15B.4 Improving Multi-Radar Multi-Sensor (MRMS) Precipitation Estimates for Orographically Enhanced Rainfall in Hawaii and the Western United States

Thursday, 16 January 2020: 4:15 PM
253A (Boston Convention and Exhibition Center)
Andrew P. Osborne, CIMMS/Univ. of Oklahoma and NOAA/NSSL, Norman, OK; and J. Zhang, S. B. Cocks, M. J. Simpson, and K. W. Howard

High accuracy precipitation products are an important source of information for flash flood forecasting, water resource management, and climatological applications. The Multi-Radar Multi-Sensor (MRMS) radar-based Quantitative Precipitation Estimation (QPE) has provided government agencies, academia and private sectors with high-resolution national precipitation data for various applications and the QPE accuracy varies with space and time. There is a challenge for the radar QPE to accurately capture the rainfall amounts seen in atmospheric river events in the western US and tropical precipitation in Hawaii. The uplift provided by the mountainous terrain in these areas tends to enhance precipitation rates. The high terrain also causes large radar blockages such that the important lower-level processes in this type of high-efficiency rainfall are not adequately captured. This work explores the potential utility of both physical and statistical based approaches for improving the underestimations often seen from the MRMS products in this scenario. Several physical based approaches are explored including a vertical rainfall correction, the use of more aggressive rainfall rate relations, and the consideration of environmental variables such as a wind-slope interaction parameter to adjust the QPE output. The statistical study can also help identify important variables for a machine learning based radar QPE model for these precipitation regimes. Case studies from several atmospheric river and tropical rain events will be presented along with evaluations of each technique within the MRMS system.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner