Abstract: Using on the CN05.1 monthly average temperature observation data set from 1961 to 2014 and the output data from 19 global climate models from Coupled Model Intercomparison Project Phase 6 (CMIP6), the simulation ability of CMIP6 models on the climatology spatial distribution and interannual variability of temperature in Southwestern China was systematically evaluated by means of Taylor diagram, Taylor index and interannual variability skill score. The variation characteristics of future temperature in this area were predicted under SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5 scenarios. The results showed that: (1) compared with other seasons, most CMIP6 models had the best performance in simulating the spatial distribution of autumn temperature climatology during 1961-2014; and CMIP6 models underestimated the interannual variability of seasonal and annual average temperature. Among the 19 models, the best models simulated the temperature in Southwestern China were ACCESS-CM2, CMCC-CM2-SR5 and CMCC-ESM5. (2) The multi-model ensemble mean(MME) of 3 best-fit models simulated the climatology spatial distribution and interannual variability of average temperature better than the MME of 19 models. (3) Compared with the multi-year average temperature observed in the same period during 1961−2014, the seasonal and annual average temperature in Southwestern China showed an upward trend in the future under the four climatic scenarios, seasonal and annual average temeprature increased by 0.94−3.48℃. Under the four scenarios, the increase of average temperature in summer was the largest(2.17−3.48℃) and the interannual fluctuation range was the smallest, the increase of temperature in winter was the smallest(0.94−2.24℃) and the interannual fluctuation range was the largest. (4) In the early of 21st century, there was little difference in the increase of seasonal and annual average temperature under 4 scenarios. During the middle of the 21st century, the upward trend of seasonal and annual average temperature in high radiation forcing scenarios was gradually larger than that in low radiation forcing scenarios. (5) Under the four scenarios, the anomaly values of multi-year average temperature at the early (2015−2034), middle (2045−2064) and end (2081−2100) period of 21st century and the historical(1961−2014) observed temperature showed the spatial distribution characteristics that the northwest was greater than southeast of this region, and the high latitude and high altitude areas were greater than the low latitude and low altitude areas. With the passage of time, at the end of 21st century, the temperature anomaly in the same region was significantly higher under high forcing scenarios than that in low forcing scenarios.

Key words: CMIP6 models, Southwestern China, Surface air temperature change, Evaluation, Projection