9424397 Stensrud Due to nonlinearities in the physics of the atmosphere, deterministic predictions of the atmosphere have their limitations. A technique known as ensemble forecasting is believed to provide a mechanism for improving forecasts of the atmosphere. Ensemble forecasting involves running a series of numerical predictions by changing, within reason, model initial conditions or model physics. The objective is to provide forecasts that define the range of possible solutions. These techniques have shown promise for forecast improvement. A joint project between the Cooperative Institute for Mesoscale Meteorological Studies and the University of Colorado is proposed with the objective to develop and test ensemble forecasting techniques for mesoscale models. The proposed research will focus upon the difficulties involved in accurately predicting mesoscale convective weather events. Three specific objectives of the research are: 1) to develop a technique to modify the mesoscale model initial conditions to define the ensemble group that maximizes the span of model solutions across the predictability phase space using a minimum number of simulations, while requiring that the ensemble initial conditions remain constrained to the error characteristics of the available observations; 2) to explore the results of this technique for both strongly and weakly forced large-scale environments; 3) to determine the applicability of using varying model physics in ensemble forecasting methods.. Knowledge gained from this study will provide necessary information on how to efficiently produce an ensemble forecast using a mesoscale model. It is anticipated that the model results also will provide some guidance on predicting the model skill for given convective weather events. ***
