Evaluation of hydrometeorological forecast skill during Hurricane Harvey (2017)

Abstract:

Hurricanes bring heavy rain and lead to catastrophic flooding. The damage and fatalities due to Hurricane Harvey underscore the urgency for the understanding and improving forecast skill for the decision-making process such as hospital patient evacuation and electric grid resilience. In this study, we evaluate the performance of the WRF-Hydro/National Water Model in predicting floods during Hurricane Harvey in 2017. In an effort to attribute uncertainties in the streamflow simulations, we evaluate the predicted hurricane track, precipitation, and streamflow over hurricane-prone areas of Texas. We will demonstrate the forecast skill at lead times of 1-10 days, and assess the impact of initial conditions on the predictability. The preliminary results show that forecast skill generally decreases with lead time, but there exist circumstances in which forecast skill increases. By linking synoptic weather forecasts to streamflow forecasts, we seek to provide a comprehensive understanding of the performance of the integrated hydrological modeling framework.