Effect of bias adjustment and rain gauge data quality control on radar rainfall estimation

Thirty major storms that passed over Goodwin Creek, a small research watershed in northern Mississippi, were analyzed to assess the bias between radar rainfall estimates at rain gauge locations and the gauge amounts. These storms, each contributing at least 10 mm of storm total rainfall, accumulated approximately 785 mm of rain, which corresponds to about half the average annual rainfall amount for the area. The focus of this study was to demonstrate the importance of (1) bias adjustment of the radar rainfall estimates and (2) the quality control of the rain gauge data used for bias adjustment. The analyses are based on Memphis Weather Surveillance Radar-1988 Doppler radar data, tipping-bucket rain gauge data, and raindrop spectra information collected within the Goodwin Creek catchment. Because of measurement and rainfall estimation uncertainties, radar observations are often combined with rain gauge data to obtain the most accurate rainfall estimates. Rain gauge data, however, are subject to characteristic error sources: for Goodwin Creek, malfunctioning of the tipping-bucket rain gauges was frequently caused by biological and mechanical fouling, and human interference. Therefore careful quality control of the rain gauge data is crucial, and only good quality rain gauge information should be used for adjusting radar rainfall estimates. By using high-quality gauge data and storm-based bias adjustment, we achieved radar rainfall estimates with root-mean-square errors (RMSE) of approximately 10% for storm total rainfall accumulations of 30 mm or more. Differences resulting from radar data processing scenarios were found to be small compared to the effect caused by bias adjustment and using high-quality rain gauge data.