Composite-weighted Precipitation Estimation based on Multiple Polarimetric Estimation Methods Using the WSR-88D Radar Data over Iowa

For several decades, weather radar has been used as a primary observational instrument for weather monitoring and forecasting over large areas. The major goals of weather radar applications in hydrology have been to improve the accuracy of quantitative precipitation estimation (QPE) and subsequent streamflow prediction for efficient flood management. The polarimetric upgrade of U.S. Weather Surveillance Radar-1988 Doppler (WSR-88D) radars has provided a myriad of opportunities to improve QPE accuracy by addressing several challenges associated with data quality control, melting layer detection, and hydrometeor classification. However, it is still not well understood which polarimetric estimation method performs best. In this study, we use several polarimetric estimators and evaluate their estimation performance. We acquire data from several WSR-88D radars that cover the Iowa domain and generate multiple QPE products by applying specific attenuation-, specific differential phase-, and reflectivity and differential reflectively-based estimators. We then use the data from a variety of rain gauge networks in Iowa as ground reference. The evaluation analysis demonstrates each polorimetric estimator's strengths and weaknesses compared to the conventional estimator using a power-law relationship between radar reflectivity and rain rate. Because a threshold-based composite method often leads to discontinuity, we explore a weighted linear combination of multiple estimators by accounting for the error structure of each estimator. We also discuss estimation improvements accomplished by the proposed combination method against the cases where a single estimator is applied.