Link to CIG Homepage

Seminar Abstract

Paul Whitfield - January 29, 2002


Analysis and modelling of recent climate driven changes in streamflow in British Columbia and Yukon

During recent decades variations in climate conditions have occurred coincident to significant changes in streamflow in British Columbia. Modelling these variations provides an insight into how rivers and streams might behave in a changed climate. We investigate the ability of empirical downscaling models to resolve these changes using ensemble neural networks forced with large-scale atmospheric circulation conditions from the NCEP/NCAR atmospheric model reanalysis project. Five-day average streamflow data from British Columbia and the southern Yukon are predicted using atmospheric circulation and moisture fields from 1965-1986 as model inputs. Ability of the models to predict streamflow during the 1987-1998 test period is then evaluated using a combination of model performance statistics, comparisons between long-term averages, and results from non-parametric statistical tests. Correspondence between modelled and observed changes in long-term average streamflow is assessed using results from a recent study of regionalization of hydrologic change in Canada. In particular, the ability of the models to capture various aspects of the hydrologic regime in the different watersheds is demonstrated.

Back to top | CIG Home