Établissement d'un modèle mathématique hydrologique servant à la prévision et l'annonce des crue du Rio Guadalquivir (Espagne)

Construction of a hydrological mathematical model for a Rio Guadalquivir flood warning and forecasting study

Ingénieurs à la SOGHEAH, Grenoble.

Résumé

1. Basic model design concepts. The object of the investigations was to find ways of protecting Seville against floods by the following means : (i) Flood warning by calculation of river level rise propagation ; (ii) Medium-term flood forecasting by a rainfall/runoff transformation calculation for the tributary basins ; (iii) Flood protection by suitable structures of the river bed and floodable area, and the effect of such works on flow conditions ; (iv) Flood control by appropriate dam operation, based on foreseeable inflows at the dams. Simulation of flood formation and propagation by a hydrological mathematical model is the means required to ascertain the potential effectiveness of hydraulic works against floods. If combined with an economic probability study of the flood prevention works and an operational research study of optimal dam operation, the model would be an important part of the means required in planning hydraulic development in the Guadalquivir basin. 2. Progress and results to date. The first steps in the study, which started in 1964, were to collect and prepare any available data on rainfall, floods, dams and topography, to evaluate frictional resistance to flow, to list structures causing local head losses, etc. The basic propagation model was constructed by the implicit solution method for the Saint Venant equations. The model represents a 400 km stretch of the Guadalquivir from the Dona Aldonza dam upstream to the Alcala del Rio dam downstream. The model was adjusted by running a certain number of observed floods on it ; it can be used to simulate major floods all the way down to the Alcala del rio dam if the upstream and downstream tributary flow rates are known. The model adjustment showed up gaps in the data and helped to guide the rehabilitation of the hydroclimatological observation network in the basin. As a result, the construction and equipping of 32 gauging stations and 45 detailed shower observation stations was put in hand and a field measurements programme was organised. After systematic application of shower data already available from 405 stations, 58 were picked out as representative for the determination of critical rainfall indices for flood warning purposes, enabling the Seville flood warning system to be set off in good time (floods in the Seville region are apt to assume quite alarming proportions at times). Systematic tests carried out with a view to studying flood attenuation by storage dams on tributaries showed that present dam operating rules (calling for some storage space to be kept available in the reservoir and the latter to be made to absorb floods fully from the moment they start until the reservoir is full) are already giving good results. In practice, however, optimum attenuation of the Guadalquivir flood will not be ensured until the inflows at certain dams ont the tributaries can be predicted from the rainfall data. Although the field measurements programme is only just starting, it has already yielded data which have enabled a start to be made with a study of daily dam inflow prediction from rainfall data. A general rainfall/discharge transformation method has been produced, and further field measurements are expected to provide the data required to establish final values for the parameters in the prediction formulae. Consideration of possible ways of developing a simple practical flood forecasting instrument from the present hydrological model has led to the adoption of the more expeditious Muskingum calculation method, which is based on volume analysis and can be ajusted by the basic model. This method seems to be better matched to the flood forecasting problem and could be applied by digital computation requiring no more than a computer of limited capacity, or by an electrical or hydraulic analogy system. 3. Possible subsequent lines of investigation. Further investigation could be confined to development of a practical flood forecasting and warning system, but the hydrological model has also been designed to serve for the construction of an economic study model if required. In any case, there is urgent need for fresh hydrological data, and priority should be given to a thorough field measurements programme. Requirements for a study confined to forecasting will be to put the finishing touches to a practical flood forecasting system, to establish final values of the parameters in the inflow prediction method, and to study organisation of the forecasting and warning network and the necessary data-transmitting and processing facilities. For a study with a wider scope, the basic model (now only calibrated for major floods) will require fine adjustment in order to provide a suitable means of simulation for use in studies of a general nature.