Positive RESTOM bias in the latest CESM tag

[adamrher]

Here's a brief history of how we ended up with a positive RESTOM bias in the latest tag. Run 100 was the first bug free version of RRTMGP, and RESTOM is in balance. We then took a recommendation from @tilmes to lower the seasalt_emis_fact = 1.5->0.75 (low ss emis), which increased clear sky SW absorption due to less scattering from aerosols. To compensate for this now positive RESTOM, I thickened the low clouds by increasing clubb_c8=4.25->4.95 (high c8), which balanced RESTOM in run 104.

The next run we did was 112, which changed the code base from beta02 -> alpha03d. It has new MOM6 defaults, new sea-ice physics, and I believe this is the tag in that ctsm5.3 came in (and its new params file). RESTOM is still balanced as in 104, despite these changes. We then did run 114, which is the same atm params as 112, but out of the box everything else. The SSTs are cooler in this run, and I think something did change in the ocean from 112->114 (@gustavo-marques).

@megandevlan did an analysis of ENSO in 104/112 and found unrealistically low variability, which we hypothesized was due to cooler equatorial SSTs. Based on the runs 100-104, the cooler SSTs were largely the result of using a high value of c8, and so in the next run 116, we did a partial revert back to run 100 atm params (ss emis=1.0, c8=4.4), plus we brought in the new dust model (Leung_2023 + danny fact: dust_emis_fact=2.3).

Part of the reason we can't get back to a balanced RESTOM must be from the only partial revert back to run 100 and possibly the new dust scheme. Increasing the ss emis = 1.0->1.5 should reduce FSNTC by ~0.5 W/m2, and this may be further reduced by reverting to the old dust scheme, or by reducing dust_emis_fact =2.3 -> ? in the new scheme. Lowering c8 to 4.25 should warm up the SSTs and increase OLR / FLNTC, but my experience is that it will reduce low clouds and the magnitude of SWCF even more, ultimately leading to a more positive RESTOM. I would suggest doing an experiment increasing c8 slightly = 4.4->4.5, and seasalt_emis_fact=1.0->1.5. This is the same as run 118, but the ocean mixing scheme is a deal breaker in that run. We could consider lowering the dust_emis_fact to bring down RESTOM further. And heck, perhaps the full c8 revert -- a small reduction in c8=4.4->4.25 -- would cause OLR to increase more than the weakening of SWCF, so that could be another thing to try.

@PeterHjortLauritzen @JulioTBacmeister @cecilehannay @dlawrenncar

[adamrher]

The positive RESTOM bias in the latest tag has mostly been resolved, primarily through reverting the ocean mods introduced after run 112. The latest run (121) has a positive RESTOM of +0.16 W/m2, which is similar to the CESM2 B1850 RESTOM of +0.13 W/m2 (see below). The CESM2 case refers to the first 100 years of the output in the CMIP6 repo. Run 121 SSTs are notably cooler than CESM2, although it is showing a warming trend in final 50 years of the run. The run is continuing and so we'll see if SSTs continues to warm.

Here are the components of the top of model energy budget. Clear-sky absorption also seems to be increasing after yr 50 in run121.