P85 An analysis of the sensitivity of precipitation forecasts to microphysical parameterization and grid-resolution at convective-allowing and cloud-allowing scales

Lynn, Barry H., The Hebrew University of Jerusalem, Israel, Ruby Leung, Pacific Northwest National Laboratory, and Amir Givati, Hydrological Service, Israel

Twelve “convection-allowing” and “cloud-resolving” deterministic and ensemble precipitation forecasts were compared during the 2010-11 winter rainy season over Israel. The WRF (bulk-parameterization) Single Moment Scheme (WSM6) and Thompson Microphysics (Thompson) were used in a set of deterministic and ensemble forecasts, while Spectral (bin) Microphysics provided a baseline reference. Comparison of probability forecasts for specific threshold values were made for observing stations in five different topographical zones in Israel. The Thompson scheme produced very small biases in the accumulated precipitation over all regions, and using it makes a significant improvement in forecast Fraction Skill Scores (FSS) for high (daily) threshold values. Comparing the Thompson and WSM6 forecasts to the SBM implies that changes to the Thompson scheme over several years have led to improvements in this scheme relative to the WSM6. The FSS values of both the Thompson and SBM were significantly better than with the older WSM6 scheme. Moreover, the analysis indicated that cloud-resolving ensemble forecasts can add value to forecast skill when compared to the 4 km ensemble, but only when the Thompson scheme was used instead of WSM6. An addition reanalysis forecast for January 2013 confirmed these results, and also revealed the importance of melting processes on accurate precipitation forecasts.