Decomposing the total uncertainty in wheat modeling: An analysis of model structure, parameters, weather data inputs, and squared bias contributions

Abstract

CONTEXTThe comparison of agricultural models and the conduct of crop improvement research have garnered significant attention in recent times. One of the primary objectives in this field is to pinpoint and mitigate the uncertainties inherent in modeling the effects of climate on crop growth and productivity. Enhancing the precision and reliability of crop models has emerged as a critical concern. OBJECTIVEIn this study, we calibrate and validate four wheat phenology models using wheat phenology data from 1990 to 2009. More importantly, we explain three significant sources of uncertainty in wheat phenology models, namely model structure, model parameters, and weather data inputs. METHODSThis study examines four wheat models—the GLAM-Wheat model, APSIM-Wheat model, SPASS-Wheat model, and WOFOST model—to simulate phenological changes across 32 agricultural meteorological stations in the North China Plain. Additionally, the three main sources of uncertainty in the model are quantified using the Markov Chain Monte Carlo (MCMC) method. RESULTS AND CONCLUSIONSThe results indicate that all four wheat phenological models effectively simulate the growth of wheat in the study area, with an average RMSE ranging from 4.4 to 5.2 days for the heading stage and from 4.7 to 5.6 days for the maturity stage. The uncertainty analysis encompasses parameters, squared bias, weather data inputs, and model structure. During the heading stage, the overall contributions of these uncertainties are 8.9 %, 40.8 %, 47.4 %, and 2.9 %, respectively. During the maturity stage, these contributions are 11.2 %, 51.2 %, 35.0 %, and 2.6 %, respectively. Weather data inputs are identified as the primary sources of uncertainty. SIGNIFICANCEThis study quantifies the uncertainty within wheat phenology models, a critical step towards enhancing the precision and dependability of crop models. Such efforts hold substantial importance in shaping agricultural policies and refining management practices, ultimately aiding in tackling the challenges posed by impending climate change.