Optimized Flood Control in the Columbia River Basin for a Global Warming Scenario Se-Yeun Lee1, Alan F. Hamlet 1,2, Carolyn J. Fitzgerald3, Stephen J. Burges1 1. Dept. of Civil and Env. Engineering, UW, 2. CSES Climate Impacts Group, UW, 3. U.S. Army Corps of Engineers, Seattle District Overview Calibration Results Optimization Results Anticipated future temperature patterns will cause reduced spring snow pack, earlier melt, earlier spring peak flow and lower summer flow in transient rain-snow and snowmelt areas. In the context of managed flood control, these systematic changes are likely to disrupt the balance food control and reservoir refill in existing reservoir systems . For example, right figure shows conceptual diagram of operational changes associated with stream- flow timing shifts for a hypothetical dam located on the west slop of the Cascade Mountains. Without adjustment of refill schedules, the hypothetical dam does not refill for the altered streamflow timing simulated for a warming scenario. By moving the refill timing one month earlier, however, the hypothetical dam successfully refills for the altered flow regime. This illustration shows that refill timing and evacuation requirement for flood control may need to be modified to adapt to these hydrologic changes associated with global warming. This work poses a significant systems engineering problem especially for large water systems. To solve this problem, there is a need to develop an objective and well-defined method to maintain or improve the current Level of system performance for climate change scenarios for complex flood control system. Using the Columbia River basin (shown in left figure) as an example, an optimization-simulation procedure is used for restoring the balance between flood control and reservoir refill in complex, multi-objective reservoir systems in response to streamflow timing shifts under a simple global warming scenario. In comparison to current flood control curves for 20th century climate (20th_ Scen_CurFC), HEC-PRM derived flood control curves for 20th century climate (20th_Scen_HecFC) show lower storage deficits for major dams in Columbia River Basin and lower flood risks at flood check points: Bonners Ferry, Columbia Falls and The Dalles Dworshak storage (Dworshak Apr-Jul flow volume >2.6 MAF) a) 20th Century Climate b) Climate Change Scenario Apr Mar : Current Climate : + 2.25 oC plus adaption : + 2.25 oC No adaption + 2.25 oC b) a) Bonners Ferry Libby storage (Libby Apr-Aug flow volume >5.5 MAF) c) Columbia Falls The Dalles d) a) 20th Century Climate b) Climate Change Scenario Apr Apr Creating New Flood Control Curves The Columbia River Basin To develop continuous flood control curves, the maximum flood space for each water year is determined using the optimization model results. This is then expressed as a non linear function of seasonal flow volume occurring in each water year (see right upper figure). The overall seasonal shape of the rule curve and the typical refill timing at each project was established from the optimized results (with a monthly time resolution) by examining the 50th percentile value, and the nearest end of month date for the initiation of refill was used in the simulations (see right lower figure). This seasonal shape was then scaled to produce the maximum flood space required for each flow condition. Simulation Results Below figures show a) Simulated Storage Deficits for major dams in Columbia River Basin and Flood Frequency Curves at b) Bonners Ferry, c) Columbia Falls and d) The Dalles for simulated 20th century climate using current flood control curves (20th_Cent_CurFC) and the climate change scenario climate using current flood control curves (CC_Secn_CurFC) and HEC-PRM derived flood control curves (CC_Scen_HecFC). b) a) Tool and Method Climate Change Scenario The U.S. Army Corps of Engineers Hydrologic Engineering Center’s Prescriptive Model (HEC-PRM) is used as an existing monthly time step optimization model for the Columbia basin. Of all penalties representing all purpose of the Columbia River Basin, only flood control and reservoir refill penalty functions are used here to solve the water alloca- tion problem in terms of a balance between flood control and reservoir refill. A monthly time step reservoir simulation model for Columbia River Basin (ColSim) is used to test the performance of the new flood control curves inferred from optimization runs. Two penalty functions in the optimization model are calibrated for the 20th century flow regime using the procedure outlined in right figure. The objective function is then held fixed and optimized flood control curves for a climate change scenario are created. A simple climate change scenario was developed using VIC model (See right figure) to test the effectiveness of the approach for developing new flood rule curves that are appropriate for altered streamflow timing. The climate change scenario was created by removing the historic monthly temperature trends from the daily time step forcing data and then increasing temperature to approximately 2° C. The precipitation in each year is assumed to be identical to the unperturbed historical meteorological forcing data. These altered streamflows are then used to drive the reservoir optimization and simulation models. d) c) No Set up flood-and storage-related penalty functions Run optimization Infer flood control rule curves from optimization run (Optimized FCs) Compare simulated flood risks and storage deficits using Optimized FCs with those using Current FCs Apply Optimized FCs into simulation model Stop Do Optimized FCs show higher storage deficits than Current FCs? Do Optimized FCs show higher flood risk than Current FCs? Decrease storage penalties or increase flood penalties Yes Decrease flood penalties or increase storage penalties Conclusions Optimization studies provide an objective method for rebalancing flood control and refill objectives in complex reservoir systems in response to hydrologic changes. The changes in flood control rule curves are different for different projects, corresponding to different changes in flow volume and timing associated with warming in each sub basin. Optimized flood control rule curves show reduced flood evacuation and earlier refill timing; up to one month earlier for a climate change scenario, compared with 20th century climate. For the climate change scenario, optimized flood rule curves decrease storage deficits while providing comparable levels of local and system-wide flood protection in comparison to the current flood control rule curves. Left figures show average reservoir inflows to Dworshak and Libby dams corresponding to the historical record and the climate change scenario. Peak flows under the climate change scenario are reduced and occur earlier in the spring in comparison with the 20th century climate. The flow volumes differ markedly at the two sites.