Assessment of Climate Change Impact on Flood Risk in the Red River Basin

Abstract:

In recent years, there have been a number of large spring floods in the Red River basin in the states of North Dakota and Minnesota, and in the Province of Manitoba. These recent floods have led to speculation that increased greenhouse gas concentrations may be changing precipitation patterns and impacting the frequency of floods. In this study, we investigate whether this is a reasonable assumption based on global climate model output. A regression model is developed to predict spring peak discharge on the Red River at a streamflow gage located at the border of the US and Canada. The predictor variables include antecedent fall precipitation used as a proxy for soil moisture at freeze-up, winter snow accumulation, and spring precipitation during the period of melt. Data from the North American Regional Reanalysis (NARR) have been used to calibrate the model. Bias-corrected projections from the CMIP5 GCM model ensemble are then used to predict floods in future years. The predicted floods are modeled using non-stationary frequency analysis. The use of multiple GCMs and multiple Representative Concentration Pathways (RCPs) allow for an estimate of uncertainty to be associated with the results. It is concluded that climate change will likely have an impact on floods in the Red River basin, but the uncertainty surrounding this assessment is rather large.