Real-time Quality Control of 1-hour Rain Gauge Data Using a Multi-Senor Approach

Real-time quality control (QC) of rain gauge data is among the most challenging aspects of developing accurate, timely, high-resolution spatial rainfall data products, such as gauge-adjusted radar precipitation. Gauges can suffer from a number of different problems including freezing precipitation, windy conditions, gauge siting (e.g., obstructions around the gauge), under-measurement by tipping bucket gauges in high intensity rainfall, and gauge maintenance. Furthermore, persistent problems cannot be easily identified over short periods of time (e.g. 1 hour), but can grow into significant sampling errors over longer-durations (e.g. 30 days or more).

In an effort to mitigate many of the challenges of real-time rainfall gauge QC, we developed innovative and effective algorithms to QC gauge data based on several variables. The QC algorithms operate in real-time on over 12,000 gauges across the United States and adjacent portions of Canada and Mexico from sources such as MesoWest, MADIS and the USGS. The 1-hour rainfall measurement QC is based on surrounding gauges, QC'd radar data and the National Weather Service Stage IV gauge-adjusted radar-estimated precipitation data. The complexities of gauge QC often prevent a binary (correct or not correct) decision to be made, therefore our QC algorithms provide a QC confidence flag to each 1-hour gauge value. The QC confidence flag ranges from 0 to 1, where 1 means the value is likely correct while 0 means the value is likely incorrect. The resulting rainfall data and QC flag can be used in a variety of hydrologic applications, including gauge-adjusted radar-estimated precipitation algorithms, identification of gauge malfunctions, alert notifications of rainfall intensities that meet/exceed thresholds, detailed post-event analyses and climatological studies. This presentation will provide an overview of the QC process, its output and impact on spatial precipitation data products.