Abstract: Based on the observed temperature data at 162 weather stations in the northeast China and five IPCC AR4 coupled climate models, the simulation abilities of global climatic model and multi model ensemble average for surface temperature were evaluated, and the variation of temperature in the future was projected under SRES B1, A1B and A2 three emission scenarios. The results show that the annual variation and spatial distribution of the surface temperature over the northeast China could be simulated accurately by the global climate models.  However, the simulated values are systematically lower than the observed values in the study area. The simulated value in summer is 1.16 ℃ lower than the observed value, while it is better than that in winter. The multi-model ensemble (MME) suggests that climate over the northeast China will has a warming trend in the middle and last period of 21st century under the three emission scenarios. The increasing amplitude of temperature is larger in the last period than in the middle period, so is in winter than in other seasons. Among the three scenarios, it is the largest under SRES A2 scenario and smallest under B1 scenario, respectively. The increasing amplitude of temperature increases from south to north, and the most obvious warming area is located in Xiaoxing'anling mountain of Heilongjiang province. The mean annual surface temperature in the northeast China would increase 2.39 ℃ under SRES B1 scenario, 3.62 ℃ under SRES A1B scenario and 4.43 ℃ under SRES A2 scenario in the last of 21st century, respectively.

Key words: Global climate model, Surface temperature, Projection