Impact of implementing the Local Climate Zone urban classification scheme in WRF on weather simulations in Houston

Abstract:

As the most populated city in Texas, Houston is vulnerable to extreme rainfall, floods, hurricanes, and extreme heat. The risk from these hazards will likely increase in the future under the influence of urbanization and climate change. The spatial heterogeneity of urban land surface creates significant uncertainties for accurate and reliable weather forecasts in this coastal, urban region. This study investigates the effect of urban spatial heterogeneity on weather simulations using the Weather Research and Forecasting (WRF) model with different strategies in characterizing land surface properties. The first strategy employs the updated MODIS land use land cover (LULC) dataset, while the second replaces the urban class in MODIS LULC dataset from the detailed urban Local Climate Zone (LCZ) classification. The dataset of 10 different urban LCZs was generated at 300-m spatial resolution using 30-m Landsat 8 radiance data and the random forest supervised classification machine learning algorithm under the World Urban Database and Access Portal Tools (WUDAPT) framework. The 6-hour FNL (Final) Operational Global Analysis data were supplied as the initial and boundary conditions for driving the WRF coupled with a single-layer urban canopy model. The simulations under clear-sky days, rainy days, and Hurricane Harvey will be presented. Initial results show that including the LCZs can simulate both day- and nighttime land surface temperatures with a higher spatial correlation, and a lower root-mean-square error than using the MODIS LULC alone under clear-sky days.