CERES-Wheat作物模型参数全局敏感性分析

作物生长模型广泛应用于区域作物估产研究与应用之中，如何选择最佳作物模型优化参数是获得较好模拟预测结果的关键之一。研究选择河南洛阳为试验区，应用扩展傅里叶振幅灵敏度检验（EFAST）法对CERES-Wheat模型作物参数及田间管理参数进行了全局敏感性分析。结果表明，完成一片叶生长所需积温、最适温度条件下通过春化阶段所需天数、光周期参数、最佳条件下标准籽粒质量参数、开花期单位株冠质量的籽粒数参数等指标具有较高敏感性，系为模型参数“本地化”的关键参数。播种日期、播种密度、施肥日期、播种深度、灌溉日期是模型区域化应用的最佳优化变量。研究表明，EFAST敏感性分析是模型参数“本地化”和选择最佳“区域化”优化变量的有效方法。

Global sensitivity analysis of CERES-Wheat model parameters

Crop growth models have been applied extensively in the regional crop yield prediction and estimation. It is very important to select the most sensitive model parameters for the model optimization and better model output. The Extend Fourier Amplitude Sensitivity Test (EFAST) was used to analyze the sensitivity of CERES-Wheat model parameters in a study region in Luoyang, Henan province. The sensitivity of crop and field management parameters were analyzed. The results show that these parameters including the interval between successive leaf tip appearances, days at optimum vernalizing temperature required to complete vernalization, percentage reduction in development rate in a photoperiod 10 hour shorter than the threshold relative to that at the threshold, standard kernel size under optimum conditions, kernel number per unit canopy weight at anthesis are the key sensitive parameters which should be firstly selected for the model localization. The optimal parameters selected for application of model in regional scale are planting date, planting density, fertilization date, planting depth and irrigation date. The research showed that the global sensitivity analysis in EFAST is effective for parameter selection in the crop growth model optimization to improve its performance at regional scale.