Wheat crop models underestimating drought stress in semi-arid and Mediterranean environments

Authors: WEBBER, H - Institute Of Landscape Systems Analysis, Leibniz Centre For Agricultural Landscape Research; COOKE, D - Institute Of Landscape Systems Analysis, Leibniz Centre For Agricultural Landscape Research; WANG, C - Institute Of Landscape Systems Analysis, Leibniz Centre For Agricultural Landscape Research; ASSENG, P - University Of Munich; MARTRE, P - Institute Of National Research For Agriculture; EWERT, F - Institute Of Landscape Systems Analysis, Leibniz Centre For Agricultural Landscape Research; KIMBALL, B - Retired ARS Employee; HOOGENBOOM, G - University Of Florida; Evett, Steven; CHANZY, A - Université D’Avignon Et Des Pays De Vaucluse; GARRIGUES, S - European Centre For Medium-Range Weather Forecasts (ECMWF); OLIOSO, A - Université D’Avignon Et Des Pays De Vaucluse; Copeland, Karen; STEINER, J - Kansas State University; CAMMARANO, D - Aarhus University; CHEN, Y - Chinese Academy Of Sciences; CRÉPEAU, M - Agriculture And Agri-Food Canada; DIAMANTOPOULOS, E - University Of Bayreuth; FERRISE, R - University Of Florence; MANCEAU, L . . . - Institute Of National Research For Agriculture; WHITE, J - University Of Florida

Submitted to: Field Crops Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/7/2025
Publication Date: 6/13/2025
Citation: Webber, H., Cooke, D., Wang, C., Asseng, P., Martre, P., Ewert, F., Kimball, B., Hoogenboom, G., Evett, S.R., Chanzy, A., Garrigues, S., Olioso, A., Copeland, K.S., Steiner, J.L., Cammarano, D., Chen, Y., Crépeau, M., Diamantopoulos, E., Ferrise, R., Manceau, L., White, J. 2025. Wheat crop models underestimating drought stress in semi-arid and Mediterranean environments. Field Crops Research. 332. Article 110032. https://doi.org/10.1016/j.fcr.2025.110032.
DOI: https://doi.org/10.1016/j.fcr.2025.110032

Interpretive Summary: Crop growth models are useful tools for assisting in the management of agricultural crops as well as for predicting the likely effects of droughts and other adverse weather. An important aspect that determines the ability of crop growth models to simulate growth and yield is their ability to simulate the rate of water consumption or evapotranspiration (ET) of the crop, especially for rain-fed crops. If, for example, the simulated ET rate is too high, the simulated crop may exhaust its soil water supply before the next rain event, thereby causing growth and yield predictions that are too low. Prior maize model intercomparisons revealed large variability among the models and poor performance by many of the models. Similar variability was found among wheat models, and a large, systematic bias was revealed, whereby the model median underestimated water use in all environments evaluated. Collectively, the results suggest the need to improve the ability of the models to simulate ET in order to avoid underestimating projected impacts of drought and to estimate the required water resources for irrigated systems. This research will help present-day and future farmers and agricultural researchers, and of course all food consumers.

Technical Abstract: Under increasingly extreme weather, projections of water demand and drought stress from process-based crop models can provide risk management and adaptation strategies. Previous studies investigating maize crop models demonstrated considerable error in simulation of water use, but and no similar evaluation of wheat crop models exists. The aims here were to (1) evaluate wheat model performance in reproducing observed daily evapotranspiration (ET) of wheat for Mediterranean and semi-arid environments; and (2) identify factors and processes associated with model error and uncertainty across environments. These were assessed with an ensemble of wheat crop models for two experiments conducted in Bushland, Texas, USA (three seasons, deficit and full irrigation) and Avignon, France (four seasons rainfed) with winter bread and durum wheat, respectively. Models were calibrated with all observed data for crop growth. A sensitivity analysis for each environment varied intermediate process values related to water use. Reflecting a large, systematic bias, the model median underestimated water use in all environments evaluated. Relative error in underestimating daily ET was constant across levels of atmospheric evaporative demand; therefore, the absolute error was higher for days with larger evaporative demand. This implies errors in the soil water balance increase more rapidly under high evaporative demand conditions. The use of a potential versus reference evapotranspiration approach did not explain relative model performance, although the sensitivity analysis indicated that simulation of demand terms explained much more uncertainty in seasonal water use than terms related to soil depth or root growth. Errors in simulated leaf area index were associated with errors in daily ET, but the relationship varied with growth stage. Collectively, the results suggest the need to improve simulation of evaporative demand to avoid underestimating projected impacts of drought or required water resources for irrigated systems.