A Numerical Study of Hurricane Maria 2017 Using a Fast Version Weather Research and Forecasting (WRF) Model

A set of numerical simulations are conducted to analyze various spatio-temporal resolution, physics sensitivities, and model setup (domain size, and spinup) of Hurricane Maria 2017 (HM17) using a fast version of the WRF model over a limited model domain. The WRF model is Graphics Processing Unit (GPU) enabled with limited options for each physics parameterization. We evaluated the model performance in hurricane track, intensity, as well as structures of HM17 using multiple source of observational datasets, including dropsonde, radar, as well as observation based data products. The questions we aim to answer include (1) whether the hurricane can be reasonably captured if the model domain is not large enough to cover the genesis of the hurricane; (2) what is/are the most important physics in WRF for hurricane simulations concerning their tracks, intensities, and structures? In particular, would the model performs better when the radiative transfer models are called more frequently given the speedup by GPU WRF?; (3) what the HM17 would be like in future with impacts of global warming?