Add option to use a prescribed snow albedo with Noah-MP

[spencerkclark]

Description

This PR builds on #46 and adds an option to also prescribe the snow albedo used by SHiELD when running with Noah-MP. The existing options, which use schemes to dynamically set the snow albedo, result in an underestimate of the snow albedo, particularly in Antarctica. This new option, which can be toggled by setting gfs_physics_nml.iopt_alb = 3, pipes in the maximum snow albedo read in from climatological fix files and uses that as the snow albedo within Noah-MP. This happens to be the approach taken in SHiELD when using the Noah land surface model with ialb = 1 (see here), so there is precedent.

cc: @kaiyuan-cheng @lharris4 I rebased this branch now that #46 has been merged. This should now be ready for review too.

How Has This Been Tested?

This has been tested by running a one month simulation initialized January 1st, 2020 from GFS analysis with prescribed observed sea surface temperatures and sea ice and computing the surface albedo bias relative to ERA5. We can see that during this season that it makes a particular difference over Antarctica. When coupled with the soil color update in #46 this results in land albedo biases more in line with those of SHiELD when run with Noah—compare the left and right-most panels of the plot below.

This climatological snow albedo is already stored and loaded from restart files by default, so no changes should be needed to ensure restart reproducibility.

Checklist:

Please check all whether they apply or not

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published in downstream modules

[lharris4]

Thank you for assembling this elegant fix.

[kaiyuan-cheng]

Thank @spencerkclark for working on this. The code looks good. However, iopt_alb = 3 does not run well with my C96 test case. I was getting errors like "emitted longwave <0; skin t may be wrong due to inconsistent input of shdfac with lai".

[spencerkclark]

Thanks @kaiyuan-cheng for giving it a try. I submitted a couple year-long runs at C96 resolution with iopt_alb = 3 and found things were able to run stably. Could I take a look at your configuration to see what might be different? Did your case crash immediately or did it take some time for the crash to occur?

[kaiyuan-cheng]

My apologies. I did something wrong when I checkouted the code from this MR. Now I confirm that the MR works fine with my test case.

[spencerkclark]

All good—thanks @kaiyuan-cheng and @lharris4!

[lharris4]

@spencerkclark Thank you for getting this together so quickly.