论文部分内容阅读
集合卡尔曼滤波(EnKF)算法在地下水数据同化领域中的应用受到了越来越广泛的关注。作为同化系统的重要组成部分,观测数据的空间/时间密度的配置直接影响滤波运算结果。本文构造了一个理想二维地下水流算例考察空间/时间密度对传统EnKF和局域化EnKF的影响。研究结果表明:随着空间密度的增大,局域化EnKF运算精度增高,而传统EnKF运算精度无此改进倾向。总体趋势上时间密度增大使EnKF运算精度增高,但对不同数目的观测井方案,这种精度增高的幅度有所变化,观测井越多,增高越不明显。由此得出结论:局域化改进EnKF能够有效同化更多的观测井数据,给出更精确的结果;模拟初期水头变化波动较大,观测数据价值较高;在一定时间密度配置下,低空间密度局域化EnKF运算精度可以接近甚至超过高空间密度配置。
The application of ensemble Kalman filter (EnKF) algorithm in the field of groundwater data assimilation has drawn more and more attention. As an important part of the assimilation system, the spatial / temporal density of the observed data directly affects the result of the filtering operation. In this paper, an ideal two-dimensional groundwater flow example is constructed to investigate the effect of space / time density on traditional EnKF and local EnKF. The results show that with the increase of spatial density, the accuracy of local EnKF operation increases, while the traditional EnKF operation accuracy does not improve. The overall trend of increasing the time density EnKF operational accuracy increased, but for different numbers of observation well program, the accuracy of the increase in the magnitude of change, observation wells, the more obvious increase. It is concluded that EnKF can effectively assimilate more well data and give more accurate results. In the early stage of simulation, the variation of water head is large and the value of observation data is high. Under a certain time density configuration, Spatial Density Localization EnKF computational accuracy can approach or exceed the high spatial density configuration.