A Retrospective Analysis of Historical Reservoir Operations in Conus

Abstract

In the contiguous United States (CONUS), there are over 52,000 reservoirs ranging from 0.5 to 243 meters high. Collectively they hold 600,000 MCM of water. These structures have dramatically affected the river dynamics of every major watershed in the country. The operations of these structures depend on a variety of climatic, anthropogenic and reservoir drivers, however, the lack of a comprehensive national dataset makes interpolation of the importance of these drivers hard to analyze. To start to address this gap we have assembled a national dataset of historical reservoir operations called ResOpsUS. Here we present the dataset, titled ResOpsUS, and evaluate static reservoir properties against the existing global reservoir database, GRanD. We then evaluate trends in operations by region and reservoir purpose. ResOpsUS contains reservoir inflow, outflow, and storage values for 679 reservoirs across CONUS. These reservoirs cover 55% of the total storage in large (greater than 1 MCM) reservoirs across the US. We show that observations agree well with the static reservoir properties reported in Grand; however, there are some systematic biases. Specifically, maximum storage capacities align with the maximum storage of irrigation reservoirs, but not necessarily flood control reservoirs, while releases of flood control reservoirs most closely align with naturalized flows. Analysis of transient operations highlighted regional and purpose-based differences in strength of the relationship between precipitation and storage and reservoir releases. Most specifically, it demonstrated that regions with large correlations between inflow and precipitation are more likely to have low correlations between cumulative storage and releases. These regional trends are most closely associated with the main use of reservoirs, however, some areas, specifically Texas, see correlations that are more nuanced due to multipurposed reservoirs.