Ensemble Variability in Rainfall Forecasts of Hurricane Irene (2011)

As tropical cyclones (TCs) move from the tropics into the midlatitudes, they are often associated with extensive regions of heavy precipitation. This precipitation can lead to widespread, devastating flood events, as occurred with Hurricane Irene (2011) over the northeastern United States. Despite the high-impact nature of these events, there are relatively few studies that explore the sensitivity of precipitation forecasts to model initial conditions, choosing instead to focus on the variability in TC track. The goal of this work is to understand what modulates precipitation forecasts over the northeastern United States during Hurricane Irene. This is investigated using the 3 km Weather Research and Forecasting (WRF) model to downscale members of the Global Forecasting System ensemble prediction system, initialized at 0000 UTC 27 August 2011. The ensemble members that forecast the largest precipitation totals (i.e., wet members) over the Catskill Mountains of New York State (where over 15” of rain were observed) are then compared to the members that predicted the least precipitation (i.e., dry members), in order to diagnose the processes that lead to the rainfall differences. Results indicate that the amount of rainfall is tied to storm track, with wetter members clustered on the western side of the track envelope, and drier members on the east. Due to the influences of terrain, however, the WRF members feature a more complex relationship between precipitation and storm track than a simple east–west correlation. The wettest members are characterized by lower-tropospheric winds that are directed perpendicular to the northeastern face of the Catskills (where the heavy rainfall occurred), allowing maximum upslope forcing during the period of highest rainfall rates. In addition, the wet members have greater southeasterly moisture transport into the region, providing the water vapor required for sustained heavy rainfall over the area.