Examining the Effects of Mosaic Land Cover on Extreme Events in Historical Downscaled WRF Simulations

The representation of land use and land cover (hereby referred to as “LU”) is a challenging aspect of dynamically downscaled simulations, as a mesoscale model that is utilized as a regional climate model (RCM) may be limited in its ability to represent LU over multi-decadal periods. Such is the case with the Weather Research and Forecasting (WRF) model, which utilizes static LU data. While the topography of coastal features and terrain is better represented in a fine-resolution RCM [as compared to a global climate model (GCM)] the representation of LU must also be robust over long time periods in order to demonstrate added value within a downscaling application. As the treatment of LU influences fluxes of momentum, heat, and moisture from the surface, a realistic representation of LU is needed to capture historical conditions and to project future changes in both mean conditions and extreme events, as well as the resulting impacts on air quality, human health, and ecosystem services. Here, WRF is used with the Noah land surface model and various treatments of LU are tested within a dynamical downscaling application.

Historical WRF simulations are conducted over a three-year period with a set of nested 36- and 12-km domains that cover the continental U.S. The 0.75° × 0.75° ERA Interim is downscaled and treated as a proxy for a similarly coarse GCM. Simulations are conducted with Noah’s default approach (wherein the most prolific LU type is assigned as the single dominant LU type present in the grid cell) and this is contrasted with results with the mosaic LU field. The mosaic option is set up such that up to six LU types can be present within a grid cell. All simulations use the NLCD 2011 LU data. Additionally, updated vegetation and soil information within the Noah look-up tables will be included to test whether the simulations are improved, as well as how this sensitivity compares with WRF’s sensitivity to the number of LU categories represented.

The ability of the simulations to accurately simulate monthly mean conditions, as well as extremes in minimum and maximum daily 2-m temperatures and heavy rainfall events will be investigated. Although the improvement of the 12-km simulations is the main area of focus, the 36- and 12-km outputs will be compared to assess whether the simulation of near-surface fields are improved with the use of finer grid spacing.