Examining WRF’s Sensitivity to Contemporary Land-Use Datasets across the Contiguous United States Using Dynamical Downscaling
Citation:
Mallard, M., T. Spero, AND S. Taylor. Examining WRF’s Sensitivity to Contemporary Land-Use Datasets across the Contiguous United States Using Dynamical Downscaling. JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY. American Meteorological Society, Boston, MA, 57(11):2561-2583, (2018). https://doi.org/10.1175/JAMC-D-17-0328.1
Impact/Purpose:
Land use representation plays a critical role in simulating air-surface interactions that affect meteorology and regional climate. In this study, two land use datasets within the Weather Research and Forecasting (WRF) model, the U.S. Geological Survey (USGS) and the 2006 National Land Cover Dataset (NLCD), are used to drive three-year simulations to assess the sensitivity of dynamically downscaled WRF runs to the land use representation.
Description:
Land use (LU) representation plays a critical role in simulating air-surface interactions that affect meteorology and regional climate. In the Noah LSM within the WRF model, LU categories are used to set the radiative properties of the surface and to influence exchanges of heat, moisture, and momentum between the air and land surface. Previous literature examined the sensitivity of WRF simulations to LU using short-term meteorological modeling approaches. Here, the sensitivity to LU representation is studied using continental-scale dynamical downscaling, which typically uses longer temporal and larger spatial scales. Two LU datasets, the USGS and the 2006 National Land Cover Dataset (NLCD), are utilized in three-year dynamically downscaled WRF simulations over a historical period. Precipitation and 2-m air temperature are evaluated against observation-based datasets for simulations covering the contiguous U.S. The WRF-NLCD simulation tends to produce lower precipitation than the WRF-USGS run, with slightly warmer mean monthly temperatures. However, WRF-NLCD results in more notable increases in the frequency of hot days (i.e., days with temperature >90°F). These changes are attributable to reduction of forest and agricultural area in the NLCD relative to USGS. There is also subtle but important sensitivity to the method of interpolating LU data to the WRF grid in the model preprocessing. In all cases, the sensitivity resulting from changes in the LU is smaller than model error. While this sensitivity is small, it persists across spatial and temporal scales.
URLs/Downloads:
DOI: Examining WRF’s Sensitivity to Contemporary Land-Use Datasets across the Contiguous United States Using Dynamical Downscaling