Burned area in CESM3 & CLM6

[wwieder]

Keith just put together these diagnostics that compare our CESM2 PI control to a longer CESM3 beta simulation (#127) and I noticed that the burned area was more than 2x larger, largely in tropical savannas, but also in agricultural areas (map below).  
We're having a number of issues with the coupled model climate that I won't go into here, but I also wondered if we may end up needing to tune the fire model for the CESM3 climate?

Previous coupled experiments in #82 didn't show much of a difference in burned area, but this was before we improved the survival of C4 grasses by relaxing N limitation in more recent CTSM5.3 parameter sets e.g. #89 & #91. See also #90.  That said, improving C4 grass survival didn't have much / any increase in burned area in land only runs based on these ILAMB results by Sam L.

I just wanted to put this on our radars as we get closer to parameterizing CESM3 & CLM6.

[samsrabin]

For what it's worth, the new burned fraction in tropical savannas—or at least, Africa and northern Australia—seems much more realistic. Here's the mean annual observed burned fraction for 2001–2020 from GFED5:

As you suggested, this may well be due to the increase in fuel due to higher survivability of C4 grass. Its effects might be modulated in the historical period (as in runs #89 and #91) due to the landscape fragmentation effect that came in with the fire model updates, as well as by the pre-existing population density effects.

I'm not sure what could be causing the increase in agricultural areas—crops are on during run 127, right? The fire model updates made it so crop burns wouldn't happen during the growing season, but I don't think that should have changed the total amount of burned crop area.

[wwieder]

Thanks Sam. How difficult may it be to include a CUPiD notebook that includes evaluation with GFED5? Alternatively, I wonder if ILAMB is considering including GFED5 in upcoming releases?

[samsrabin]

I wouldn't be surprised if that's in the ILAMB plan, but I don't know anything about it. Adding a CUPiD notebook would be pretty easy, and I can plan to do that—what sort of timeline were you thinking?

[lifang0209]

The GFED5 burned area is twice that of GFED4s. Since I calibrated the new fire model using GFED5, I am not surprised that the simulated burned areas for global regions, tropical savannas, and croplands in CESM3 are much higher than in CESM2.
Additionally, the global cropland burned area has been recalibrated in CTSM5.3 based on GFED5 (83 Mha/yr) instead of the 18 Mha/yr used in CLM5. I set crop fires occur outside the growing season, so I don't think they will significantly impact crop simulations.
Should we recommend that ILAMB use GFED5 to replace GFED4s?

[dlawrenncar]

Do we have a CESM3 beta simulation comparison with a land-only run? I think that is the bigger question right now, if the increase in burned area is intentional to better match GFED5. Also, has anyone looked at carbon emissions from fire? I can push to get GFED5 into ILAMB.

…

On Tue, Mar 25, 2025 at 7:54 AM lifang0209 ***@***.***> wrote: The GFED5 burned area is twice that of GFED4s. Since I calibrated the new fire model using GFED5, I am not surprised that the simulated burned areas for global regions, tropical savannas, and croplands in CESM3 are much higher than in CESM2. Additionally, the global cropland burned area has been recalibrated in CTSM5.3 based on GFED5 (83 Mha/yr) instead of the 18 Mha/yr used in CLM5. I set crop fires occur outside the growing season, so I don't think they will significantly impact crop simulations. Should we recommend that ILAMB use GFED5 to replace GFED4s? — Reply to this email directly, view it on GitHub <#92 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFABYVDJTSU7R64ZLOGCVR32WFNZLAVCNFSM6AAAAABZVIQED2VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTENRRGU3DSOA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[olyson]

I will run diagnostics for the last 30 years of 138 compared to the last 30 years of the land-only pSASU part of #91 to help address the burned area question.

[olyson]

Land diagnostics for this are here
If you need seasonal/annual climatologies for plotting purposes, they are here:

/glade/derecho/scratch/oleson/ANALYSIS/climo/b.e30_alpha06b.B1850C_LTso.ne30_t232_wgx3.138_33_62
/glade/derecho/scratch/oleson/ANALYSIS/climo/ctsm53026_BNF_pSASU_131_160

[olyson]

Regarding the question yesterday about this:

COL_FIRE_CLOSS(PgC/y) total column-level fire C loss 3.961 3.564
PFT_FIRE_CLOSS(PgC/y) total pft-level fire C loss 2.809 2.551

Looking at the code, the reason why they are different, is that COL_FIRE_CLOSS is the patch carbon losses to fire (PFT_FIRE_CLOSS), plus the "vertically-integrated (diagnostic) decomposing C fire loss (gC/m2/s)".

[wwieder]

Thanks Keith, this suggests we're burning off litter and CWD too.

[lifang0209]

COL_FIRE_CLOSS represents the total carbon loss from fire
COL_FIRE_CLOSS = PFT_FIRE_CLOSS (burning of vegetation tissues) + burning of litter and CWD