Use of Sequent Peak Algorithm Drought Severity Index and Hydroclimatic Reconstructions from Tree-Rings to Inform Water Supply Reliability Planning

Abstract:

California is in the midst of a severe drought with below average runoff since WY 2012. Within this context, many water resource managers are scrutinizing water supply reliability assumptions for planning studies. Severe droughts represent a relatively rare phenomenon, occurring only a handful of times within our limited 100-year period of watershed runoff records. Furthermore, droughts may have different runoff magnitudes and durations that inherently present a challenge for direct comparisons of one drought with another. We use the sequent peak algorithm as a drought severity index (SPADSI) that accounts for both drought magnitude and duration relative to an assumed minimum release policy and fixed level-of-development (LOD) demand modeling framework. The SPADSI allows direct, quantitative evaluation of different policy options for lessening drought severity where, for example, layering a customer rationing policy onto model results reduced the SPADSI for the historical 1976-77 drought from 520 to 450 thousand acre-feet (TAF) and 1987-92 drought from 650 to 415 TAF for 2015 LOD. A strong correlation (R2 = 0.96) between Mokelumne River watershed runoff and tree-ring hydroclimate reconstructions for neighboring American and Stanislaus watersheds from Meko et al. (2014) was the basis for an extended 1100-year historical reconstruction of Mokelumne Watershed annual runoff. The reconstructed runoff timeseries is used to investigate extended historical drought durations for the Mokelumne Watershed where shorter one- to three-year droughts are most probable durations (>90%) whereas longer duration droughts lasting as long as 10 years such as occurred in 1776-85 are also possible, though much less likely. Applying the SPADSI to the reconstructed runoff timeseries showed that recent droughts e.g. 1929-34, 1976-77, and 1987-92 are all relatively severe within this millennial context, falling on the distribution tail of the extended SPADSI dataset. These findings are consistent with Meko et al. (2014) in their analysis of other watersheds in the region. These findings and other insights from the reconstructed runoff timeseries along with the SPADSI provide valuable information for water resource managers evaluating water supply reliability assumptions for future drought planning efforts.