基于PEST的土壤水力参数数值反演研究

发布时间：2018-02-02 22:04

本文关键词： 土壤水力参数 SWAP模型 PEST 土壤含水量实际蒸散发　出处：《中国地质大学(北京)》2017年硕士论文　论文类型：学位论文

【摘要】：土壤水力参数的确定是农业水文模型应用的重要环节。近年来,数值反演方法被广泛的应用于土壤水力参数的估计。该方法虽然避免了人工试错法的费时费力,但也存在着局部收敛和“异参同效”等问题,获得的结果的准确性仍然存在争议。因此,研究土壤水力参数的数值反演,多方面的对优化的参数以及模拟结果进行评估,对数值反演方法的应用以及模型精度的提高具有重要意义。本文将农业水文模型SWAP(Soil-Water-Atmosphere and Plant)与数值反演程序PEST耦合,利用北京市通州区张家湾试验站的三个土壤剖面,首先进行SWAP模型的正向模拟,得到土壤含水量和实际蒸散发值,将这些值作为实际观测值并认为没有误差,用于反演模拟。其次,本文探讨了数值反演的单一目标优化。从土壤含水量观测误差,观测频率和观测点空间设置三个方面对估计参数的不确定性和模拟结果进行评估。结果表明,当只用土壤含水量来反演估计土壤水力参数时,土壤含水量观测误差应在37.5%以内,且土壤水力参数的不确定性随着土壤水分观测误差的增加而增加;土壤水力参数的反演对观测频率变化不敏感,土壤含水量的观测频率最好不要超过四周;观测点的增加对土壤含水量的拟合精度有所提高,当每层各增加一个点时,模拟结果最好,但如果条件不允许时,在第一层加观测点,对各层反演精度的提高贡献更大。在此基础上,考虑到单一变量的不可靠性,我们进一步探索数值反演的多目标优化。本文在应用土壤含水量的基础上,增加实际蒸散发量(ETa)作为另一反演变量,比较了在不同作物体系下,分别用土壤含水量(θ)、土壤含水量和实际蒸散发量(ETa)、实际蒸散发量(ETa)反演得到的各反演变量的模拟结果。结果表明,在假设的现实条件下(土壤含水量观测误差为10%,观测频率为一周,观测点为每个土壤层一个),只用ETa反演得到的结果均是最差的,尤其是对于土壤含水量的模拟;ETa的加入对反演变量模拟精度的提高并不明显,甚至在轮作时,会有反作用,这说明,在大多数情况下,只用土壤含水量进行参数反演就足够了,ETa对于土壤含水量和ETa的拟合效果的提高并没有太大作用。
[Abstract]:The determination of soil hydraulic parameters is an important step in the application of agricultural hydrological models. In recent years, numerical inversion method has been widely used in the estimation of soil hydraulic parameters. However, there are also some problems such as local convergence and "heteroparametric equivalence", and the accuracy of the obtained results is still controversial. Therefore, the numerical inversion of soil hydraulic parameters is studied, and the optimized parameters and simulation results are evaluated in many aspects. It is of great significance for the application of numerical inversion method and the improvement of model precision. In this paper, the agricultural hydrological model SWAP(Soil-Water-Atmosphere and plant is coupled with the numerical inversion program PEST, and the three soil sections of Zhangjiawan Experimental Station in Tongzhou District, Beijing, are used in this paper. First, the forward simulation of the SWAP model is carried out to get the soil moisture content and the actual evapotranspiration value. These values are regarded as the actual observed values and there is no error, so they can be used to inverse the simulation. In this paper, the single objective optimization of numerical inversion is discussed. The uncertainty and simulation results of the estimated parameters are evaluated from three aspects: the observation error of soil moisture content, the observation frequency and the space setting of observation points. When only soil water content is used to estimate soil hydraulic parameters, the observation error of soil water content should be less than 37.5%, and the uncertainty of soil hydraulic parameters increases with the increase of soil moisture observation error. The inversion of soil hydraulic parameters is insensitive to the change of observed frequency, and the observation frequency of soil water content should not exceed four weeks, and the fitting accuracy of soil water content is improved with the increase of observation points, when one point is added to each layer, The simulation results are the best, but if conditions do not permit, adding observation points in the first layer contributes more to the improvement of inversion accuracy in each layer. On this basis, considering the unreliability of a single variable, We further explore the multi-objective optimization of numerical inversion. Based on the soil moisture content, the increase of actual evapotranspiration (ETa) is used as another inversion variable to compare different crop systems. The simulated results obtained from the inversion of soil water content (胃), soil moisture content and actual evapotranspiration (ETaan) and actual evapotranspiration (ETaa) are obtained respectively. In the hypothetical realistic condition (the observation error of soil water content is 10, the observation frequency is one week, the observation point is one for each soil layer, the results obtained by ETa inversion are the worst. In particular, the addition of ETa to the simulation of soil moisture content does not improve the simulation accuracy of the inversion variables obviously, even in rotation, which indicates that, in most cases, Only soil water content can be used for parameter inversion is enough to improve the fitting effect of soil water content and ETa.
【学位授予单位】：中国地质大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S152.7

【参考文献】