Merge tag 'ctsm5.1.dev122' into ctsm.newtestgrids

Rework handling of evaporation constraint in SoilFluxes Occasionally, h2osoi_ice was going significantly negative in UpdateState_TopLayerFluxes - see ESCOMP#1979. As noted in that issue, this seems to be due to h2osoi_ice having a very different magnitude from h2osoi_liq, leading to greater-than-roundoff-level differences from zero final state in a relative sense (i.e., relative to the magnitude of h2osoi_ice) - I think because of the appearance of the sum (h2osoi_ice + h2osoi_liq) in the equations that limit fluxes. To try to deal with this, I have reworked the handling of the evaporation constraint to directly limit both the liqevap and solidevap, so that both of them should result in the equivalent liq or ice states going to 0 within roundoff. To do that, I needed to move the partitioning of the total flux into liquid and solid to earlier in the subroutine and then recalculate those partitioning fluxes in conditions where we're applying an evaporation constraint. Note that I applied a max of 0 to the new fluxes because many initial conditions files have roundoff-level negative H2OSOI_LIQ, so without this limit, we were getting roundoff-level negative fluxes. Resolves ESCOMP#1979
adamrher · Apr 27, 2023 · 7572d07 · 7572d07
2 parents bbe21f4 + cdb6187
commit 7572d07
Show file tree

Hide file tree

Showing 9 changed files with 367 additions and 93 deletions.
diff --git a/Externals_CLM.cfg b/Externals_CLM.cfg
@@ -2,7 +2,7 @@
 local_path = src/fates
 protocol = git
 repo_url = https://github.com/NGEET/fates
-tag = sci.1.61.0_api.25.0.0
+tag = sci.1.65.3_api.25.4.0
 required = True
 
 [externals_description]

diff --git a/bld/namelist_files/namelist_defaults_ctsm.xml b/bld/namelist_files/namelist_defaults_ctsm.xml
@@ -493,7 +493,7 @@ attributes from the config_cache.xml file (with keys converted to upper-case).
 <!-- FATES default parameter file                                       -->
 <!-- ================================================================== -->
 
-<fates_paramfile>lnd/clm2/paramdata/fates_params_api.25.0.0_12pft_c221128.nc</fates_paramfile>
+<fates_paramfile>lnd/clm2/paramdata/fates_params_api.25.4.0_12pft_c230327.nc</fates_paramfile>
 
 <!-- ========================================================================================  -->
 <!-- clm 5.0 BGC nitrogen model                                                                -->

diff --git a/cime_config/testdefs/ExpectedTestFails.xml b/cime_config/testdefs/ExpectedTestFails.xml
@@ -51,13 +51,6 @@
     </phase>
   </test>
 
-  <test name="FUNITCTSM_P1x1.f10_f10_mg37.I2000Clm50Sp.cheyenne_intel">
-    <phase name="RUN">
-      <status>FAIL</status>
-      <issue>#1972</issue>
-    </phase>
-  </test>
-
   <!-- fates test suite failures -->
 
   <test name="ERS_Lm12.1x1_brazil.I2000Clm50FatesCruRsGs.cheyenne_intel.clm-FatesFireLightningPopDens">

diff --git a/cime_config/testdefs/testlist_clm.xml b/cime_config/testdefs/testlist_clm.xml
@@ -1876,7 +1876,7 @@
       <machine name="cheyenne" compiler="gnu" category="fates"/>
     </machines>
     <options>
-      <option name="wallclock">00:40:00</option>
+      <option name="wallclock">00:50:00</option>
     </options>
   </test>
   <test name="ERS_Lm13" grid="f45_f45_mg37" compset="I2000Clm50Fates" testmods="clm/FatesColdNoComp">

diff --git a/doc/.ChangeLog_template b/doc/.ChangeLog_template
@@ -121,15 +121,10 @@ infrastructure should be run when appropriate, as described below.
 
     clm_pymods test suite on cheyenne - 
 
-  regular tests:
-      - aux_clm: https://github.com/ESCOMP/CTSM/wiki/System-Testing-Guide#pre-merge-system-testing
-      - Fortran unit tests on cheyenne (until https://github.com/ESCOMP/CTSM/issues/1972 is resolved): from src, run:
-          ../cime/scripts/fortran_unit_testing/run_tests.py --build-dir `mktemp -d --tmpdir=. unit_tests.XXXXXXXX`
-
-    aux_clm on cheyenne ------------
-    aux_clm on izumi ---------------
-    Fortran unit tests on cheyenne -
+  regular tests (aux_clm: https://github.com/ESCOMP/CTSM/wiki/System-Testing-Guide#pre-merge-system-testing):
 
+    cheyenne ----
+    izumi -------
 
   fates tests: (give name of baseline if different from CTSM tagname, normally fates baselines are fates-<FATES TAG>-<CTSM TAG>)
     cheyenne ----

diff --git a/doc/ChangeLog b/doc/ChangeLog
@@ -1,4 +1,275 @@
 ===============================================================
+Tag name: ctsm5.1.dev122
+Originator(s): sacks (Bill Sacks)
+Date: Sun Apr 23 19:36:37 MDT 2023
+One-line Summary: Rework handling of evaporation constraint in SoilFluxes
+
+Purpose and description of changes
+----------------------------------
+
+Occasionally, h2osoi_ice was going significantly negative in
+UpdateState_TopLayerFluxes - see
+https://github.com/ESCOMP/CTSM/issues/1979. As noted in that issue, this
+seems to be due to h2osoi_ice having a very different magnitude from
+h2osoi_liq, leading to greater-than-roundoff-level differences from zero
+final state in a relative sense (i.e., relative to the magnitude of
+h2osoi_ice) - I think because of the appearance of the sum (h2osoi_ice +
+h2osoi_liq) in the equations that limit fluxes.
+
+To try to deal with this, I have reworked the handling of the
+evaporation constraint to directly limit both the liqevap and solidevap,
+so that both of them should result in the equivalent liq or ice states
+going to 0 within roundoff.
+
+To do that, I needed to move the partitioning of the total flux into
+liquid and solid to earlier in the subroutine and then recalculate those
+partitioning fluxes in conditions where we're applying an evaporation
+constraint.
+
+Note that I applied a max of 0 to the new fluxes because many initial
+conditions files have roundoff-level negative H2OSOI_LIQ, so without
+this limit, we were getting roundoff-level negative fluxes.
+
+Significant changes to scientifically-supported configurations
+--------------------------------------------------------------
+
+Does this tag change answers significantly for any of the following physics configurations?
+(Details of any changes will be given in the "Answer changes" section below.)
+
+    [Put an [X] in the box for any configuration with significant answer changes.]
+
+[ ] clm5_1
+
+[ ] clm5_0
+
+[ ] ctsm5_0-nwp
+
+[ ] clm4_5
+
+
+Bugs fixed or introduced
+------------------------
+CTSM issues fixed (include CTSM Issue #):
+- Resolves ESCOMP/CTSM#1979 (Need some changes to avoid negative h2osoi_ice in UpdateState_TopLayerFluxes)
+
+
+Testing summary:
+----------------
+
+  regular tests (aux_clm: https://github.com/ESCOMP/CTSM/wiki/System-Testing-Guide#pre-merge-system-testing):
+
+    cheyenne ---- OK
+    izumi ------- OK
+
+    Tests passed, some baseline differences as expected.
+
+Answer changes
+--------------
+
+Changes answers relative to baseline: YES
+
+  Summarize any changes to answers, i.e.,
+    - what code configurations: potentially all
+    - what platforms/compilers: potentially all
+    - nature of change (roundoff; larger than roundoff/same climate; new climate):
+      roundoff
+
+   Differences were only observed in a few tests:
+   - ERP_P36x2_Ld30.f45_f45_mg37.I2000Clm51FatesSpCruRsGs.cheyenne_intel.clm-FatesColdSatPhen
+   - ERI_D_Ld9_P48x1.T31_g37.I2000Clm50Sp.izumi_nag.clm-reduceOutput
+   - SMS_D_Ln9_P36x3.f19_g17.IHistClm50Sp.cheyenne_intel.clm-waccmx_offline
+   - SMS_D_Ln9_P36x3_Vmct.f19_g17.IHistClm50Sp.cheyenne_intel.clm-waccmx_offline
+
+   If bitwise differences were observed, how did you show they were no worse
+   than roundoff?
+
+     Only two tests had greater-than-roundoff-level differences in the
+     cprnc output:
+     SMS_D_Ln9_P36x3.f19_g17.IHistClm50Sp.cheyenne_intel.clm-waccmx_offline
+     and the mct equivalent,
+     SMS_D_Ln9_P36x3_Vmct.f19_g17.IHistClm50Sp.cheyenne_intel.clm-waccmx_offline.
+     To verify that differences were fundamentally no greater than
+     roundoff-level, I introduced temporary code; this minimal diff
+     ended up being enough to give just roundoff-level differences from baseline:
+
+       diff --git a/src/biogeophys/SoilFluxesMod.F90 b/src/biogeophys/SoilFluxesMod.F90
+       index c316d30fe..6a958c0ee 100644
+       --- a/src/biogeophys/SoilFluxesMod.F90
+       +++ b/src/biogeophys/SoilFluxesMod.F90
+       @@ -45,7 +45,7 @@ subroutine SoilFluxes (bounds, num_urbanl, filter_urbanl, &
+            ! Update surface fluxes based on the new ground temperature
+            !
+            ! !USES:
+       -    use clm_time_manager , only : get_step_size_real
+       +    use clm_time_manager , only : get_step_size_real, get_nstep
+            use clm_varcon       , only : hvap, cpair, grav, vkc, tfrz, sb 
+            use landunit_varcon  , only : istsoil, istcrop
+            use column_varcon    , only : icol_roof, icol_sunwall, icol_shadewall, icol_road_perv
+       @@ -79,7 +79,9 @@ subroutine SoilFluxes (bounds, num_urbanl, filter_urbanl, &
+            real(r8) :: t_grnd0(bounds%begc:bounds%endc)                   ! t_grnd of previous time step
+            real(r8) :: lw_grnd
+            real(r8) :: evaporation_limit                                  ! top layer moisture available for evaporation
+       -    real(r8) :: evaporation_demand                                   ! evaporative demand 
+       +    real(r8) :: evaporation_demand                                   ! evaporative demand
+       +    real(r8) :: qflx_liqevap_orig
+       +    real(r8) :: qflx_solidevap_orig
+            !-----------------------------------------------------------------------
+
+            associate(                                                                & 
+       @@ -291,6 +293,7 @@ subroutine SoilFluxes (bounds, num_urbanl, filter_urbanl, &
+                       qflx_evap_soi(p)   = qflx_evap_soi(p) - frac_sno_eff(c)*(evaporation_demand - evaporation_limit)
+                       qflx_liqevap_from_top_layer(p)   = max(h2osoi_liq(c,j)/(frac_sno_eff(c)*dtime), 0._r8)
+                       qflx_solidevap_from_top_layer(p) = max(h2osoi_ice(c,j)/(frac_sno_eff(c)*dtime), 0._r8)
+       +
+                       ! conserve total energy flux
+                       eflx_sh_grnd(p) = eflx_sh_grnd(p) + frac_sno_eff(c)*(evaporation_demand - evaporation_limit)*htvp(c)
+                    endif
+       @@ -307,6 +310,24 @@ subroutine SoilFluxes (bounds, num_urbanl, filter_urbanl, &
+                       qflx_ev_snow(p)    = qflx_evap_soi(p)
+                       qflx_liqevap_from_top_layer(p)   = max(h2osoi_liq(c,j)/dtime, 0._r8)
+                       qflx_solidevap_from_top_layer(p) = max(h2osoi_ice(c,j)/dtime, 0._r8)
+       +
+       +               if (h2osoi_liq(c,j) + h2osoi_ice(c,j) > 0._r8) then
+       +                  qflx_liqevap_orig = max(qflx_evap_soi(p)*(h2osoi_liq(c,j)/ &
+       +                       (h2osoi_liq(c,j)+h2osoi_ice(c,j))), 0._r8)
+       +               else
+       +                  qflx_liqevap_orig = 0._r8
+       +               end if
+       +               qflx_solidevap_orig = qflx_evap_soi(p) - qflx_liqevap_orig
+       +               if (qflx_solidevap_from_top_layer(p) == 0._r8 .and. &
+       +                    qflx_solidevap_orig < 0._r8 .and. &
+       +                    qflx_solidevap_orig > -1.e-16_r8) then
+       +                  write(iulog,'(a, i0, 1x, i0, 1x, 5e24.17)') &
+       +                       'WJS adj urb: solid orig le 0, new 0: nstep, p, orig, new, qflx_evap_soi, h2osoi_liq, h2osoi_ice = ', &
+       +                       get_nstep(), p, qflx_solidevap_orig, qflx_solidevap_from_top_layer(p), &
+       +                       qflx_evap_soi(p), h2osoi_liq(c,j), h2osoi_ice(c,j)
+       +                  qflx_solidevap_from_top_layer(p) = qflx_solidevap_orig
+       +               end if
+       +
+                       ! conserve total energy flux
+                       eflx_sh_grnd(p) = eflx_sh_grnd(p) +(evaporation_demand -evaporation_limit)*htvp(c)
+                    endif
+
+     (Note that the diffs in
+     ERP_P36x2_Ld30.f45_f45_mg37.I2000Clm51FatesSpCruRsGs.cheyenne_intel.clm-FatesColdSatPhen
+     were ambiguous as to whether they were roundoff-level due to the
+     single-precision output in that test; I reran with double precision
+     for the baseline and the branch and was able to verify that the
+     diffs were only double-precision roundoff-level.)
+
+Other details
+-------------
+
+Pull Requests that document the changes (include PR ids):
+https://github.com/ESCOMP/CTSM/pull/1987
+
+===============================================================
+===============================================================
+Tag name: ctsm5.1.dev121
+Originator(s): glemieux (Gregory Lemieux,LBL/NGEET,510-486-5049)
+Date: Wed Apr  5 13:34:09 MDT 2023
+One-line Summary: Changes soil moisture initialization logic for FATES
+
+Purpose and description of changes
+----------------------------------
+
+This PR changes the logic for soil moisture initialization to initialize
+with wetter soils (75% of saturated water content, as opposed to 15% of
+absolute water content) for all FATES configurations. The rationale for
+this is that in FATES-nocomp simulations, Jessica Needham was finding very
+high initial mortality rates in some seasonal tropical forest regions which
+she traced it back to the initial soil moisture killing off plants before the
+ecosystem could get established.
+
+This also updates the fates externals pointer to the latest tag which includes
+a number of science updates since the last tag update and updates the the
+default parameter file.
+
+Significant changes to scientifically-supported configurations
+--------------------------------------------------------------
+
+Does this tag change answers significantly for any of the following physics configurations?
+(Details of any changes will be given in the "Answer changes" section below.)
+
+    [Put an [X] in the box for any configuration with significant answer changes.]
+
+[ ] clm5_1
+
+[ ] clm5_0
+
+[ ] ctsm5_0-nwp
+
+[ ] clm4_5
+
+
+Bugs fixed or introduced
+------------------------
+
+Externals issues fixed (include issue #):
+    Partially addresses FATES#994 -- Bare ground establishment problem and increased soil moisture
+
+Known bugs found since the previous tag (include issue #):
+    #1979 -- Need to loosen tolerance on near-zero truncation of h2osoi_ice in UpdateState_TopLayerFluxes
+
+Notes of particular relevance for users
+---------------------------------------
+
+Changes made to namelist defaults (e.g., changed parameter values):
+   fates_paramfile updated to fates_params_api.25.4.0_12pft_c230327.nc
+
+Notes of particular relevance for developers:
+---------------------------------------------
+
+Caveats for developers (e.g., code that is duplicated that requires double maintenance):
+    This changes the starting level for a COLD start of soil moisture for ALL FATES cases to a much
+    higher value than for non-FATES. In the long run we'd like to have these the same and/or
+    have the value changable on the namelist.
+
+Testing summary:
+----------------
+
+ [PASS means all tests PASS; OK means tests PASS other than expected fails.]
+
+  regular tests (aux_clm: https://github.com/ESCOMP/CTSM/wiki/System-Testing-Guide#pre-merge-system-testing):
+
+    cheyenne ---- OK
+    izumi ------- OK
+
+  fates tests: (give name of baseline if different from CTSM tagname, normally fates baselines are fates-<FATES TAG>-<CTSM TAG>)
+    cheyenne ---- OK
+    izumi ------- OK
+
+If the tag used for baseline comparisons was NOT the previous tag, note that here:
+
+  FATES tests run against fates-sci.1.65.3_api.25.4.0-ctsm5.1.dev120 baseline
+
+
+Answer changes
+--------------
+
+Changes answers relative to baseline:
+
+   Changes answers in fates suite for all non-hydro fates tests since the soil
+   moisture initialization matches that of fates hydro now.  Changes answer
+   for all fates testmods in the aux_clm suite as the science tag has iterated
+   forward by 4 minor version updates.  All diffs accounted for with prior fates
+   suite tests.
+
+Other details
+-------------
+
+List any externals directories updated (cime, rtm, mosart, cism, fates, etc.):
+
+- FATES: sci.1.61.0_api.25.0.0 -> sci.1.65.3_api.25.4.0
+
+Pull Requests that document the changes (include PR ids):
+(https://github.com/ESCOMP/ctsm/pull)
+
+  https://github.com/ESCOMP/CTSM/pull/1962 -- Cold start moisture for FATES increased
+  https://github.com/ESCOMP/CTSM/pull/1978 -- revert some commits now that FUNITCTSM works again
+
+===============================================================
+===============================================================
 Tag name: ctsm5.1.dev120
 Originator(s): sacks (Bill Sacks)
 Date: Sat Mar 25 17:49:27 MDT 2023

diff --git a/doc/ChangeSum b/doc/ChangeSum
@@ -1,5 +1,7 @@
 Tag                   Who      Date  Summary
 ============================================================================================================================
+    ctsm5.1.dev122    sacks 04/23/2023 Rework handling of evaporation constraint in SoilFluxes
+    ctsm5.1.dev121 glemieux 04/05/2023 Changes soil moisture initialization logic for FATES
     ctsm5.1.dev120    sacks 03/25/2023 Update externals and minor fixes
     ctsm5.1.dev119   slevis 03/16/2023 Allow gross unrepresented land use transition (PR #309)
     ctsm5.1.dev118   slevis 02/05/2023 Use conda environment rather than ncar_pylib with the fsurdat_modifier system test