根系水质模型中土壤与作物参数优化及其不确定性评价

Optimizing and uncertainty evaluation of soil and crop parameters in root zone water quality model

农业系统模型参数优化存在很高的不确定性,是模型应用研究的重点和难点。该研究利用自动优化程序PEST(parameter estimation software)对根系水质模型(root zone water quality model,RZWQM)中土壤参数(土壤水力学参数和根系生长参数)和作物遗传参数进行了优化,结果表明PEST优化模拟结果明显优于传统试错法的校正结果,且具有较高的参数优化效率。模型参数优化不确定性评价表明校正数据和参数初始值的选择、土壤水力学参数估算方法、不同类型参数间的相互作用以及优化目标方程(误差来源计算)都明显影响模型模拟结果。以上过程中土壤水力学参数优化值差异较小,但其土壤水分特征曲线却明显不同。通过以上评价分析提高了RZWQM相关参数优化结果的可靠性及其模拟功能,降低了模型参数优化的不确定性,为PEST优化其他模型参数提供了重要支持。

Reducing uncertainty in optimizing agricultural system model parameters is the key issue for model applications. An automated parameter estimation software （PEST） was used to calibrate the soil parameters and crop genetic parameters in the root zone water quality model （RZWQM）. The simulation results optimized by PEST were better than the calibration results via manual trial and error method, and showed higher efficiency. Parameterization uncertainty analysis of the model by PEST showed that the calibration data selection, initial parameter value, soil hydraulic parameter estimation method, interactions between these different parameter types and objective functions （error resources） had significant influences on PEST optimization results. Similar optimized soil hydraulic parameters were obtained from above processes, but produced different soil water retention curve. By above assessment, the uncertainty in RZWQM parameter optimization was reduced with improved soil water and crop yield predictions. This result can help optimize parameters of other similar models by PEST.