Impact of initial/boundary conditions and model configuration to predict storms for the Northeast United States

Abstract:

Storms that induce high winds and heavy precipitation cause severe effects to the economy and society by electric power disruption along with other infrastructure losses. Especially in the NE US, severe weather conditions directly affect electric power outages. Improving numerical weather prediction (NWP) leads to improvements in the prediction of power outages, thus, enables accurate disaster preparedness and strategy. This study presents the impact of using different initial and boundary conditions and model configuration options to improve prediction of rain-wind and thunderstorm events over the NE US. Two different versions of the Weather Research and Forecasting (WRF) model, version 4.1.3 and version 3.8.1 have been implemented using the North American Mesoscale Forecast System (NAM) 12 km analysis and ERA5 reanalysis fields at 30 km as initial and boundary conditions. The derived wind speed and wind gust along with temperature, pressure, and humidity have been evaluated to understand the impact of different initializations, analysis nudging, model version and model configuration. Precipitation evaluation is conducted using the Global Historical Climatology Network Daily data (GHCND) and Stage IV data. We also evaluate specific weather variables (maximum and sustained wind speed, maximum and sustained wind gust, etc.) that are required by the Outage Prediction Model (OPM) of the Eversource Energy Center at the University of Connecticut to assess improvements that can be achieved for the electric utility if weather prediction performance is improved.