Question Regarding Surface Temperature Bias in Icepack Standalone v1.5.3

[Ajay Bhadran]

Dear Icepack Consortium Team,

I am currently running the latest standalone version of Icepack (v1.5.3) using ERA5 atmospheric forcing data. I have attached my icepack_in configuration file, forcing file (ERADATA.txt), and the model run log for reference.

While analyzing the simulations, I noticed that the model-simulated surface temperature is consistently colder than the observed/forcing atmospheric temperature. I would like to clarify whether this behavior is physically acceptable in standalone Icepack simulations or whether it may indicate an issue in my setup or forcing configuration.

I am particularly concerned because this surface temperature bias could affect the surface energy balance, especially the sensible and latent heat flux calculations. Since these fluxes depend strongly on the temperature gradient between the surface and atmosphere, I suspect that the colder surface temperature may be enhancing sublimation and contributing to an underestimation of snow depth in my simulations.

In my runs, the model is not able to reproduce the observed snow depth evolution well, and sublimation appears relatively high. I am unsure whether this is:

1. A normal limitation or expected behavior in standalone Icepack simulations,
2. Related to uncertainties or biases in the atmospheric forcing,
3. Associated with parameter choices in the thermodynamic scheme, or
4. Indicative of a mistake in my setup or forcing preparation.

Could you please clarify whether this type of surface temperature bias is expected or whether I may be missing something important in the model configuration or forcing data processing?

Thank you very much for your time and guidance.

Best regards,
Ajay Bhadran

 

[dbailey]

Are you using your own customized ERA forcing? What is the latitude and longitude of the point you are looking at? There are several tunable parameters within Icepack to make it look more like observations. However, one cannot expect this to be an apples to apples comparison.

 

[mansi]

Hi, I just went through your input files.
I noticed that atm_data_type = 'CFS'. Have you added a subroutine for ERA5? Otherwise, the model may still be reading the forcing as CFS or possibly falling back to a default setting, which could explain the colder temperatures you’re seeing. Let me know if you modified icedrv_forcing.F90 to read ERA5 data or changed the forcing input logic in any way.

 

[Ajay Bhadran]

Are you using your own customized ERA forcing? What is the latitude and longitude of the point you are looking at? There are several tunable parameters within Icepack to make it look more like observations. However, one cannot expect this to be an apples to apples comparison.

Thank you for your response.

I am currently using ERA5 atmospheric data converted into the prescribed CFS-style forcing format for the standalone Icepack driver, without modifying the default tunable parameters at this stage.

My study region is in East Antarctica near Prydz Bay (69.24°S, 76.12°E).

I would appreciate your suggestion on which sections or parameters should be the primary focus while investigating the surface temperature bias in the standalone simulation.

Thank you again for your guidance.

 

[Ajay Bhadran]

Hi, I just went through your input files.
I noticed that atm_data_type = 'CFS'. Have you added a subroutine for ERA5? Otherwise, the model may still be reading the forcing as CFS or possibly falling back to a default setting, which could explain the colder temperatures you’re seeing. Let me know if you modified icedrv_forcing.F90 to read ERA5 data or changed the forcing input logic in any way.

Thank you for checking the input files.

At this stage, I have not added a separate ERA5 subroutine or modified icedrv_forcing.F90. I am currently using ERA5 data converted into the prescribed CFS-style forcing format while keeping atm_data_type = 'CFS'.

Also, if possible, could you please suggest which tunable parameters or sections would be most important to focus on for improving the standalone simulation and reducing the surface temperature bias?

Thank you again for your guidance.

 

[mansi]

I think you might need to change the icedrv_forcing.F90 for the forcing. I can send you the ERA subroutine that I added for my work , it’s working on my end, so you can try using that. I just don’t have the files handy right now, but I should be able to send the instructions in a day or so, if that works for you.
In the meantime, you might want to try modifying the read function in the CFS section of icedrv_forcing.F90, that could also help.
On line 638:
read (nu_forcing, '(6(f10.5,1x),2(f10.8,1x))')
This format works for CFS, but ERA does not read it correctly.