| Popis: |
The present chapter is an overview of our recent works in the area of water resources modeling for ends of prediction in Algeria. It focuses on a particularly interesting type of models that can support not only the stochastic nature of the hydrological processes but also their temporal variability as well as the nonlinear character of the hydrological system. Such models are mostly required in water resources design and management because they provide a helpful tool for decision and policy makers in Algeria. The objective here is to showcase some of our recent results regarding the extent of applicability of discrete Kalman filter (KF) to the modeling and prediction of water resources in Algeria. For this end, two important hydrological variables have been investigated: rainfalls in the Cheliff watershed and stream flows in the northern Algeria. The corresponding time series data in the annual and monthly scales have been utilized to build some mathematical models based on the discrete KF structure. For each of the two hydrological variables, annual and monthly models have been set for prediction. In all cases, the obtained model is an online prediction operation where the variable predictor is not bound to time or space, but rather adapts itself recursively to evolving conditions related to any climatic and physiographic factors in the study area. The obtained models provide with predictions that respect the variables stochastic character and undertake also the nonlinear nature of the hydrological system. Moreover, the obtained results are optimum in more ways than one: (1) To be initiated, calculations need only a minimum objective information. (2) Recursivity in the time domain gives to the model an adaptive character that can be used in real-time forecasting. (3) The prediction covariance error is provided exactly at each iteration calculations. The obtained results for either the spatial or the temporal variability of the considered variables are satisfactory, and the associated errors are quite acceptable. |
