Mapping the Convective Watersheds: Assessing the Predictability of Convective Evolution with Idealized Numerical Simulations

Thunderstorms provide numerous hazards throughout the United States, including lightning, flooding, hail, wind, and tornadoes. Convective evolution can be difficult to predict, however, due to high sensitivity to small perturbations on all scales. Storms grow in scale and shape  according to many factors, including profiles of buoyancy, wind, and moisture. Previous studies exploring these profile parameters have shown variations in sensitivity of supercell evolution within parameter space. It is extreme sensitivity to small movements within parameter space that denotes the areas of lowest predictability. Hence, this presentation shows results from hundreds of idealized WRF simulations of convective growth across very small intervals of parameter space, and the parameter-space regions that correspond to the most difficult forecasts.

Consider a terrain map, showing mountain peaks, steep slopes, and broad watershed basins. A rainstorm over a peak would yield high uncertainty in which watershed the falling rain would flow into. In the same way, our study draws from hundreds of idealised numerical simulations to plot storm characteristics in regions of parameter space to map “convective watersheds”. Storms that evolve in regions of steep slopes and lone peaks will be most sensitive to initial conditions, suggesting inherently low predictability. Conversely, broad regions of parameter space with low gradients indicate strong attraction and hence high predictability. Our results are then compared to observed convective events, forecast performance, and attendant atmospheric conditions.