-
Notifications
You must be signed in to change notification settings - Fork 244
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Big RMSE of 2 meter temperature #2116
Comments
@YongjunZHENG Can you provide more information? Which application is the result from, model version, coupled runs or atm-only runs, resolution, start time model version, etc? Thanks |
Could this issue be related to this one: ufs-community/ufs-srweather-app#1004 ? E.g. running the GFSv17 suite in the "short range weather app"? If so, I suggest discussing it in that context as it must be related to some configurational setting related to that specific application. @YongjunZHENG could you provide more information? |
lisa-bengtsson,
Thank you for your reply. Here is some infomation which may be useful to figure out the incorrect configuration in my setting.
1. I cloned the latest version of the UFS Weather Model in August of 2022:
git clone --recursive https://github.com/ufs-community/ufs-weather-model.git
2. Succefully built the standalone ATM:
export CMAKE_FLAGS="-DAPP=ATM -DCCPP_SUITES=FV3_GFS_v16"
3. And run the UFS model for global medium range weather forecasts without any problems..
4. When I evaluated the performance of UFS forecast by comparing with EC forecast, the RMSE and correlation of precipitation and 10 meter wind are good, but the RMSE of 2 meter temperature is significantly larger than that of EC as shown in the previous figure.
5. I enclosed two configurational files: input.nml and suite_FV3_GFS_V16.xml
Best regards,
Yongjun Zheng
At 2024-01-25 23:08:12, "lisa-bengtsson" ***@***.***> wrote:
Could this issue be related to this one: ufs-community/ufs-srweather-app#1004 ?
E.g. running the GFSv17 suite in the "short range weather app"? If so, I suggest discussing it in that context as it must be related to some configurational setting related to that specific application. @YongjunZHENG could you provide more information?
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
I am not familar with how to use the github issue editor. So I uploaded two configurational files (input.nml and suite_FV3_GFS_V16.xml) in Email. Sorry for the inconvenience.
At 2024-01-25 23:08:12, "lisa-bengtsson" ***@***.***> wrote:
Could this issue be related to this one: ufs-community/ufs-srweather-app#1004 ?
E.g. running the GFSv17 suite in the "short range weather app"? If so, I suggest discussing it in that context as it must be related to some configurational setting related to that specific application. @YongjunZHENG could you provide more information?
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Updates committed to the latest ufs-weather-model main repository have not always been tested for the GFSv16 physics package (FV3_GFS_v16). Please instead use suite_FV3_GFS_v17_p8.xml. If you do want to run GFSv16, please check out this tagged version https://github.com/ufs-community/ufs-weather-model/tree/GFS.v16.3.1. It has been known that GFS T2m biases are usually larger than that of IFS/ECMWF, due to both DA and model physics differences. Improvements have been made in both fronts in recent years. They will be included in the next GFS implementation. |
@YongjunZHENG I'm also a bit curious how you obtained the input.nml file? There are a couple things that need to match for a given model configuration. E.g. the source code (could be a tag), a namelist, the field_table (table of prognostic tracers) and initial conditions. If you run the model outside of a workflow that provides this, you may see unexpected results. |
@lisa-bengtsson
I obtained the input.nml from a FV3_GFS testcase and made some appropriate changes.
I enclosed the field_table for your reference.
I used chgres to make the initial condition from the NCEP analysis. For example, I downloaded the following two files:
https://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/gfs.20240128/12/atmos/gfs.t12z.atmanl.nc
and
https://nomads.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/gfs.20240128/12/atmos/gfs.t12z.sfcanl.nc
@fanglin Yang
I will try suite_FV3_GFS_v17_p8.xml with the latest ufs-weather-model.
Could you please present me the configurational files (input.nml, field_table, and suite_FV3_GFS_v17_p8.xml) which are consistent with the latest ufs-weather-model or a stable version/tag? I would like to test whether it works.
Thank you so much!
Yongjun Zheng
At 2024-01-27 06:01:29, "lisa-bengtsson" ***@***.***> wrote:
@YongjunZHENG I'm also a bit curious how you obtained the input.nml file?
There are a couple things that need to match for a given model configuration. E.g. the source code (could be a tag), a namelist, the field_table (table of prognostic tracers) and initial conditions. If you run the model outside of a workflow that provides this, you may see unexpected results.
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
Description
The text was updated successfully, but these errors were encountered: