Runoff forecast error uncertainty and some ways it can affect snowmelt water scheduling decisions in the Sierra

Forecasting runoff and scheduling reservoir releases for PG&E's hydroelectric system, located in California's Sierra and Southern Cascade Watersheds, requires that watersheds be subdivided into subbasin reaches between diversion dams to facilitate computation of daily unimpaired sidewater inflow for each subbasin reach. Computation of daily unimpaired subbasin sidewater flow allows hydro schedulers to construct a feasible operation plan that optimizes water releases from upstream reservoirs for purpose of maximizing hydroelectric value while satisfying multiple constraints such as are often imposed for instream flows, whitewater rafting, planned maintenance, and recreation. What occurs with subbasin daily flow computation is large variance among successive daily flows. This variance is seldom explainable by hydrologic events alone. However an explanation can be found, when one reviews propagation of error uncertainty and daily storage change taking place at upstream reservoirs. In addition the April through July seasonal runoff forecast has a tendency to carry the full standard error of the estimate forward into the final ablation phases of snowmelt. At that point, the relative magnitude of the error compared with remaining flow yet to come causes large error uncertainty that often affects optimal final seasonal filling of reservoirs.