The development and utility of a database of mesonet wind observations for use in the RTMA system

NCEP's Real Time Mesoscale Analysis (RTMA) is designed in part to provide a better representation of mesoscale phenomena, to serve as a verification of the NWS gridded forecast database (NDFD), and to create an analysis of record (AOR). Mesonet winds continue to present a quality control challenge within the RTMA. At present, a majority of wind observations from mesonet sites are excluded from the analysis due to a perceived low bias ostensibly from poor siting. Hence, RTMA wind analyses reflect only the somewhat sparsely populated METAR sites and other official observations (e.g., RAWS). As a result, a regional quality control/quality assurance (QC/QA) system is developed to create a larger list of acceptable observation sites for use within the RTMA. The number of observations identified as 'acceptable' have increased and their impact on the RTMA are being evaluated.

The new system uses a mySQL database that is populated by mesonet wind observations from which relevant statistics are calculated. Observation sites over a specific region (Florida) and time period (1 August 2008 – 31 July 2009) are examined. QC checks are run within the database by using simple SQL queries and a flagging system. Initial QC checks are primarily based on comparing mesonet site observations with known high-quality observations (i.e. METAR sites) and the first guess field.

The primary goals of this study are twofold: to create an expanded static list of mesonet sites to be used in the RTMA, and to create a dynamic list that changes based on observed wind direction. Observed wind speed biases (with respect to the background field) in 8 directional bins are compared with those of nearby METAR sites. Sites with similar bias to that of nearby METAR sites are identified as acceptable. This dynamic list represents a new tool that can be used to increase the number of quality observations in the RTMA.

The benefits of using a database approach for quality control purposes are demonstrated. The system can be applied in a computationally inexpensive and fairly straightforward manner. The database system allows for multiple QC/QA checks to be performed in quick succession, and could easily be adapted for use in other assimilation systems and thus could easily be adapted to run in real time.