Predicting Basin-Wide Tailwater Temperature Patterns by Combining Mechanistic and Statistical Water Quality Modeling

A simplified method for predicting water temperature patterns downstream of a peaking hydropower dam is described for the Fort Randall Dam tailwater of the mainstem Missouri River as part of studies by the U.S. Army Engineer Division, Missouri River (MRD) to revise and update system-wide reservoir operations manuals. The method has two steps. First, a one-dimensional longitudinal, time-varying, riverine water quality model, CE-QUAL-RIV1Q, simulated downstream water temperatures at about 1.0 mile intervals under upstream boundary conditions and meteorology expected for future operations. Second, step one output was statistically evaluated using a combination of linear and nonlinear regression. Nonlinear regression was used to estimate the value of two coefficients providing the best fit of model predictions to the logistics equation for each of 108 scenarios. The coefficients were then linearly regressed against the boundary conditions of the scenarios. The resulting linear regression equations were coded into a short program and used as a simplified procedure to predict downstream water temperatures. Independent verification data shows that the screening model consistently matched water temperature predictions of the mechanistic model to within 1.0°C.