GC41E-0635:
Taking climate change into estimation of long-term flood risks: A case of Devils Lake of North Dakota, USA

Thursday, 18 December 2014
Gehendra Kharel and Andrei Kirilenko, University of North Dakota, Grand Forks, ND, United States
Abstract:
Terminal lakes are heavily impacted by regional changes in climate. Devils Lake (DL) is a terminal lake located in the northeastern North Dakota of the US. Since 1990, following a shift in regional precipitation pattern, DL has encountered a 10 m water level rise, with over 400% increase in surface area and 600% increase in water volume, costing over $1.5 billion in mitigation. Currently, the lake is <1.5 m from spillover level to the nearby Sheyenne River with potential negative consequences for downstream water quality and flooding. Recently, the artificial outlets have been constructed and operated to divert DL water to the Sheyenne River amid legal and political pressure. Outlet construction however did not take into consideration possible changes in local climate. We modeled the DL basin (~9,800 km2) hydrology using the Soil and Water Assessment Tool (SWAT) and estimated future water levels of DL for different outlet scenarios under three Intergovernmental Panel on Climate Change (IPCC) SRES scenarios (A1B, B1 & A2) for 2020s and 2050s. We evaluated model performance by comparing SWAT simulated daily streamflow outputs against the observed streamflow data recorded at 6 USGS water gauge locations within the basin. Future climate conditions in the region were estimated by combining historical weather data (1981-2010), 15 CMIP3 General Circulation Model projections from the IPCC data center, and stochastic downscaling methodology (LARS-WG). Our results indicate significant likelihood (7.3% ̶ 20.0%) of uncontrolled DL water overspill in the next few decades in the absence of outlets, with some members of GCM integration ensemble carrying over 85.0% and 95.0% overspill probability for 2020s and 2050s respectively. However, full-capacity outlets show radical reduction in overspill probability to partially mitigate the flooding problem by decreasing the average lake level by approximately 1.9 m and 1.5 m in 2020s and 2050s. Moreover, had there been outlet operation from the beginning of the flood episode since 1990s, not only the future overspill risks but also the current flooding extent would have been reduced significantly (Fig. 1).