Incorporating ensemble-based probabilistic forecasts into a campaign simulation in the Weather Impact Assessment Tool (WIAT)

Abstract

Stochastic (probabilistic) forecasting techniques give forecasters a means to transmit information about certainty and a range of forecast possibilities to operational decision makers. Previous studies have shown value in probabilistic forecasting through series of independent hypothetical events, or through comparative period forecasts. This thesis demonstrates the value of stochastic forecasting through a series of operational events, in the context of a Strike Warfare campaign in the Weather Impact Assessment Tool (WIAT), a campaign simulator. Simulated ensemble members and probability files were created to study six weather parameters and their impacts on various Strike Warfare missions. Tests were run comparing deterministic and stochastic forecasts, incorporating varying levels of forecast error and sampled probability thresholds. Metrics within and external to WIAT were employed to analyze the results of the forecasting strategies. Constraints in WIAT's structure and campaign modeling yielded results that suggest, but do not definitively prove, enhanced campaign performance as a result of incorporating probabilistic forecasting. Programming adjustments to WIAT are recommended in order to provide a higher-quality test bed for future studies of both deterministic and stochastic forecasting techniques.