Floods in a Changing Climate: A Case Study From the Red River Basin

Abstract:

In many parts of the world, the frequency of large flood events appears to have increased. Although it is impossible to associate any particular event with climate change, there is a general perception that climate change may be at least part of the reason for changes in the statistical distribution of floods. The Red River, sometimes called the Red River of the North, originates at the borders of North Dakota, South Dakota, and Minnesota and flows north through the Province of Manitoba before emptying into Lake Winnipeg. There have been several severe spring floods on the Red River in recent years, creating speculation that increased greenhouse gas concentrations are changing the frequency of floods. In this study, we investigate whether this is a reasonable assumption based on global climate model output. A regression model has been 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 CMIP5 GCM model ensemble are used to determine change factors for the predictor variables. The modified predictor variables are then used to produce scenarios of floods in a changed climate. The use of multiple GCMs and multiple Representative Concentration Pathways (RCPs) allows for an estimate of uncertainty to be associated with the results.