Drought Analysis with Reservoirs Using Tree-Ring Reconstructed Flows

This paper presents a drought study of the Sacramento River of California considering the presence of reservoirs. A total of 508 years (1485-1992) of annual flow data are used. The data prior to 1872 was reconstructed from tree-ring data of Oregon and California using statistical and stochastic techniques. Runs analysis is used to investigate the drought characteristics for three cases: (1) No reservoir; (2) a reservoir with a myopic release rule; and (3) a reservoir with an anticipatory release rule. It is found that: (1) The tree-ring reconstructed flows in this study are better than a previous data reconstruction; (2) the reconstructed data showed that other severe droughts occurred prior to 1872, when historical records were available, thus there are risks involved in relying solely on the historical data; (3) the Sacramento River basin experienced the most severe droughts around the 1580s and 1930s; and (4) the results of the cases with reservoirs have significantly longer interarrival times between drought events, a much smaller number of drought events, and much less severe water shortages and cumulative deficits in contrast to the results with no reservoir.