Abstract

The world's reservoirs are losing capacity at a rate of 0.5-1%a year due to sedimentation processes. The strategies to reverse this trend must include an accurate estimation of sedimentation rates in terms of Specific Sediment Yield (SSY). This research develops an empirical and statistical model based on data from the CEDEX (Spanish Studies and Experimentation Centre for Public Works). From an initial number of 131 reservoirs studied in the period 19,672,004, a group of 26 reservoirs with medium-sized basins (750 to 1750 km(2)) was selected for analysis. Reservoir catchments were described with 11 explanatory variables, representing the production, transport and deposition of sediment, although the calibrated model considers only six of these variables: Fournier Index (Rainfall Torrentiality), Drainage Length, Reservoir Coefficient (relation between reservoir capacity and area), C (USLE Land Cover Factor), Yearly Average Rainfall and Slope. SSY and the explanatory variables were transformed during the calibration process, and the resulting model shows a non-linear relation between them. Compared to other models calculated with CEDEX data with a determination coefficient of between 17% and 80%, this model has a determination coefficient of 84%, is statistically consistent, validated by means of a jackknife analysis and contrasted with other models. However, the model is not cross validated with information on additional reservoirs, and shows substantial uncertainty and instability deriving from the definition of the explanatory variables and the quality of the data set, so extrapolation to other reservoirs is only possible under supervision and local calibration. However, the model reveals a strong non-linear relation between land cover, rainfall amount and torrentiality, slope gradient, drainage length and reservoir depth with SSY, and provides valuable information for exploring the effect of watershed alterations on sedimentation. (C) 2019 Elsevier B.V. All rights reserved.