Using Extreme Events to Compare USGS and NLCD Land Use Data Sets in WRF for Dynamical Downscaling

Land use changes occur frequently due to deforestation and construction for the growing population; however, some land use data sets used as input for model simulations may not reflect those changes. Assessing model sensitivity to this input is necessary because land use differences among datasets can have an effect on localized weather conditions and processes. This becomes important for downscaling studies of regional areas where local variables such as temperature, low level winds and precipitation may be affected. This study poses an important question of which land use data set produces the best model performance between the United States Geological Survey (USGS) land use data and the 2006 National Land Cover Data (NLCD) when using the Weather Research and Forecasting (WRF) model for dynamical downscaling. Extreme climate events (i.e. drought, tropical storms, hard freezes, flooding, and severe weather outbreaks) during 1988-1990 were selected as case studies to determine how WRF simulations driven by these land use data sets react to different atmospheric scenarios and/or conditions. The three-year, 36-km WRF simulations of the contiguous United States using these two land use data sets were compared to Climate Prediction Center (CPC) precipitation data and North American Regional Reanalysis (NARR) data to evaluate model performance. Although the results demonstrate that both simulations are similar, the WRF run driven by the NLCD data set is better able to represent 2 meter temperatures while the USGS driven run simulates precipitation distribution slightly better.