Study on the impact of low‐temperature stress on winter wheat based on multi‐model coupling

Abstract Crop growth models, such as the WOrld FOod STudies (WOFOST) model, mimic the mechanistic processes involved in crop development, growth, and yield production. The accuracy of simulation is decreased in unfavorable low‐temperature settings because these models do not accurately represent crop response processes in low‐temperature stress. Enhancing the WOFOST crop growth model's accuracy in simulating crops' responses to cold temperatures is the aim of this work. Given its vulnerability to low temperatures, the inquiry uses winter wheat in Henan Province as a focal point. It integrates the WHEATGROW wheat phenology model with the Frost model of Lethal Temperature 50 (FROSTOL) inside the framework of the crop growth model. This link aims to improve simulation accuracy and supplement the model's mechanisms, particularly when it comes to the impact of low temperatures on crop development. The study uses Long Short‐Term Memory networks to build a yield model that integrates remote sensing data with information from simulated crop models. Under low temperatures, the leaf area index, total above ground biomass, and total weight of storage organs of the model WWF—which combines FROSTOL and WHEATGROW with WOFOST—show a considerable decline. It was discovered that there is a greater improvement in simulation accuracy of the linked model WWF relative to the WOFOST model in frost years than in normal years, based on a comparison analysis between typical frost years and normal years. To be more precise, the improvement is 8.03% in frost years and 1.98% in regular years. When all is said and done, the coupled model advances our knowledge of how winter wheat is impacted by low temperatures.