In this paper, we put forward a new method to reduce the calculation amountof the gain matrix of Kalman filter in data assimilation. We rewrite the vector describing the totalstate variables with two vectors whose dim...In this paper, we put forward a new method to reduce the calculation amountof the gain matrix of Kalman filter in data assimilation. We rewrite the vector describing the totalstate variables with two vectors whose dimensions are small and thus obtain the main parts and thetrivial parts of the state variables. On the basis of the rewrittten formula, we not only develop areduced Kalman filter scheme, but also obtain the transition equations about truncation errors, withwhich the validity of the main parts acting for the total state variables can be evaluatedquantitatively. The error transition equations thus offer an indirect testimony to the rationalityof the main parts.展开更多
文摘In this paper, we put forward a new method to reduce the calculation amountof the gain matrix of Kalman filter in data assimilation. We rewrite the vector describing the totalstate variables with two vectors whose dimensions are small and thus obtain the main parts and thetrivial parts of the state variables. On the basis of the rewrittten formula, we not only develop areduced Kalman filter scheme, but also obtain the transition equations about truncation errors, withwhich the validity of the main parts acting for the total state variables can be evaluatedquantitatively. The error transition equations thus offer an indirect testimony to the rationalityof the main parts.
