Comparison of automated and manual quality control techniques for surface observations over complex terrain

The Real-Time Mesoscale Analysis (RTMA) is the first step in a multi-year effort to develop an “Analysis of Record”. The project is motivated by a need across the National Weather Service (NWS) for a high resolution surface objective analysis to support forecast operations. The RTMA is created through a 2DVar analysis of thousands of surface observations collected by the Meteorological Assimilation Data Ingest System (MADIS).

As part of the RTMA quality control procedures, observations are withheld from the analysis, if they are included on one of two reject lists. The first list is based on MADIS quality control flags that are generated using automated methods (i.e., validity, internal consistency, temporal consistency, statistical spatial consistency, and spatial “buddy” checks). The list is generated monthly and observations are flagged, if they fail MADIS automated checks at least 25% of the time during a given month. The second list contains observations that are reported by Weather Forecast Offices (WFO) on a quarterly basis. These lists are subjectively generated based on forecaster input from looking at local objective analyses and surface observational data.

This poster will compare the quality control reject lists created by automated methods to those subjectively generated by NWS forecast offices between January and June 2008. The comparison will be broken down by WFO (in NWS Western Region) and by major reporting networks. Local WFO quality control procedures will also be analyzed.