Evaluation of storm initiation uncertainty information from a high-resolution time-lagged ensemble and its application to short-term forecasting

There is a need for improved short term (0-8 hour) forecasts of precipitation intensity and more specifically thunderstorms in a number of aspects of society including aviation planning, flash-flood forecasting, outdoor recreation planning, construction and agriculture. Forecasting the location and timing of initiation of strong storms that produce heavy rain, lightning and/or hail is inherently a probabilistic problem owing to the chaotic nature of the interplay of dynamics and physics that spans a range of scales. The goal of this study is to ascertain if the likelihood of storm initiation can be derived from a high-resolution time-lagged ensemble. The NOAA/GSD High Resolution Rapid Refresh (HRRR, a WRF-based NWP model that includes radar-data assimilation and cycling) will be analyzed to assess the frequency of occurrence of observed storm initiation as a function of the number of time-lagged members that predicted storm initiation. Storms are defined as an area of greater than 100 km2 that exceeds 40 dBZ. The analyses will be performed by time of day, region, synoptic conditions and storm type (with a focus on larger-scale storm systems). Preliminary analyses indicate that the HRRR has appreciable skill in predicting initiation along cold fronts and stationary fronts but has difficult forecasting secondary initiation along outflow boundaries. The work performed here will aid in the future development of CoSPA (Consolidated Storm Prediction for Aviation – which blends NWP and radar extrapolation/evolution models) by improving the treatment of storm initiation in the 2-4 hour time frame.