文章摘要
刘琨,黄冠华.基于集合卡尔曼滤波法的二维土壤水流状态变量和参数联合估计[J].水利学报,2019,50(3):399-408
基于集合卡尔曼滤波法的二维土壤水流状态变量和参数联合估计
Joint state and parameter estimation of two-dimensional soil water flow model based on Ensemble Kalman Filter method
投稿时间:2018-10-26  
DOI:10.13243/j.cnki.slxb.20180953
中文关键词: 集合卡尔曼滤波  二维土壤水流  参数估计
英文关键词: Ensemble Kalman Filter  two-dimensional soil water flow  parameter estimation
基金项目:国家自然科学基金项目(51639009);国家重点研发计划(2017YFC0403301)
作者单位E-mail
刘琨 中国农业大学 水利与土木工程学院, 北京 100083
中国-以色列国际农业研究培训中心, 北京 100083 
 
黄冠华 中国农业大学 水利与土木工程学院, 北京 100083
中国-以色列国际农业研究培训中心, 北京 100083 
ghuang@cau.edu.cn 
摘要点击次数: 1857
全文下载次数: 186
中文摘要:
      集合卡尔曼滤波方法(EnKF)显式地考虑了模型输入、输出以及模型结构等因素的不确定性,近年来被广泛应用于水文模型参数估计研究中。本文基于EnKF方法开展二维土壤水流运动模型状态变量和参数联合估计研究,设计数值实验探究了在线源入渗条件下EnKF方法对粉壤土、壤土和砂壤土的饱和导水率和进气值参数的估计以及压力水头的同化效果,分析了观测点布置方式和观测点数量对同化效果的影响。研究结果表明,粉壤土条件下观测点垂向布置方式更好;壤土和砂壤土条件下,在0~30cm深土壤中水平向布置观测点可以得到较好的参数估计值。观测点水平向布置时应尽量靠近地表,同化系统可以有效地利用观测信息更新状态向量,参数更快地收敛于真值,但压力水头的同化效果仅限于一定深度的土壤。增加观测点数量可以有效地减小参数估计偏差,进而提高土壤剖面压力水头的预测精度。
英文摘要:
      The Ensemble Kalman Filter method (EnKF) explicitly considers the uncertainties such as in-put,output and model structure,and has been widely used in the parameter estimation problem in hydrolo-gy.The objective of this study was to extend the use of EnKF to state and parameters estimation of two-di-mensional soil water flow model.Numerical experiments were conducted to assess the performance of EnKF on soil hydraulic parameters estimation and pressure head assimilation under the condition of line source in-filtration for silt loam,loam and sandy loam.The influence of the arrangement of observation points and the number of observation points on assimilation results was further analyzed.The results show that the ver-tical arrangement of observation points can obtain better parameter estimation in comparison with that of hor-izontal arrangement for silt loam.The saturated hydraulic conductivity and shape parameter can be well esti-mated when the observation points are arranged horizontally in 0-30cm deep soil for loam and sandy loam.The observation points should be arranged as close as possible to the soil surface,so that the assimilation system can update the state as soon as possible and the parameters converge to the true value more quick-ly.However the assimilation effect on the pressure head is limited to a certain depth of soil.The predic-tion errors of soil pressure head in the areas with observation points are smaller than that of areas without observation points.Increasing the number of observation points can improve the prediction of soil pressure head and the estimation of soil hydraulic parameters.The result of this study indicates that the EnKF is an effective method for parameter estimation in two-dimensional soil water flow model.
查看全文   查看/发表评论  下载PDF阅读器
关闭