1989 年 33 巻 p. 211-216
Two methods, one uses the second-order filter and the other the extended Kalman filter are proposed, which are used to predict the concentration distribution and to identify the parameters of the one-dimensional stochastic convective-dispersion equation is discussed. It is shown that the proposed methods identify the parameters effectively and predict the concentration distribution accurately. It is clarified that the computational time for the second-order filter requires about two times longer than that for the extended Kalman filter, whereas the accuracy of prediction of the second-order filter is almost the same as that of the extended Kalman filter.