论文部分内容阅读
集合卡尔曼滤波(En KF)能够便捷地根据观测信息改进土壤水的预测精度,但在非线性问题中存在不一致性问题:当同时更新变量(水头或含水率)和非饱和水力参数时,En KF会导致参数和变量之间不再服从Richards方程关系。本文以Borden试验场的饱和渗透系数数据为基础,构造土柱试验,研究了不一致性对土壤水数据同化带来的破坏以及相应的迭代型解决方案。研究结果表明:在非均匀土壤中,En KF可能会引发强烈的不一致性,对含水率预测和参数估计的精度造成破坏;不一致性的峰值位置与水流锋面保持一致,且受边界条件影响;对于同类型的观测与待求参数,不一致性一般是随着观测点数量与待求参数数量的比例的增大而减小;当观测与水头或含水率之间具有强烈的非线性时(如强烈的干湿交替),推荐采用能在保证计算效率的前提下有效降低不一致性影响的CEn KF或MREn KF方法;当观测信息相对充足,不一致性可忽略时,En KF方法更具优势。
EnKF (Ensemble Kalman Filter) can easily improve the prediction accuracy of soil water based on observational information, but there is an inconsistency problem in nonlinear problems. When En (variable head (water head or moisture content)) and unsaturated hydraulic parameters are updated simultaneously, En KF causes the parameters and variables to no longer follow the Richards equation. Based on the data of saturation permeability coefficient of Borden test site, this paper constructs the soil column test, and studies the damage caused by the inconsistency on soil water data assimilation and the corresponding iterative solution. The results show that: Inhomogeneous soil, En KF may lead to strong inconsistencies, the accuracy of water content prediction and parameter estimation damage; inconsistencies peak position consistent with the water front, and the boundary conditions; for The same type of observations and the parameters to be determined, the inconsistency generally decreases with the increase of the ratio of the number of observation points to the number of parameters to be solved. When there is a strong nonlinearity between the observation and the water head or moisture content , It is recommended to use the CEn KF or MREn KF method which can effectively reduce the effect of inconsistency while ensuring the computational efficiency. En KF method is more advantageous when the observation information is relatively sufficient and the inconsistency is negligible.