Real-Time Flood Forecasting in the River Section with Ungaged Tributaries

Abstract

In the contribution, the solution of the problem of the adaptive flood forecasting of tributaries of large-scale water resource systems is described. The water resource system is divided into interbasins limited by significant specific profiles. The basic dynamic model of interbasin behaviour is formulated by the state space model and solved by the adaptive Kalman filtration algorithm. In view of fast changes in the behaviour of hydrological processes, the results of adaptive filtration mostly based on the analysis of innovation sequence exhibit typical delays and/or instability which depreciate hydrological forecasts. In real time operation therefore the following procedure can be advantageous: /i/ to estimate the type of situation in which the hydrological process is positioned, /ii/ to use, for the given type of hydrological situation, the corresponding basic model and the values of the unknown parameters, and to let the adaptive algorithm operate around these values till the subsequent change in the type of hydrological situation sets in. The problem of the effective utilization of inaccurate and incomplete information on the behaviour of hydrological processes in ungaged interbasins is solved by the two-level estimator. The algorithm is applied to the real-time estimation and forecasting problem in the Czechoslovak section of the Danube river and other Czechoslovak rivers.