Developing a Probabilistic Heavy-Rainfall Guidance Forecast Model for Great Lakes Cities

Annual flooding and flash flooding as a result of heavy rainfall events pose a significant threat to property and life, with increased vulnerabilities for urban communities. Therefore, improved predictability of these heavy rainfall events is vital for increasing situational awareness for forecasters and mitigating risks. More specifically, this study aims to improve heavy rainfall predictability for the Milwaukee and Chicago County Warning Areas (CWAs) through the development of a zero to 24-hour probabilistic forecast tool using an Artificial Neural Network (ANN). Predictor variables for the neural network are obtained using the 32-km, 3-hourly meteorological reanalysis data for the warm-season (April-October) between 2002-2016 from the North American Regional Analysis (NARR) dataset. Historical rainfall data is obtained from the hourly, multi-sensor National Centers for Environmental Prediction (NCEP) 4-km National Precipitation Dataset and interpolated to the NARR grid. Results following from this study will be presented at the conference.