ECONOMIC OPTIMUM RECORD LENGTH

Abstract

Uncertainty in hydrologic design parameters is reflected as an increase in expected project costs. The Bayesian or statistical decision approach produces a minimum-cost decision for a specific design, yielding the expected opportunity loss, EOL, a measure of the uncertainty inherent in the decision process. The uncertainty stems from the fact that the population value of the design parameters is unknown. Additional information can be obtained by collecting more data, and the value of these data is measured by the decrease in EOL. But there are costs involved in getting additional data: the cost of continued data collection and the cost of delaying the construction of the project (benefits foregone). The expected economic optimum record length is defined at the point where the marginal benefits of additional data are equal to the marginal costs of obtaining those data. The theory is applied to sediment load data used to design the sediment storage portion of a reservoir. Based on an economic efficiency criterion, the expected optimum record length is found to be 12 years, given that 5 years of data are available at the time of the analysis. Objectives like environmental quality and social benefits are disregarded in this analysis.