Towards the Development of an Operational Mesoscale Ensemble System for the DoD Using the WRF-ARW Model

Ensemble based forecasting has been proposed as a means to objectively encapsulate the inherent uncertainty associated with weather forecasting (Leith, 1974). Uncertainty arises due to an inability to truly specify the initial conditions and due to limitations with the forecast models themselves (Bjerknes et al., 1911; Lorenz, 1963 and 1969)., Within an ensemble system, a suite of diverse yet equally likely initial conditions and forecasts models are used to produce a suite of diverse yet equally likely forecasts. These forecasts are then used to describe a probability density function of specific environmental forecast variables (e.g., temperature, wind, pressure). The goal of the ensemble design is to ensure the forecast solution set encompasses observations yet is narrower in scope than the climatological solution set. Mesoscale ensembling is a both a design and computational challenge. Given the scope of the challenge, a joint proposal was submitted and selected for a Dedicated HPC Project Investment (DHPI) to investigate the ensembles for the Department of Defense. The goal of this project is to design and test a joint (multi-center) mesoscale ensemble prototype system to demonstrate the utility of stochastic weather intelligence to the decision-making process of the DoD warfighter at all echelons of combat. Half way through that project, this paper describes the progress thus far and discusses some of the design lessons learned as well as some subjective evaluation of the output.