Cross-scale evaluation of dynamic crop growth in WRF and Noah-MP-Crop

Abstract

Abstract Accurately representing croplands in climate models is important for simulating water and energy fluxes between the land and atmosphere, as well as evaluating the impacts of climate change on agriculture. The recent integration of dynamic crop growth in the Noah land surface model with multiparameterization (Noah-MP-Crop) has the potential to substantially advance Earth system modeling and is included in the latest release of the Weather Research and Forecasting (WRF) regional climate model. The addition of dynamic crop growth to WRF provides a unique opportunity to simultaneously evaluate biases associated with a crop model coupled to a regional climate model and address outstanding questions regarding the role of agroecosystems in modulating regional climate. Here, we analyze dynamic crop growth in WRF across three simulated spatial scales (25km, 5km, and 1km) for growing seasons with precipitation above (2010), below (2012), and approximately equal (2015) to the seasonal average. Including dynamic crop growth in WRF significantly reduces biases in simulated leaf area index over croplands in the central U.S. relative to observations. However, there is no substantial difference in calculated daily evapotranspiration, average growing season temperature, or total growing season precipitation between WRF simulations with dynamic crops (WRF-Crop) compared to the dynamic vegetation module without crops (WRF-DV). Simulated corn (soy) mean absolute error (MAE), as a percentage of observed annual average yield, ranges from 24.7% -101% (28.1% - 109%) depending on year and spatial resolution, with the most significant biases in highly irrigated counties. Forcing Noah-MP-Crop with observed climate substantially reduces the range of corn (soy) yield MAE to 9.5% - 55.1% (15.0% - 37.5). Increased model resolution consistently leads to lower corn and soy yield estimates within WRF-Crop.