd oscillations.
c**** changes for pilps: (reversed)
c**** use soils100.com instead of soils45.com
c**** set im=36,jm=24
c**** set sdata,vdata and fdata to real*4
c**** divide canopy heat capacities by 10.d0
c**** change aroot of grass to 1.0d0, to reduce root depth.
c**** end of changes for pilps
c**** changes for pilps: (kept)
c**** modify gdtm surface flux timestep limits
c**** define new diagnostics
c**** zero out diagnostics at start of advnc
c**** end of changes for pilps
c**** modified for 2 types of bare soils
c****
c**** soils62 soils45 soils45 cdfxa 04/27/95
c**** same as soils45 but with snowmelt subroutine snmlt changed
c**** to melt snow before 1st layer ground ice.
ccc end comments from new soils
c**** also corrects evaps calculation.
c**** also includes masking effects in radation fluxes.
c**** modifies timestep for canopy fluxes.
c**** soils45 10/4/93
c**** uses bedrock as a soil texture. soil depth of 3.5m
c**** everywhere, where layers can have bedrock.
c**** requires sm693.data instead of sm691.data.
c**** sdata needs to be changed in calling program.
c**** soils44b 8/25/93
c**** uses snow conductivity of .088 w m-1 c-1 instead of .3d0
c**** soils44 8/16/93
c**** adds bedrock for heat calculations, to fill out the
c**** number of layers to ngm.
c**** soils43 6/25/93
c**** comments out call to fhlmt heat flux limits.
c**** uses ghinij to return wfc1, eliminates rewfc.
c**** soils42 6/15/93
c**** adds snow insulation
c**** soils41 5/24/93
c**** uses snow masking depth from vegetation height to determine
c**** fraction of snow that is exposed.
c**** reth must be called prior to retp.
c**** soils40 5/10/93
c**** removes snow from canopy and places it on vegetated soil.
c**** soils39 4/19/93
c**** modifications for real*8 or real*4 runs. common block
c**** ordering changed for efficient alignment. sdata,fdata,
c**** and vdata are explicitly real*4. on ibm rs/6000, should
c**** be compiled with -qdpc=e option for real*8 operation.
c**** to run real*4, change implicit statement in include file.
c**** soils38 2/9/93
c**** adds heat flux correction to handle varying coefficients
c**** of drag.
c**** soils37 1/25/93
c**** changes soil crusting parameter ku/d from .05 per hour to .1d0,
c**** to agree with morin et al.
c**** soils36 11/12/92
c**** calculates heat conductivity of soils using devries method.
c**** changes loam material heat capacity and conductivity
c**** to mineral values.
c**** soils35 10/27/92
c**** includes effect of soil crusting for infiltration by
c**** modifying hydraulic conductivity calculation of layer
c**** 1 in hydra.
c**** soils34 8/28/92
c**** uses effective leaf area index alaie for purposes of
c**** canopy conductance calculation.
c**** soils33 8/9/92
c**** changes canopy water storage capacity to .1mm per lai from 1.d0
c**** soils32 7/15/92
c**** 1) for purposes of infiltration only, reduces soil conductivity
c**** by (1-thetr*fice) instead of (1-fice).
c**** 2) betad is reduced by fraction of ice in each layer.
c**** 3) transpired water is removed by betad fraction in each layer,
c**** instead of by fraction of roots. prevents negative runoff.
c**** 4) speeds up hydra by using do loop instead of if check,
c**** by using interpolation point from bisection instead of logs,
c**** and by avoiding unnecessary calls to hydra. also elimates call
c**** to hydra in ma89ezm9.f.
c**** soils31 7/1/92
c**** 1) fixes fraction of roots when soil depth is less than root
c**** depth, thus fixing non-conservation of water.
c**** soils30 6/4/92
c**** 1) uses actual final snow depth in flux limit calculations,
c**** instead of upper and lower limits. fixes spurious drying
c**** of first layer.
c**** Added gpp, GPP terms 4/25/03 nyk
c**** Added decks parameter cond_scheme 5/1/03 nyk
