Tuning Leung_2023 dust cycle for clm6_0_cam7.0

[dmleung]

What is the feature/what would you like to discuss?

(See a short summary in the next message)
This is an issue for round 2 of Leung_2023 dust cycle tuning for CESM3. We completed round 1 one year ago (ESCOMP/CTSM#2803) in CTSM, and we are now attempting to tune the dust in both CAM and CTSM.
@dmleung examined the dust budget for the cesm3_0_alpha07e tag. For climatology, dust is expected to have a global total PM10 emission of ~1.5–3 Pg/yr from CTSM, which generates a global mean dust AOD (AODDUSTdn) of ~0.03, as constrained by observational data (David Ridley et al., 2016). We previously tuned CESM3 dust emissions for CLM6-SP/CAM7 physics to yield a global mean AODDUSTdn of ~0.03 in the 2000s. However, dust PM10 emission is way too high (>5 Pg/yr, see the following Figure). This means that CESM requires extremely high CTSM dust emissions (and CAM dust burdens) to sustain a reasonable global dust AOD in CAM. There are multiple reasons from both CAM and CTSM sides that lead to the low slope/sensitivity of d(dustAOD)/d(dustemis). (I can illustrate more reasons if needed, but two reasons are included in the next paragraph.) However, the current objective is to reduce CTSM dust emission and CAM dust PM burden without significantly altering the current level of dust AOD.
Longlei Li (@L3atm) and Natalie Mahowald have implemented many science and tuning changes in the Cornell CESM sandbox that can help scale down dust emission with little change in CAM dust AOD, which will be beneficial to our tuning. Recent observations and theories suggested that coarse-mode aerosols (like dust and sea spray) are aspherical in shape. Dust asphericity not only (1) increases dust's optical thickness (scattering+absorption) by ~28 % compared with spherical dust (Kok et al., 2017), but also reduces (2) dust's gravitational settling velocity by ~25% (Huang, Kok et al., 2020). These two effects mean that we can reduce ~30–35 % dust emissions/burden to generate roughly the same global dust optical thickness/AOD. (2) also means that dust particles can stay longer in the atmosphere and be transported a bit further away, leading to slightly higher global dust AOD.

@L3atm implemented the original code changes (for Longlei Li et al. 2022). Since CESM assumes spherical aerosols for calculations of aerosol deposition and optics and does not support aspherical aerosols, for now we simply use two tuning factors to represent the dust asphericity effect for scaling the gravitational settling velocity and dust AOD. @dmleung followed @L3atm's implementation and adopted it in the cesm3_0_alpha07e tag. The attached figure below shows a control case (right column) and a test case with modifications (left). Testing with a CLM6-SP/CAM7_LT (FHIST_LTso) case, this successfully scaled down CTSM dust emission (DSTSFMBL = SFdst_a1+SFdst_a2+SFdst_a3 in CAM) from 5.29 Pg/yr to 3.79 Pg/yr while maintaining the global mean AODDUSTdn at ~0.029. This change also reduced surface dust PM10 by > 30 %, which may help CESM/MUSICA air quality modelers (e.g., @rrbuchholz @lkemmons) obtain a slightly more realistic surface dust level. 3.79 Pg/yr for DSTSFMBL is still a higher-than-expected value, and surface dust PM10 is still very high over certain regions (especially the Sahara). This requires future work in CTSM and CAM to tune down dust emission and burden further.

@dlawrenncar asked the CESM3 dust tuning to be done soon. @dmleung will work on the CTSM dust emission change next. I wonder if this change can go in the next CESM alpha tag (e.g., cesm3_0_alpha08) together with the upcoming dust emission tuning inside CTSM with @ekluzek (#XXXX, a placeholder for the CTSM issue).

Relevant cases if any

original FHIST_LTso case with CLM6 physics (SP mode) /CAM7 physics (low top)
/glade/work/dleung/derecho_cases/dust_tuning/251017_tuning_FHIST_LT_SPmode
modified case for testing
/glade/work/dleung/derecho_cases/dust_tuning/251020_tuning_FHIST_LT_SPmode_asphericalDAODdrydep

Is there anyone in particular you want to be part of this conversation?

@fvitt
@tilmes @dlawrenncar @cecilehannay @wwieder @ekluzek @slevis-lmwg (I tag the CTSM people too so they are aware of this)
@L3atm @jfkok

Will this change (regression test) answers?

Yes

Will you be implementing this enhancement yourself?

Yes

[dmleung]

Summary/TLDR:
The dust budget in cesm3_0_alpha07e shows that global total dust PM10 emission (> 5.2 Pg/yr) is too high to generate a reasonable global dust AOD (mean ~ 0.03) in the 2000s. Multimodel agreement of a reasonable global total dust PM10 emission is ~1.5–3 Pg/yr. We need to scale down dust emission while maintaining the current reasonable dust AOD level. (This means we need an enhanced d(dustAOD)/d(dustemis) sensitivity.)
We try to achieve this by scaling down dust gravitational settling by 20 % and enhancing dust extinction (= absorption + scattering) by 28 % for the coarse-mode dust. This is backed by theory and observations, since observed coarse-mode dust is aspherical, which means they are optically thicker and have higher aerodynamic resistance.
Testing these changes in the cesm3_0_alpha07e tag shows that it allows a 30 % reduction in dust emission if keeping global dust AOD at the same level.

[wwieder]

Thanks for this discussion, Danny. Please let us know if there are change we should expect on the CTSM side?

[dmleung]

For this issue no, but I will work on another issue and PR on the CTSM side very soon, thanks