An implicit but very important assumption for downscaling large-scale projection to regional and local scales is that the projected large-scale change represents future changes in the climate due to anthropogenic forcing. This assumption, unfortunately, has rarely been verified. Without proper verification, it is difficult to gauge the confidence on the projected future change. In this paper, we present a framework in which model projected anthropogenic large-scale climate changes are verified with and constrained by the observed climate change, improving the confidence in the projected future changes. Using simulations from 14 GCMs participating the IPCC 4th assessment, observations over 400 stations in Canada, over 700 stations in China, and 10 stations in the Republic of the Congo, we found that influence from external forcing is clearly detectable in the mean temperatures for all three countries and that external forcing influence may have also manifested at local level. We found that external forcing may have increased the risk of the record annual mean temperatures that mostly occurred in the past two decades for more than three times in those countries. By using model projected future changes in the regional mean temperature as predictors, we found that the current 50-yr return annual mean temperature will occur much more often in the future, becoming a 5 to 10 years event in Canada and China by 2025 and 5 or less years event in Congo by 2025.