Citation:

Mallard, M. AND T. Spero. Representing Contemporary Land Use in WRF Regional Climate Simulations. American Meteorological Society's 102nd Annual Meeting, NA, Texas, January 23 - 27, 2022.

Impact/Purpose:

To summarize previous research on the impact of various choices of land use representation on regional climate simulations done with the Weather Research and Forecasting model  

Description:

The representation of land use (LU) in regional climate modeling can have significant effects on fluxes of heat, moisture, and momentum between the air and atmosphere, influencing the model’s ability to properly simulate rainfall and temperature in both historical and future realizations. Inadequate treatment of LU can, therefore, hinder the model’s ability to simulate future changes in extreme rainfall and temperature events and project their effect on ecosystem services and human health. This presentation summarizes prior works that focused on representation of LU in historical simulations utilizing the Weather Research and Forecasting (WRF) model with the Noah land surface model (LSM).  The model’s sensitivity to 1) LU dataset source, 2) the method of interpolating LU data to the modeled domain, and 3) the use of the Noah LSM’s mosaic capability to represent subgrid-scale LU variability are assessed using 3-year historical WRF runs and contemporary LU datasets.     Mallard et al. (JAMC, 2018) used WRF version 3.8 to compare simulations driven by the NLCD 2006 versus the USGS LU data.  These runs were performed over a 36-km contiguous U.S. (CONUS) domain. Generally, compared with WRF-USGS, WRF-NLCD was found to increase mean 2-m temperatures and reduce monthly rainfall, with an accompanying increase in the number of days with extreme temperatures. This was found to be driven by reduced forest and agricultural LU types within the WRF-NLCD representation relative to WRF-USGS. Similar LU differences in key categories, like forest LU, were also found when the NLCD data was applied to the target model grid using an alternative interpolation method in the WRF preprocessing (the default method when the USGS data is chosen for LU representation). This suggests that differences in interpolation method, along with differences in the LU data source, alter the WRF LU representation. The Mallard et al. (JAMC, 2018) study was conducted using only a single most-dominant LU category in each 36-km grid cell. In a subsequent study, Mallard and Spero (JGR-A, 2019) ran12-km CONUS-wide WRF simulations using the Noah LSM’s mosaic capability to represent LU variability using subgrid-scale tiles to represent multiple LU types present in each grid cell. These were compared to the previously used most-dominant LU category approach.  All simulations were driven by a common LU source, the NLCD 2011 data. The study found that impacts on near-surface meteorology from using mosaic LU were primarily associated with modeling urban LU types. The use of the mosaic capability resulted in widespread increases in the coverage of low-intensity urban LU with accompanying reductions in forest and agricultural types. This resulted in widespread increases in 2-m temperatures and sensible heat flux, while precipitation and latent heat fluxes were generally reduced. Within large cities, the opposite trend was found, as use of mosaic LU enabled WRF to represent green space within areas previously treated as solely urban under the dominant LU treatment. Overall, the use of the mosaic approach elevated the significance of LU categories, like urban types, that were previously only found in highly developed areas and were under-represented in suburban and rural areas under the dominant treatment of LU employed exclusively in previous versions of WRF.

URLs/Downloads:

Record Details: