Optimal Configurations of Ensemble Prediction Systems for Short-Range QPF

University Of Arizona, Tucson AZ

Investigators

Abstract

Ensemble Forecasting is a technique under development at a number of laboratories and universities to try to deal with the fact that different forecast models with different strengths and weaknesses and different initial assumptions, pararneterizations, etc., give different results. The ensemble forecast is a sort-of average of all the individual forecasts by different models. Verification studies have shown that ensemble forecasts can be better than any of the individual forecasts within the ensemble. This research will undertake to improve the reliability of ensemble forecasts of precipitation 12 to 36 hours in advance of its occurrence by addressing the issue of optimal configurations of mixed ensemble prediction systems for mesoscale limited-area models. This program of study will examine the sensitivity of model forecasts to variations in boundary layer and cumulus parameterizations as well as stochastic perturbations. The examination will involve multiple cases (between 30 and 60 per season) of mesoscale model forecasts at a horizontal grid spacing near 30 km for the cool and the warm seasons to allow for a rigorous statistical assessment of error growth characteristics and ensemble performance. This research is important because it will lead to better estimates of predictability limits for precipitation and provide guidance on how to construct optimum mesoscale ensemble prediction systems to improve operational forecasts of precipitation with limited computational resources.