Probabilistic forecasting for isolated thunderstorms using a genetic algorithm: the DC3 campaign

Researchers on the Deep Convective Clouds and Chemistry (DC3) field campaign in summer 2012 sought in-situ measurements using aircraft of isolated thunderstorms in three different study regions: northeast Colorado, northeast Alabama, and a larger region extending from central Oklahoma through northwest Texas. Experiment objectives required thunderstorms that met several criteria. To sample thunderstorm outflow, storms had to be large enough to transport boundary-layer air to the upper troposphere and have a lifetime long enough to produce a large anvil. The storms had to be small enough to safely sample and isolated enough that experimenters could distinguish the impact of a particular thunderstorm from other convection in the area. A decision-making recommendation tool was developed to aid in the optimization of daily flight decisions on a regional scale, producing probabilistic forecasts of suitable flight conditions for each of the three regions, accompanied by a recommendation to fly to one of the three regions or to not fly. Model-forecasted atmospheric variables for each region were converted to probabilistic forecasts of suitable conditions using fuzzy logic trapezoids, which delineated the favorability of each variable. In parallel, the trapezoid parameters were tuned using a genetic algorithm and the favorability values of each of the atmospheric variables were weighted using a logistic regression. Results indicate that the probabilistic forecasting method shows predictive skill over climatology in each region, with Brier Skill Scores as high as 31%.