Effects of urbanization on extreme rainfall in an arid/semiarid region

In this study, an extreme rainfall event that occurred mainly on October 2, 2018, in the Phoenix metropolitan area, southwestern United States, is simulated with the Weather Research and Forecasting (WRF) model, version 4.0, with multiple microphysics and boundary layer schemes. Paired simulations are conducted by running the model first using a realistic land cover and land-use dataset (URB run), and second, by replacing the urban land cover with open shrubland cover (called NOURB run). The model simulations are first evaluated against radar Stage IV (hereafter Stage IV) data and multiple radar and multiple sensors (MRMS) rainfall to validate the structure, amplitude, and location (SAL) of the rainfall field for both URB runs and NOURB runs. These evaluation results indicate that the model performance varies considerably depending on the physics schemes employed. Based on SAL values, the average of model simulations captures the main features (values of S, A, and L are less than 0.1) both for URB runs and for NOURB runs. Comparisons of hourly rainfall rates show that modeled rainfall maximums occurred about 1 hr earlier than both Stage IV and MRMS data over the Phoenix metro area, while the maximum rainfall amounts from the model are close to MRMS data but lower than Stage IV data. The differences of rainfall and other meteorological variables between URB runs and NOURB runs are analyzed. The differences of rainfall between URB and NOURB vary depending on the physics scheme employed. Our study also indicates that urbanization-induced rainfall changes are small within the urban area but are significant at some locations outside of it; an explanation for this pattern needs further investigation. This result implies that using a single physics scheme to address the effects of urbanization on extreme rainfall events may result in inappropriate conclusions.