Development of optimized flood control rule curves for the Columbia River Basin in response to climate change and interannual climate variability

Water resources operating policies that attempt to preserve a balance between flood risk and other system objectives are usually predicated on assumptions of stationary climate conditions derived from historic streamflow records. Recent climate research has demonstrated, however, that systematic changes in temperature are occurring in response to global warming and other factors, and that regional climate varies on interannual and decadal time scales. These short- to long-term non-stationary elements of the climate system are investigated to determine if it is possible to improve dam operations for flood control. An optimization-simulation approach is developed and explored as an objective and well-defined procedure to improve system operations in response to the effects of global warming and by creating flood rule curves conditioned by seasonal to interannual climate forecasts at a monthly time scale and a daily time scale.;The optimization model is calibrated using eighty-six years of 20 th century monthly flow by tuning the objective function to reproduce the current reliability of reservoir refill without increase of flood risks. After the optimization model is calibrated, the same objective function is used to develop flood control curves for a global warming scenario which assumes an approximately 2°C increase in air temperature. Robust decreases in system storage deficits are simulated for the climate change scenario when optimized flood rule curves replace the current flood control curves, without increasing monthly flood risks.;A daily time step simulation model is used in Kootenai Basin portion of the upper Columbia Basin to test and refine the optimized flood control curves derived from monthly time step. This basin is chosen because releases from two major dams (Libby and Duncan) are constrained by water levels at Cora Linn Dam that were developed in a 1938 agreement of the International Joint Commission (IJC). For daily time step simulation which doesn't consider hydropower generation, general release from Libby and Duncan Dams is required during winter and spring to avoid the conflicts with the 1938 IJC agreement and required flood evacuation at Libby and Duncan Dams, while the maximum flood space and refill timing are kept from the monthly analysis. After modifying the evacuation schedule, the conclusions from the monthly time scale prove robust at daily time scales. Due to decrease of July storage deficits, additional benefits such as more revenue from hydropower generation and more July and August outflow for fish augmentation are observed when the optimized flood control curves are used for a climate change scenario.;The benefit of incorporating El Nino-Southern Oscillation (ENSO) information (cool, neutral, warm) in designing flood control curves is also investigated at a monthly time step. ENSO conditioned simulated flood risk and storage deficits under current operating policy are used to calibrate penalty functions for each ENSO state. ENSO conditioned rule curves using a unique objective function for each of the three ENSO states offer advantages in terms of storage deficits over current operations, and over optimal flood rule curves based on an unconditioned, single objective function.