基于Downhill-Simplex算法的观测数据与作物生长模型同化方法研究

摘要

以夏玉米叶面积指数(LAI)、贮存器官干重(WSO)、地上总干重(TAGP)以及土壤水分含量(SM)为结合点,建立了基于Downhill-Simplex算法的作物生长模型WOFOST同化多种地面观测数据的一般方法或流程:开展观测数据与作物生长模型同化方法的正确性验证→利用Downhill-Simplex算法进行WOFOST模型的敏感性分析→选择敏感参数组合→通过优化效果确定待优化参数→利用新的观测数据对待优化参数进行优化,从而实现了观测数据与作物生长模型的同化,提升了模型的模拟能力.同化过程中遴选出的WOFOST模型的待优化参数主要包括比叶面积、最大CO2同化速率、初始地上部总干物重、根深最大日增量和初始土壤有效水等.%The scheme of assimilation of multivariate observation data and crop growth model based on Downhill -Simplex algorithm was established while LAI, dry weight of living storage organs ( WSO) , total above ground production ( TAGP) and soil moisture ( SM) as the point of integration for summer maize in Hebei. Correctness verification of assimilation of the observational data and crop growth model was firstly performed. Then, sensitivity of all parameters and initial value of the state variables in WOFOST were analyzed based on the Downhill - Simplex algorithm and the parameters to be optimized were determined through selection of parameter groups and optimization results. The optimal value of those parameters was at last obtained by means of optimization of new observed data. So, assimilation of measured data and crop growth model was achieved and simulated accuracy of crop growth model was improved. In addition, parameters to be optimized in data assimilation mainly included specific leaf area, leaf maximum CO2 assimilation rate, the initial total crop dry weight, maximum daily increase in rooting depth, and initial amount of available water in total root zone.