Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Convergence Issues in Regional MPAS-A

ragy

New member
Hi there,

We're running a test simulation of the regional MPAS-A focused on South America, using intermediate data processed with the cmip6-to-wrfinterm tool available at cmip6-to-wrfinterm/ReadMe.md at master · lzhenn/cmip6-to-wrfinterm (with minor modifications to include the variables needed for MPAS). We used this tool to convert CMIP6 data to the WRF format, and it worked perfectly; however, when using the same intermediate data in MPAS-A with physical parameterizations enabled, the model does not run correctly.

The mesh was cropped from the initial mesh 60-15 km mesh (535554 horizontal grids).

Although the .init.nc and lbc. files are generated without errors, during the atmospheric simulation the "w" values eventually become NaN, stopping the simulation without a clear error message. When running only the dynamic component, everything works fine; the issue arises only when parameterizations are activated.

We appreciate any guidance that can help us resolve this issue. Attached are the logs for your review.
 

Attachments

  • log.atmosphere.0000.out.txt
    36.4 KB · Views: 0
  • log.atmosphere.0000_none.out.txt
    151.7 KB · Views: 0
  • namelist.amosphere.txt
    1.7 KB · Views: 1
I am suspicious that this issue is caused by the mismatch of landmask between your CMIP6 and MPAS static data. We did see the same 'w' NaN feature when landmask in the forcing data is inconsistent with that in MPAS static data.

I would suggest that you recompile MPAS in debug mode, then rerun this case. You will be able to find when and where the NaN value first pops up, and check all related variables at that specific point.

Please keep me updated about your progress. Thanks.
 
Top