Changes in aerosol effective radiation forcing from CESM2 to CESM3_development. Short wave or long wave?

[oyvindseland]

Hi

Although NorESM has a different aerosol parameterisation than CESM, the numbers for ERF have been fairly close for released versions. The same seems to hold true for NorESM3-development / CESM3 development as both are decreasing compared to NorESM2 / CESM2. My latest estimate for NorESM3-development version is -0.7 W m-2 and if I have not misunderstood the same reduction is also found in CESM3. However when I look more into the details it seem that the reason for change is not some much a reduction in cloud short wave effect but a quite large positive LW cloud effect. Is this the case also for the CESM3 development version?

What I also find interesting is that the increase in LWCF is less local to the aerosol regions than the more negative SWCF so it looks more like a circulation response than impact from the aerosols themselves. In particular there seems to be a southward shift in Atlantic and Eastern Pacific ITCZ even without the change in SST pattern as one typically finds in a coupled simulation with increased NH aerosols. As in CAM we do not in general underestimate / overestimate the precipitation in Amazon / West Africa in uncoupled simulations but the impact of the added aerosol is to increase the precipitation in the Amazon region. Do you find the same for CESM3.

[adamrher]

Hi @oyvindseland

I haven't look in detail at the ERF breakdown, but our latest estimate is about -0.7 W/m2. I just calculated the breakdown of cloudy and clear-sky components from 2x10 year runs as:

I'm not seeing a very large LWCF effect. How large is it for you?

@brianpm @cmcclus @islasimpson have you looked at all at the precipitation differences and/or changes to the circulation?

Run details are here: NCAR/amwg_dev#641

[islasimpson]

Is the question about the precip or circulation anomalies in the F1850 with imposed aerosol forcing? If so, I haven't looked at that, but I'd be happy to if someone can point me to the run.

[adamrher]

I'm not sure what SST @oyvindseland is using for the ERF runs, but we ran with PD SSTs (2000-2009).

[oyvindseland]

Thank you for the information.

I get much higher numbers except for net: dRESTOM -0.42; dSWCF -1.41; dLWCF 0.93; dFSNTC -0.17; dFLNTC=-0.22

Now granted I use a different aerosol formulation although the difference has not been that large before. Also I use 2014 -1850 aerosols but that should not matter too much.
Still using Zender dust parameterisation though since Fates (embedded in CLM) is still not set up for Leung. May have a larger impact, but still nothing. I have checked that I use the same SST in both simulations, btw.

I plan to run ERF for individual compounds next (Sulfur; BC; OM). I hope that this can give me some ideas as the dLWCF should not be anywhere near what I am getting.

[oyvindseland]

I should probably need to check the implementation of hetfreez as well, in case BC has received more weight than it should. Thank you for providing your ERF numbers.

[adamrher]

Here are our current scale factors for dust and bc:

<hetfrz_bc_scalfac>0.01D0</hetfrz_bc_scalfac>
<hetfrz_dust_scalfac>0.05D0</hetfrz_dust_scalfac>

[oyvindseland]

The SST field is a PI estimate, but I do not think that is critical (sst_HadOIBl_bc_1x1_clim_pi_c101029.nc)
We still read in forest fire emissions though, not online. That may make a difference at least in short wave.
Estimating ERF from prognostic forest fire may perhaps be a bit problematic. Given the same SST, GHG concentration and land use, emissions from prognostic forest fire emissions will be more or less the same in the twinned ERF simulation?