Although hydrological time series for different sites in a given region are usually correlated and that climate changes should have a regional impact on water resources, very little has appeared in the literature about multivariate change-point analysis. This paper generalizes the univariate Bayesian approach for the detection of a single shift in the mean level to study a change in the meanvector of a sequence of multivariate normal vectors. Two different problems are considered: the ®rst one is the estimation of the unknown regional change-point under the hypothesis that a shift occurred, while the second one is the overall assessment of change versus no change. This method is illustrated by an application to stream¯ow data series for six rivers situated in the Northern Que Âbec Labrador region.

Climate change is commonly evaluated as the difference between simulated climates under future and current forcings, based on the assumption that systematic errors in the current-climate simulation do not affect the climate-change signal. In this paper, we investigate the Canadian Regional Climate Model (CRCM) projected climate changes in the climatological means and extremes of selected basin-scale surface fields and its sensitivity to model errors for Fraser, Mackenzie, Yukon, Nelson, Churchill and Mississippi basins, covering the major climate regions in North America, using current and future climate simulations (2041-2070; A2 and IS92a scenarios) performed with two versions of CRCM.Assessment of errors in both model versions suggests the presence of nonnegligible biases in the surface fields, due primarily to the internal dynamics and physics of the regional model and to the errors in the driving data at the boundaries.In general, results demonstrate that, in spite of the errors in the two model versions, the simulated climate-change signals associated with the long-term monthly climatology of various surface water balance components (such as precipitation, evaporation, snow water equivalent (SWE), runoff and soil moisture) are consistent in sign, but differ in magnitude. The same is found for projected changes to the low-flow characteristics (frequency, timing and return levels) studied here. High-flow characteristics, particularly the seasonal distribution and return levels, appear to be more sensitive to the model version.CRCM climate-change projections indicate an increase in the average annual precipitation for all basins except Mississippi, while annual runoff increases in Fraser, Mackenzie and Yukon basins. A decrease in runoff is projected for Mississippi. A significant decrease in snow cover is projected for all basins, with maximum decrease in Fraser. Significant changes are also noted in the frequency, timing and return levels for low flows.

Daily minimum temperature increases more than the daily maximum temperature in the CCC GCM2 2xCO, equilibrium run (Zwierc and Kharin, 1998).