Global Climate Model Simulated Hydrologic Droughts and Floods in the Nelson-Churchill Watershed

Abstract:

There is uncertainty surrounding the duration, magnitude and frequency of historical hydroclimatic extremes such as hydrologic droughts and floods prior to the observed record. In regions where paleoclimatic studies are less reliable, Global Climate Models (GCMs) can provide useful information about past hydroclimatic conditions. This study evaluates the use of Coupled Model Intercomparison Project 5 (CMIP5) GCMs to enhance the understanding of historical droughts and floods across the Canadian Prairie region in the Nelson-Churchill Watershed (NCW).

The NCW is approximately 1.4 million km² in size and drains into Hudson Bay in Northern Manitoba, Canada. One hundred years of observed hydrologic records show extended dry and wet periods in this region; however paleoclimatic studies suggest that longer, more severe droughts have occurred in the past. In Manitoba, where hydropower is the primary source of electricity, droughts are of particular interest as they are important for future resource planning.

Twenty-three GCMs with daily runoff are evaluated using 16 metrics for skill in reproducing historic annual runoff patterns. A common 56-year historic period of 1950-2005 is used for this evaluation to capture wet and dry periods. GCM runoff is then routed at a grid resolution of 0.25° using the WATFLOOD hydrological model storage-routing algorithm to develop streamflow scenarios. Reservoir operation is naturalized and a consistent temperature scenario is used to determine ice-on and ice-off conditions. These streamflow simulations are compared with the historic record to remove bias using quantile mapping of empirical distribution functions. GCM runoff data from pre-industrial and future projection experiments are also bias corrected to obtain extended streamflow simulations.

GCM streamflow simulations of more than 650 years include a stationary (pre-industrial) period and future periods forced by radiative forcing scenarios. Quantile mapping adjusts for magnitude only while maintaining the GCM’s sequencing of events, allowing for the examination of differences in historic and future hydroclimatic extremes. These bias corrected streamflow scenarios provide an alternative to stochastic simulations for hydrologic data analysis and can aid future resource planning and environmental studies.