A Forecast Evaluation Of The North American Summer Monsoon Precipitation Near Arizona In Recent Years
Presenter: Dylan Girone1
Co-Author(s): Christoforus Bayu Risanto, Lourdes Mendoza Fierro, Hsin-I Chang
Advisor(s): Dr. Castro
1Department of Hydrology and Atmospheric Sciences, University of Arizona
2021 was one of the wetter North American Monsoon (NAM) years in the Southwest U.S., following the driest monsoon year of 2020. Tucson received more than 12 inches of precipitation in 2021, the third highest on record. We are investigating the NAM precipitation forecast skill of the real time University of Arizona Weather Research Forecasting (WRF) model managed by the Department Hydrology and Atmospheric Sciences during the 2021 NAM season in Arizona. This forecast system is utilized by various stakeholders in the Southwest, including utilities, flood control districts, and the National Weather Service. The goal is to determine the value added of the NAM precipitation forecasts generated by this model with a 1.8 km, convective-allowing grid spacing. Daily retrospective forecast results from the real time UA WRF model are compared with the WRF forcing data, as evaluated against the observational Multi-Radar Multi-Senor (MRMS) gridded precipitation for the time period from June 1 to September 30 during the year. All datasets used are re-gridded to a uniform 1-kmx1-km spatial resolution before comparison and metric calculations. Forecast capability skills were evaluated through a series of contingency table-based analysis: probability of detection (POD), critical success index (CSI), and false alarm rate (FAR). We will present the relevant statistical metrics and identify regions where the WRF model consistently exhibits low forecasting accuracy and analyze why this may occur in those specific regions in order to recommend adjustments to the model.