Predictability of runoff in the Mississippi River basin

Prediction of streamflow is the essential challenge of hydrology. The large and growing economic costs associated with flood and drought events provide the motivation for this study: to investigate opportunities for improving long-lead (monthly to seasonal) runoff prediction over large continental areas; and to evaluate the potential impacts of these improvements on water resources management. Potential runoff predictability as used in this study derives from climate signals and from initial conditions—soil moisture and snow water equivalent at the beginning of the forecast period. Because long-term spatially distributed observations of soil moisture and snow water do not exist, a 50-year, hydrologically consistent data set of observed and derived surface energy and moisture fluxes and state variables was derived for the continental U.S. This data set provided the basis for investigating relative influences of initial states of the climate signal, snow water content, and soil moisture on long-lead predictability of runoff across the Mississippi River basin. The potential value of runoff predictability for water management was investigated using a simulation model of the Missouri River main stem reservoirs. In the Mississippi basin, climate indicators provide a small but significant source of winter runoff predictability through a lead time of three months. Soil moisture provides the dominant source of runoff predictability at lead times of one to two months over most of the Mississippi basin, except in the snow dominated mountainous areas of the west. For smaller sub-areas, runoff predictive skill exists through a lead of two seasons, longer than is currently used in operational forecasting. There was a very small difference in Missouri River system hydropower benefits associated with perfect forecast skill, due primarily to the system's large storage capacity relative to inflow. An investigation of the effect of prediction skill relative to reservoir size found a generally inverse relationship, which is which is consistent with various previous studies. A hypothetical reduced-volume system showed greater sensitivity to runoff predictability; with knowledge of the climate state, and snow and soil moisture initial states providing an increase of {dollar}6.8 million in annual hydropower benefits, about two percent of total annual hydropower revenues.