Presentation PDF (570.4 kB)
Specifically, the analysis variable for temperature has been changed from potential to sensible temperature, and the Cressman-based analysis error has been replaced with a Lanczos version. The quality control has been improved by adding a global blacklist for stations with unreliable observations, as determined by MADIS statistics, as well as local blacklists provided by NWS field offices. The smoothing of the terrain used to build the background error covariances has been reduced to allow the analysis increments to follow the terrain more closely. Finally, the quality control for moisture observations has been modified to better handle dryline events.
These changes will be described in detail in the presentation, and cases to demonstrate the changes will be shown. Future plans to improve the RTMA will also be discussed.