Impact of Rain Gauge Location Errors on Verification of Radar-Based Precipitation Estimates

The Multi-Radar Multi-Sensor (MRMS) system, which generates a 1-km grid of quantitative precipitation estimates (QPE), can provide insight to forecasters about high resolution precipitation events at a high temporal frequency for flash flood monitoring and detection. To assess the accuracy of the MRMS QPE products, rain gauges are treated as ground truth and can confirm the occurrence of localized intense rainfall rates generated by MRMS in real time operations. One widely used dataset for this purpose is the 838 rain gauges from the Automated Surface Observing System (ASOS) located within the CONUS. It is well known that radar-based QPE values can vary from collocated observed gauge values for a variety of reasons; however, location errors of the rain gauge can also have an impact on the verification of MRMS QPE.

Using Google Earth, the metadata locations of the ASOS gauges were checked against where the station was found in the satellite imagery. ASOS location errors varied from less than 3 m to 80,163 m. When the original and corrected locations were compared to MRMS QPE, 79.31% of ASOS stations in the CONUS were found to be in a different MRMS QPE grid box after the correction. Of those stations, 19.44% were found more than 1 km away from the expected locations. Radar-based QPE values for the new and old locations were compared to the hourly ASOS gauge observations. The results showed that the correlation with the new metadata location increased from 0.777 to 0.810. Additional analysis was performed to assess the impact of the gauge location errors on the MRMS gauge-corrected QPE. These comparisons highlight the need to maintain up-to-date rain gauge metadata to improve the verification of radar-based QPE and other hydrometeorological products.