Energy fluxes do not balance in Greenland

[KeerZ]

Hi All,

I am analyzing simulation results from a case with BSSP585 Compset (but BGC-SP is used) by CESM2.1.2. I output the PFT level results and aggregated (area-weighted) the results from all tree PFTs to compute the results of 'tree land tile'.
Then I checked the energy balance of the tree land tile and rural land unit using this equation: Error=FSA-FIRA-FSH-EFLX_LH_TOT-FGR. I found that the Error can be very large (-20W/m2) in Greenland region and sometimes in Tibet plateau. For other regions, the energy balance error is very small.

Is this energy imbalance common for a fully coupled simulation? If so, I wonder what is the possible reason for the energy imbalance in Greenland? I guess this has something to do with the land ice model but I cannot figure out why...

Any help would be appreciated! Thanks!

 

[oleson]

Such an imbalance of that magnitude should not occur for a fully coupled (or any) simulation.
The energy balance check at the pft level (for non-urban patches) is:

errseb(p) = sabv(p) + sabg_chk(p) + forc_lwrad(c) - eflx_lwrad_out(p) &
; - eflx_sh_tot(p) - eflx_lh_tot(p) - eflx_soil_grnd(p)

and we know this balances to within 1.e-7 at each time step (the error check threshold) (see ERRSEB in CLM history file).

I'm not sure exactly how you are calculating the energy balance for rural, but I would think that this equation for rural should balance (with one exception):

energy_balance_r = FSA_R - FIRA_R - EFLX_LH_TOT_R - FSH_R -FGR_R

The exception I have found in my own work is that the sabg_chk term in the equation for errseb can be slightly different than sabg in some cases. FSA is reporting sabg + sabv, not sabg_chk + sabv.
So, energy_balance_r may not balance at a few gridcells unless sabg_chk is used for SABG. I think we should be using sabg_chk instead but I haven't yet pursued this robustly.
But this should not be causing the large imbalance you are seeing over the coast of Greenland. One think I can think of is that the incoming longwave from the atmosphere is downscaled over glacier columns and sometimes over other columns depending on location. See here for a bit of an explanation:

Problem With LW Downscaling in CLM5

I am experiencing a potential bug while running both CESM2 and standalone CLM5. In the case of standalone CLM5, I am running with a 1.1 km grid with topography at that resolution (and all other fields interpolated) in parts of the Andes, East Africa etc. Downscaling of fluxes and ground...

bb.cgd.ucar.edu

So maybe there is some inconsistency in how you are calculating net longwave for rural (or trees) with regard to downscaled versus not downscaled incoming longwave, or maybe net longwave.
Not sure if this helps. Let us know if you figure it out.

 

[KeerZ]

Such an imbalance of that magnitude should not occur for a fully coupled (or any) simulation.
The energy balance check at the pft level (for non-urban patches) is:

errseb(p) = sabv(p) + sabg_chk(p) + forc_lwrad(c) - eflx_lwrad_out(p) &
; - eflx_sh_tot(p) - eflx_lh_tot(p) - eflx_soil_grnd(p)

and we know this balances to within 1.e-7 at each time step (the error check threshold) (see ERRSEB in CLM history file).

I'm not sure exactly how you are calculating the energy balance for rural, but I would think that this equation for rural should balance (with one exception):

energy_balance_r = FSA_R - FIRA_R - EFLX_LH_TOT_R - FSH_R -FGR_R

The exception I have found in my own work is that the sabg_chk term in the equation for errseb can be slightly different than sabg in some cases. FSA is reporting sabg + sabv, not sabg_chk + sabv.
So, energy_balance_r may not balance at a few gridcells unless sabg_chk is used for SABG. I think we should be using sabg_chk instead but I haven't yet pursued this robustly.
But this should not be causing the large imbalance you are seeing over the coast of Greenland. One think I can think of is that the incoming longwave from the atmosphere is downscaled over glacier columns and sometimes over other columns depending on location. See here for a bit of an explanation:

Problem With LW Downscaling in CLM5

I am experiencing a potential bug while running both CESM2 and standalone CLM5. In the case of standalone CLM5, I am running with a 1.1 km grid with topography at that resolution (and all other fields interpolated) in parts of the Andes, East Africa etc. Downscaling of fluxes and ground...

bb.cgd.ucar.edu

So maybe there is some inconsistency in how you are calculating net longwave for rural (or trees) with regard to downscaled versus not downscaled incoming longwave, or maybe net longwave.
Not sure if this helps. Let us know if you figure it out.

It is very helpful. Thanks, Keith!

I found that it's because I used the grid-mean FLDS as downwelling longwave radiation when I calculate the net radiation (I did not use FSA-FIRA somehow). I thought FIRA_R always equals FIRE_R minus grid mean FLDS because I was not aware that the longwave radiation is downscaled in CISM domain. The energy balance error becomes very small when I use FSA_R-FIRA_R as net radiation or when I use column-level FLDS to compute the actual downwelling longwave radiation at the rural/tree land tiles.