Le soutien des étiages en Loire à l'aide de réservoirs situés dans le haut bassin. Application au barrage de Naussac

Upper valley reservoirs to maintain adequate low-water discharge in the river Loire The Naussac river barrage

Ingénieur hydraulicien à l'Agence Financière de Bassin Loire-Bretagne

Résumé

The river Loire has a very irregular regimen which sometimes results in an extremely low water level in the river. To remedy the situation the Loire-Brittany Development Office has produced a low-water discharge make-up programme, the first link of which is the Naussac barrage on the river Allier, a tributary of the river Loire. The investigation model allows for the following requirements : 1. Discharge of 8 m3/s M Vieille Brioude, 110 km down-river from the reservoir. 2. Highest possible discharge at Gien, 470 km from Naussac. 3. Reservoir management and operation to be consistent with development of tourism. 4. Reservoir to be built for full storage requirement (190 Mm3) right from the start and to be filled by diverting flow under gravity from the river Chapeauroux and pumping supplies from the river Allier. At the beginning of each month of the considered low-water period an attempt is made to predict the mean flow for the rest of the period, allowing for a 95 % security level. By comparing the predicted figure with observed past discharge distribution the severity of future low-water conditions can be predicted (probability that these will not be exceeded). From this probability and the amount of storage remaining in the reservoir (i.e. the theoretical deficiency that can be covered), an. "available discharge" target can be established for Gien for the considered period (see method described in section 3). A short-term forecast is then made, taking the transfer time of supplies released, from which water release requirements are calculated. Transfer times are calculated by cross-autocorrelation. Data from fifteen climatological and six river-gauging stations were used for these forecasts. Three coefficients were calculated to check the results: 1. a "release effectiveness" coefficient (KE); 2. a "deficiency coverage" coefficient (KC); and 3. a "storage utilisation" coefficient (KU). The model simulation covered the period from 1936 to 1969. Its results can be considered satisfactory, since it was found that deficiency coverage averaged 83 % and that 89 % of the releases were effective. A brief description is also given of how the scheme will operate when the second link (the Villerest barrage on the Upper Loire) becomes operational. In addition to sustaining adequate low-water discharge, this barrage will also provide effective flood control.