WIEMIP, Hillslope & CH4 setup

[wwieder]

This is a place to discuss results for WIEMIP simulations

DATM

• WIEMIP initialized with data from CRUJRA2025 (years 1850-1869) vs. default
• CRUJRA from 2024 (years 1901-1920) with data replaced over Antarctica and Greenland.
  • See Diagnostics, and comments in # 158

Hillslope fire modifications

• Flat model = no changes with suggested fire changes for the n02 tag,
  • Diagnostics vs # 157 and comments in # 161.
  • 1850 Burned area = 8.2 M km2/y for this f19 case.
• Hillslope large changes in burned area with suggested changes!
  • Diagnostics vs #147 and comments in # 159
    • 1850 Annual burned area = 11.4 M km2 for this f09 hillslope case vs.
  • Diagnostics vs flat #131 and comments in # 159
    • 1850 Annual burned area = 9.0 M km2 in this older flat f09 case

CH4 Fluxes

• Flat model wetlands.n01.ctsm5.4.029_f09noCrop_flat_158_ [AD, SASU, pSASU] #158
• Hillslope wetlands_fire.n01.ctsm5.4.029_f09noCrop_HH_159_ [AD, SASU, pSASU] #159, wetlands_fire.n01.ctsm5.4.029_f09noCrop_9HH_160_ [AD, SASU, pSASU, HIST] #160

Generic Crops

• see ctsm5.4.029_f19_wetlandCH4_157_ [AD, SASU, pSASU, HIST] #157 _cropHIST vs. _HIST (both cases use BGCcrop spinup)

[wwieder]

Table summarizing key results is linked here

[wwieder]

@lifang0209, can you help us with fire calibration now that we're running with these changes to fire calculations? Specifically, could you provide a few suggestions for fire tunings we could try (these runs are very slow)? If so, when do you think you could have this to us? Sorry for the added pressure, we're running up against deadlines for contributing to WIEMIP.

I've started several hillslope hydrology cases from the wetland branch. The spinup is still ongoing, but a quick look after 150 years of AD spinup suggest the annual burned area is too low (5.8 M km2; see issue note here)

Full diagnostics are posted here

case dir: /glade/work/wwieder/wiemip/wetlands_fire.n01.ctsm5.4.029_f09noCrop_9HH_160_test3
in scratch: /glade/derecho/scratch/wwieder/archive/wetlands_fire.n01.ctsm5.4.029_f09noCrop_9HH_160_test3/lnd

the parameter file I'm using is here /glade/work/wwieder/modified_inputs/ctsm60_params.genericCropFix.ch4cal_e.c260410.nc

[lifang0209]

@wwieder I'll try to send the tuned parameter tomorrow.

[lifang0209]

@wwieder Hi Will, could you run a 10-year test using the parameter file
/glade/u/home/fangli/ctsm60_params.genericCropFix.ch4cal_e.fire.c260422.nc
In this file, I increased the btran2 threshold slightly.
Could you also change borpeat_fire_soilmoist_denom from 0.3d00 to 0.35d00?

in your test case user_nl_clm, add:
borpeat_fire_soilmoist_denom=0.35d00
paramfile = '/glade/u/home/fangli/ctsm60_params.genericCropFix.ch4cal_e.fire.c260422.nc'

This is just a test case, so I’m not yet sure whether it is over-tuned or under-tuned. I’m traveling with my family, but I should have time tomorrow evening to check the fire simulation from the new test.

Also, where is your local path for HH_160_test3? Is it this one: /glade/u/home/wwieder/CTSM?
I’d like to run the test case today, but I’m not sure which local directory I should use to create the test case.

[wwieder]

Thanks @lifang0209, I appreciate this quick turnaround while you're traveling.
I've started an AD and test case in #163 with these suggested changes.

• the AD case is from a coldstart
• the test case has regular monthly h0 output from AD initial conditions in wetlands_fire.n01.ctsm5.4.029_f09noCrop_9HH_160_ [AD, SASU, pSASU, HIST] #160.

Yes, my sandbox for these case has been /glade/u/home/wwieder/CTSM. As of #163, this is on the wetlands branch, which is up to date with the CTSM wetlands.n02 branch tag, but results in #160 were basically from the identical code base.

I'll hopefully be able to post reults from the 163_test case tomorrow.

[wwieder]

Quick look at the simulation I did with these new parameters in #163 suggests we're getting annual burned area just under 7 M km2. Diagnostic are posted here.

[lifang0209]

@wwieder Hi Will, could you use this parameter file for the next run?
/glade/u/home/fangli/ctsm60_params.genericCropFix.ch4cal_e.fire.c260424.nc
In this file, I slightly tuned rswf_max, rswf_min for several tree PFTs, and fsr_pft, based on your 1850 simulation results.

Not only does soil moisture in natural vegetation columns change when hillslope is active, but the fire calculation structure changes as well. With hillslope off, fire is computed once for all natural vegetation in a grid cell (only one natural vegetation column), using predictors averaged across natural vegetation PFTs. With hillslope on, fire and its predictors are computed separately for each natural vegetation column and then averaged.

Once you get the historical transient run finished, I’d like to check the fire simulations again. After all, most of the observations we compare against, not just fire, are for the present day. Parameters tuned only against the 1850 run may have some issues. In particular, I cannot really evaluate deforestation and degradation fires in the 1850 run, because those depend on land-use data to calculate deforestation rates.

[wwieder]

Thanks Fang.

I'll try this file in a new run. Should I also use borpeat_fire_soilmoist_denom=0.35d00?

Just a heads up, I'm not sure we'll be able to run a historical case before we needs to start the WIEMIP simulations, so we'll have to do our best to calibrate based on 1850 conditions.

[lifang0209]

borpeat_fire_soilmoist_denom=0.3d00
needn't change to 0.35d00

[wwieder]

This gets annual burned area up to 7.8 M km2.
Diagnostics are here, but not loading on the website for some reason?

[wwieder]

This total burned area from the pSASU 163_test3 hillslope case actually compares well with what we're seeing from pSASU out of the flat case in #158 (7.9 vs. 7.4 M km2). Diagnostics are here, but again not working.

I'll try spinning up a hillslope run in #165 that uses: /glade/u/home/fangli/ctsm60_params.genericCropFix.ch4cal_e.fire.c260424.nc

[wwieder]