Examining WRF's Sensitivity to Contemporary Land-Use Datasets across the Contiguous United States Using Dynamical...

Area of Science:

• Atmospheric Science
• Climate Modeling
• Geospatial Analysis

Background:

• Land-use (LU) representation is crucial for simulating air-surface interactions influencing regional climate.
• The Noah Land Surface Model (LSM) within the Weather Research and Forecasting (WRF) Model uses LU categories to define surface radiative properties and energy/moisture exchanges.
• Previous studies explored WRF sensitivity to LU using short-term meteorological modeling.

Purpose of the Study:

• To investigate the sensitivity of WRF simulations to land-use representation using continental-scale dynamical downscaling.
• To compare the impact of two distinct land-use datasets (USGS and NLCD) on meteorological simulations over the contiguous United States.
• To evaluate the influence of LU data interpolation methods on model outcomes.

Main Methods:

• Conducted 3-year dynamically downscaled WRF simulations over a historical period for the contiguous United States.
• Utilized two land-use datasets: the U.S. Geological Survey (USGS) dataset and the 2006 National Land Cover Dataset (NLCD).
• Evaluated precipitation and 2-m air temperature against observation-based datasets.

Main Results:

• The WRF simulation using the NLCD dataset produced lower precipitation and slightly warmer mean monthly temperatures compared to the WRF-USGS run.
• The WRF-NLCD simulation showed a notable increase in the frequency of hot days (temperature > 90°F or 32.2°C).
• Observed changes were attributed to reductions in forest and agricultural areas in the NLCD compared to the USGS dataset, with subtle sensitivity to LU data interpolation methods.

Conclusions:

• Land-use representation significantly influences WRF simulations, affecting temperature and precipitation patterns, particularly the frequency of extreme heat events.
• While the NLCD dataset alters simulation outcomes, the resulting sensitivity is generally smaller than the inherent model error.
• The observed land-use sensitivity persists across various spatial and temporal scales in continental-scale dynamical downscaling.