You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The online domain check filter has been tested for atms_npp, atms_n20, and amsua_n19. The yaml file for applying domain check for satellite radiance is as bellow:
- filter: Bounds Check
filter variables:
- name: brightnessTemperature
channels: *atms_npp_channels
test variables:
- name: GeoVaLs/observable_domain_mask
flag all filter variables if any test variable is out of bounds: true
minvalue: 0.0
maxvalue: 0.5
Example: atms_n20 Before QC:
After QC:
In the above figures, the dark blue color represent the data kept after domain check. So the data points outside of the domain are still kept in hofx file and filled with missing value, instead of completely removed. The previous 'circle domain check' filter does not remove the data outside of the domain either. With the filtered data kept in the memory, the size of output hofx file become very large. We need to address this problem with a long-term solution considering the operational implementation.
The text was updated successfully, but these errors were encountered:
This PR is going to merge the ATMS NOAA_20 JEDI yaml file into RDASApp validated yaml file templates.
As the test in issue #203, atms_n20 data has been tested in MPAS_JEDI including all QC filters and bias correction. Not only the ioda data converted from regular feed bufr data has been tested, but also the ioda data converted from combined regular feed, DBNet, and ESRs bufr data has been tested too.
In addition, this atms_n20 yaml file has included the online domain check filters as described in and issue #224, and removed the 'circle domain check' filter
The online domain check filter has been tested for atms_npp, atms_n20, and amsua_n19. The yaml file for applying domain check for satellite radiance is as bellow:
Example: atms_n20

Before QC:
After QC:

In the above figures, the dark blue color represent the data kept after domain check. So the data points outside of the domain are still kept in hofx file and filled with missing value, instead of completely removed. The previous 'circle domain check' filter does not remove the data outside of the domain either. With the filtered data kept in the memory, the size of output hofx file become very large. We need to address this problem with a long-term solution considering the operational implementation.
The text was updated successfully, but these errors were encountered: