The reliability of single, historic estimates of reservoir capacity

Direct-supply reservoirs are the oldest form of artificial, man-made surface water impoundments. Hydrological analysis to determine the capacity-yield relationships of such reservoirs is often carried out empirically, using a record of observed streamflow data at the site with one of several techniques which come under the general rubric of “critical period methods”. The major problem with this approach is that it does not provide, explicitly, an estimate of the reliability of the reservoir system. Information on the reliability is extremely important for a system whose failure to meet targets is often accompanied by huge economic losses. In this paper an attempt is made, using Monte Carlo simulation techniques, to quantify the reliability of single historic estimates of reservoir capacity. The results show that the reliability is not unique but varies depending on the length of data record, the statistical characteristics of the inflow series and the yield level.