BAYESIAN DISCRIMINATION OF HYDROLOGIC FORECASTING MODELS BASED ON THE KALMAN FILTER.
Academic Article
Overview
Additional Document Info
View All
Overview
abstract
The Kalman filter algorithm very well suits real time prediction of streamflow. The state of the system is assumed to be either the ordinates of the response function of the system or streamflows themselves. In the first case assumptions have to be made about the initial state of the system, the lag structure of the model and the covariance matrix of the measurement noise. In this paper the use of Bayesian theory is proposed to discriminate alternative assumptions on the values of these variables. Controlled and real world experiments were carried out to examine the performance of these discrimination criteria and the results were quite satisfactory.