P78 Object-based verification of regional WRF simulations of summertime convection

Starzec, Mariusz, Gretchen Mullendore, David Delene, University of North Dakota, Paul A. Kucera, National Center for Atmospheric Research, and Darin Langerud, North Dakota Atmospheric Resource Board

The Polarimetric Cloud Analysis and Seeding Test (POLCAST) is an ongoing research project in eastern North Dakota focused on evaluating the effectiveness of hygroscopic seeding flares in summertime convection. Since 2010, local 3-km resolution Weather Research and Forecasting (WRF) model runs were used to predict the timing, intensity, and distribution of convection in the study region. Forecasts were spatially verified by utilizing the Model Evaluation Toolkit (MET) Method for Object-based Diagnostic Evaluation (MODE) tool to compare the simulated reflectivity field to the radar reflectivity field at a 1-km height as observed by the University of North Dakota’s polarimetric C-band Doppler radar. Traditional skill scores applied to the matched objects were low even in cases that showed good subjective agreement, so alternative methods for computing forecast skill were developed. Results from two summers of 3-km runs showed that approximately 75% of cases occurred within 10% areal coverage difference when both forecast and observed storms (objects) existed in the domain. Forecasts had a varying degree of skill with smaller complexes below 45 km2, while a trend of significant over-forecasting is seen in complexes larger than 45 km2. Preliminary results from recent summer 1-km forecasts will also be discussed.