Abstract:

In order to assess the suitability of MaizeSM crop model in Tumd Left Banner, a global sensitivity analysis method was employed to identify sensitive parameters of the model. Subsequently, localized model parameters were calibrated using maize variety data from experimental fields, meteorological observations, soil physical and chemical data, and field management records spanning 2010 to 2022. This calibration enabled accurate simulation and prediction of local maize growth processes and characteristics across different stages. Accuracy of simulation results was verified using actual yield and growth period duration indicators. The results showed that findings revealed nine sensitive parameters within the model, with k₁ (emerging-joining stage basic development coefficient) being identified as most sensitive while TR₁ (stem sheath storage transport efficiency parameter before flowering) exhibited minimal sensitivity. Strong correlations between simulated values for each growth period and actual values were observed, with normalized root-mean-square error (NRMSE) below 30% and root-mean-square error (RMSE) falling within an acceptable range. The crop model can simulate the local maize growth well. The crop model demonstrates good simulation performance for local maize growth. The localized maize growth simulation model, MaizeSM, with improved parameters, has enhanced the refined yield prediction based on station-scale agricultural meteorological services. This further strengthens the application capabilities of agricultural models in climate change impact assessment, operational services, and agricultural production in the Tumd Left Banner region. These advancements assist agricultural managers in formulating optimal planting strategies to achieve maximum production efficiency.

Key words: MaizeSM crop model, Tumd Left Banner, applicability, sensitive parameter, localization debugging, accuracy verification