Port CLM cryosphere and hydrology fixes into ELM by daliwang · Pull Request #8638 · E3SM-Project/E3SM

daliwang · 2026-08-17T14:46:29Z

Summary

Port CLM cryosphere / hydrology bug fixes into ELM (snow energy and melt compaction, bedrock heat capacity, active-element accumulator masking, surface-water runoff, snow balance accounting).
Changes are in ELM biogeophysics and accumulator code only.

Commits

Port CLM fractional-snow energy and melt-compaction fixes
Port CLM bedrock heat capacity fix in SoilTemperatureMod
Port CLM active-element masking for ELM accumulators
Bug fixes for surface water runoff calculation
Fix snow balance accounting issue

Summary

Port CLM cryosphere / hydrology bug fixes into ELM (fractional-snow energy and melt compaction, bedrock heat capacity, active-element accumulator masking, surface-water runoff, snow balance accounting).
ELM biogeophysics and accumulator code only.

Testing

e3sm_land_developer on Frontier / craygnu (project cli115), 2026-08-14.

Branch: lnd/port-clm-cryosphere-fixes-master @ fbfcc93f52
Gold generated from parent master @ a899004464, then compared on this branch
Personal baseline root (not shared Frontier gold): /lustre/orion/cli115/world-shared/wangd/kmELM/baselines

Result	Count	Notes
BASELINE DIFF	48	Expected: snow, soil T, runoff, coupled fluxes change vs `a899004464`
PASS vs gold	5	MOSART-only tests; bit-for-bit
RUN FAIL (walltime)	2	Same two r05 tests timed out on generate and compare
MEMLEAK FAIL	4	Already present on parent master; not new on this branch

Namelists (NLCOMP) unchanged. All completed ERS tests passed COMPARE_base_rest (restart is bit-for-bit with itself). Debug builds DIFF, they do not crash.

The two walltime FAILs (no gold, so no baseline compare):

ERS.r05_r05.ICNPRDCTCBC.frontier_craygnu.elm-cbudget
ERS.r05_r05.IELM.frontier_craygnu.elm-V2_ELM_MOSART_features

Allow frac_sno_eff and melt-compaction ddz3 fractional forms for all non-urban, non-lake columns when subgrid fluxes are on, matching ctsm5.4.042 behavior for glacier and wetland columns. Co-authored-by: Cursor <cursoragent@cursor.com>

Do not scale bedrock volumetric heat capacity by soil porosity (1-watsat). Below nlevbed, use solid-rock heat capacity times layer thickness, consistent with ctsm5.4.042. Co-authored-by: Cursor <cursoragent@cursor.com>

Skip inactive subgrid points when updating timeavg, runmean, and runaccum fields; track per-point nsteps for correct averaging and restart, matching the ctsm5.4.042 accumulator redesign. Co-authored-by: Cursor <cursoragent@cursor.com>

bishtgautam · 2026-08-17T19:10:09Z

+                !if (use_modified_infil) then
                  if (frac_h2osfc_act(c) <= pc .and. frac_h2osfc(c) <= pc) then 
                     frac_infclust=0.0_r8
                  else
-                      if (frac_h2osfc(c) <= pc) then
+                      if (frac_h2osfc_act(c) <= pc) then
                        frac_infclust=(frac_h2osfc_act(c)-pc)**mu
                      else
-                        frac_infclust=(frac_h2osfc(c)-pc)**mu
+                        frac_infclust=(frac_h2osfc_act(c)-pc)**mu
                      endif
                  endif
-                else
-                  if (frac_h2osfc(c) <= pc) then
-                    frac_infclust=0.0_r8
-                  else
-                    frac_infclust=(frac_h2osfc(c)-pc)**mu
-                  endif
-                endif
+                !else
+                !  if (frac_h2osfc(c) <= pc) then
+                !    frac_infclust=0.0_r8
+                !  else
+                !    frac_infclust=(frac_h2osfc(c)-pc)**mu
+                !  endif
+                !endif


It seems to me that we are making the use_modified_infil scheme as the default. So, why are we not doing it through namelist instead of commented out the old code and making the use_modified_infil as the default?

Or, if use_modified_infil needs to be made default, let's remove it from the namelist and delete the code corresponding to .not. use_modified_infil.

use_modified_infil is still used to control how the infiltration capacity is computed for different sub fractions. This modification just remove the control of use_modified_infil on the surface water runoff generation.

Copilot

Pull request overview

This pull request ports a set of CLM cryosphere and hydrology fixes into ELM, focusing on snow energy / compaction behavior, soil thermal properties, accumulator handling for inactive subgrid elements, surface-water runoff, and snow-balance accounting within ELM biogeophysics and accumulator logic.

Changes:

Update ELM accumulators to skip inactive / zero-weight subgrid elements and track active-step counts via restartable _NSTEPS.
Apply CLM fixes to bedrock heat capacity handling and fractional-snow treatments (energy and melt compaction) across glacier/wetland-related pathways.
Adjust surface-water runoff / infiltration logic and update glacier snow water-balance accounting for fractional snow cover effects.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
components/elm/src/main/accumulMod.F90	Adds active-element masking to accumulators and introduces per-point step counts persisted via restart.
components/elm/src/biogeophys/SoilTemperatureMod.F90	Fixes bedrock-layer heat capacity calculation below the soil-bedrock boundary.
components/elm/src/biogeophys/SoilHydrologyMod.F90	Updates surface-water runoff / infiltration fraction logic and runoff coefficient behavior.
components/elm/src/biogeophys/SnowHydrologyMod.F90	Adjusts melt-compaction gating to follow CLM fractional-snow behavior outside lake/urban points.
components/elm/src/biogeophys/CanopyHydrologyMod.F90	Extends fractional-snow effective coverage logic beyond soil/crop to non-urban, non-lake columns under subgrid fluxes.
components/elm/src/biogeophys/BalanceCheckMod.F90	Updates snow-balance accounting on glacier columns to reflect fractional snow-covered flux partitioning.

Suppressed comments (1)

components/elm/src/main/accumulMod.F90:588

Same issue as the single-level updater: in the multi-level timeavg path, reset and normalization are skipped for inactive points because they are inside if (accum(nf)%active(k)). This can leave stale val/nsteps for inactive points (especially after restart or dynamic reweighting) and can contaminate averages when points become active later. Apply reset/normalization for all points; only add field for active points.

       time_to_reset = (mod(nstep,accum(nf)%period) == 1 .or. accum(nf)%period == 1) .and. (nstep /= 0)
       do j = 1,numlev
          do k = beg,end
             if (accum(nf)%active(k)) then
                if (time_to_reset) then
                   accum(nf)%val(k,j) = 0._r8
                   accum(nf)%nsteps(k,j) = 0
                end if

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

+       time_to_reset = (mod(nstep,accum(nf)%period) == 1 .or. accum(nf)%period == 1) .and. (nstep /= 0)
+       do k = beg,end
+          if (accum(nf)%active(k)) then
+             if (time_to_reset) then
+                accum(nf)%val(k,1) = 0._r8
+                accum(nf)%nsteps(k,1) = 0
+             end if
+             accum(nf)%val(k,1) = accum(nf)%val(k,1) + field(k)
+             accum(nf)%nsteps(k,1) = accum(nf)%nsteps(k,1) + 1
+             if (mod(nstep,accum(nf)%period) == 0) then
+                if (accum(nf)%nsteps(k,1) > 0) then
+                   accum(nf)%val(k,1) = accum(nf)%val(k,1) / real(accum(nf)%nsteps(k,1),r8)
+                else
+                   accum(nf)%val(k,1) = accum(nf)%initval
+                end if
+             end if
+          end if
+       end do


bishtgautam · 2026-08-17T19:36:14Z

             if (h2osfcflag==1) then
                ! calculate runoff from h2osfc  -------------------------------------
-                if (use_modified_infil) then
+                !if (use_modified_infil) then
                  if (frac_h2osfc_act(c) <= pc .and. frac_h2osfc(c) <= pc) then 
                     frac_infclust=0.0_r8
                  else
-                      if (frac_h2osfc(c) <= pc) then
+                      if (frac_h2osfc_act(c) <= pc) then
                        frac_infclust=(frac_h2osfc_act(c)-pc)**mu
                      else
-                        frac_infclust=(frac_h2osfc(c)-pc)**mu
+                        frac_infclust=(frac_h2osfc_act(c)-pc)**mu
                      endif


I agree with Copilot's review: Why is the second frac_infclust=(frac_h2osfc(c)-pc)**mu being changed to frac_infclust=(frac_h2osfc_act(c)-pc)**mu?

bishtgautam · 2026-08-17T19:31:48Z

+          ! Port of CLM fractional-snow energy fix: when subgrid fluxes are on,
+          ! allow fractional frac_sno_eff for all non-urban, non-lake columns
+          ! (including glacier and wetland), matching soil/crop treatment.
+          if (subgridflag == 1 .and. .not. urbpoi(l) .and. ltype(l) /= istdlak) then


Is the default value of subgridflag 1 or 0?

bishtgautam · 2026-08-17T19:44:53Z

                if(h2osfc(c) >= h2osfc_thresh(c) .and. h2osfcflag/=0) then
                   ! spatially variable k_wet
-                   k_wet=1.0_r8 * sin((rpi/180._r8) * col_pp%topo_slope(c))
+                   k_wet=1.0e-4_r8 * sin((rpi/180._r8) * max(col_pp%topo_slope(c), 1.0e-3_r8))


What is the logic of reducing k_wet by four-orders of magnitude? Was this a bug identified by CLM that we are fixing here?

A bug from the misuse of unit

czender

These changes are beyond my level of ELM expertise. I couldn't find anything that looked problematic that had not already been noted.

daliwang and others added 5 commits August 14, 2026 00:04

Port CLM bedrock heat capacity fix in SoilTemperatureMod

3f7fd0f

Do not scale bedrock volumetric heat capacity by soil porosity (1-watsat). Below nlevbed, use solid-rock heat capacity times layer thickness, consistent with ctsm5.4.042. Co-authored-by: Cursor <cursoragent@cursor.com>

Bug fixes for surface water runoff calculation

4b7a01a

Fix snow balance accounting issue

fbfcc93

daliwang requested review from bishtgautam, czender and thorntonpe August 17, 2026 15:37

bishtgautam requested a lite review from Copilot August 17, 2026 19:05

Copilot started reviewing on behalf of bishtgautam August 17, 2026 19:05 View session

bishtgautam reviewed Aug 17, 2026

View reviewed changes

Copilot AI reviewed Aug 17, 2026

View reviewed changes

bishtgautam reviewed Aug 17, 2026

View reviewed changes

czender reviewed Aug 17, 2026

View reviewed changes

daliwang mentioned this pull request Aug 18, 2026

Port CLM cryosphere fixes and ELM h2osfc ponding/runoff adjustments onto maint-3.0 #8643

Open

rljacob added ELM land model bug fix PR labels Aug 19, 2026

rljacob assigned bishtgautam Aug 21, 2026

bishtgautam added the non-BFB PR makes roundoff changes to answers. label Aug 28, 2026

Navigation Menu

Uh oh!

Port CLM cryosphere and hydrology fixes into ELM - #8638

Port CLM cryosphere and hydrology fixes into ELM#8638
daliwang wants to merge 5 commits into
E3SM-Project:masterfrom
daliwang:lnd/port-clm-cryosphere-fixes-master

daliwang commented Aug 17, 2026

Uh oh!

bishtgautam Aug 17, 2026

Uh oh!

donghuix Aug 21, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

bishtgautam Aug 17, 2026

Uh oh!

bishtgautam Aug 17, 2026

Uh oh!

bishtgautam Aug 17, 2026

Uh oh!

daliwang Aug 20, 2026

Uh oh!

czender left a comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Uh oh!

Conversation

daliwang commented Aug 17, 2026

Summary

Commits

Summary

Testing

Uh oh!

bishtgautam Aug 17, 2026

Choose a reason for hiding this comment

Uh oh!

donghuix Aug 21, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

bishtgautam Aug 17, 2026

Choose a reason for hiding this comment

Uh oh!

bishtgautam Aug 17, 2026

Choose a reason for hiding this comment

Uh oh!

bishtgautam Aug 17, 2026

Choose a reason for hiding this comment

Uh oh!

daliwang Aug 20, 2026

Choose a reason for hiding this comment

Uh oh!

czender left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants