79 Improving Multisensor Estimation of Heavy-to-Extreme Precipitation via Conditional Bias-Penalized Optimal Estimation

Monday, 8 January 2018
Exhibit Hall 3 (ACC) (Austin, Texas)
Mohammad Nabatian, Univ. of Texas, Arlington, TX; and D. J. Seo, S. Noh, L. Tang, J. Zhang, D. Kitzmiller, and G. Fall

Accurate high-resolution quantitative precipitation estimation (QPE) is of critical importance to NWS’s weather, water and climate services and to supporting the recently launched National Water Model. In this presentation, we describe and evaluate a new technique for merging radar precipitation estimates and rain gauge data to improve multisensor QPE, in particular, of heavy-to-extreme precipitation. Unlike the conventional cokriging methods which minimize error variance only, the proposed technique, referred to as conditional bias-penalized cokriging (CBPCK), minimizes a weighted sum of error variance and expectation of the square of Type-II CB, which arises when failing to detect an existing effect, for improved performance at the tail ends of the distribution of the predictand. To evaluate CBPCK, we carry out cross validation using multiple heavy-to-extreme precipitation events in different parts of the US and visually examine the analysis fields. The CBPCK results are intercompared with the radar-gauge merging algorithm in the NWS’s Multisensor Precipitation Estimator and the radar-only QPE from the Multi-Radar Multi-Sensor (MRMS) system. The CBPCK algorithm is currently being implemented on the MRMS Testbed toward operationalization in the NWS. We present the results, describe the research-to-operations path and share issues and challenges.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner