论文部分内容阅读
由于地下水流动的发生环境通常非常复杂,通过稀少的钻孔资料难以对含水层特征进行准确的描述,而数据同化技术可以利用地下水位等动态数据反演出含水层的特征.为减小大尺度问题的抽样误差,提出了以确定性抽样技术为基础的卡尔曼滤波方法,讨论了确定性卡尔曼滤波方法在强烈非均匀介质和大尺度问题中的应用效果.研究结果表明:确定性卡尔曼滤波方法生成了唯一的样本集合,避免了传统集合卡尔曼滤波方法预测结果的不确定性;该方法能够缓解小样本条件下的系统方差快速衰减现象,并在强烈非均质介质中表现出良好的计算性能;结合局部化技术,确定性集合卡尔曼滤波方法能够很好地解决大尺度地下水系统的参数反演问题.
Because the groundwater flow is usually very complicated, it is difficult to accurately characterize the aquifer with scarce borehole data, and data assimilation techniques can use the dynamic data such as the groundwater table to retrieve the characteristics of the aquifer.In order to reduce the large-scale problem , A Kalman filtering method based on deterministic sampling technique is proposed and the application of deterministic Kalman filtering method in strongly inhomogeneous media and large scale problems is discussed.The results show that deterministic Kalman filtering The method generates a unique set of samples and avoids the uncertainty of the prediction results of the traditional ensemble Kalman filtering method. The proposed method can reduce the rapid degradation of the system variance under the condition of small samples and shows good performance in strongly heterogeneous media Combining with localization technology, the deterministic ensemble Kalman filter method can well solve the parameter inversion problem of large scale groundwater system.