Optimized Flood Control in the Columbia River Basin for a Global Warming Scenario

Abstract

In transient rain-snow and snowmelt dominant river basins in the western U.S. global climate change is expected to produce systematic changes in winter climate and mountain snowpack resulting in higher flows in winter, earlier spring peak flows, and lower flows in spring and summer. The magnitude of these hydrologic changes is largely dependent on mid-winter temperature regimes, and the impacts will vary with elevation, latitude, and proximity to the ocean. In the context of managed flood control, the associated seasonal streamflow timing shifts are likely to disrupt the balance between flood control objectives and reservoir refill for water supply and other system objectives. To adapt to these hydrologic changes, evacuation of reservoirs for flood control may need to move earlier in the year as warming progresses, and refill may also need to begin earlier. In large rivers such as the Columbia River in the Pacific Northwest, where system-wide flood control storage is distributed between a number of reservoirs in different parts of the basin, rebalancing flood control and refill objectives in response to hydrologic changes poses a significant systems engineering problem. In this study we use the Variable Infiltration Capacity (VIC) macro-scale hydrology model implemented over the Columbia River basin coupled to a monthly time step reservoir optimization model (employing the HEC-PRM solver) to produce dynamic flood rule curves for a group of projects on the Columbia River main stem and tributaries. To test the methods a simple climate change streamflow scenario consistent with projections of warming for the second half of the 21st century is constructed. The scenario is based on a hydrologic simulation using observed precipitation and detrended historic temperatures consistent with a year 2000 temperature regime, with an additional 2 degrees C added. To calibrate the optimization model for the 20th century flow regime, the objective function is tuned to reproduce (using the ColSim reservoir simulation model) the current reliability of reservoir refill in the basin while providing comparable levels of flood control (at monthly time scales) to those in the historic record. The objective function is then held fixed, and an ensemble (88 years) of optimal flood rule curves for the future flow scenario are produced. These ensemble results are then summarized by constructing new flood control evacuation and refill schedules as a function of summer streamflow volumes, a framework which is consistent with current operational practice. These procedures provide an objective and well-defined method for rebalancing flood control and refill objectives in complex water resources systems in response to global wanning.