近10年(2007—2016年)全球地表气温相对于工业革命前(1861—1890年)已上升约1 ℃,未来达到1.5 ℃温升阈值时的气候变化及其影响成为国际社会高度关注的问题。目前对未来温度的预估多依赖气候模式,但模式在区域气候预估方面尚存在较大不确定性。采用国际通用的“空间型标度(Pattern scaling)”方法,尝试基于1951—2005年历史温度观测资料,预估1.5 ℃温升阈值下全球区域地表气温相对于当前升温1 ℃的变化。由于未来气温变化的空间型可能与历史时期不完全相同,同时非线性因素亦可能令基于线性假设的空间型标度法出现偏差,故利用参加第五次耦合模式比较计划(CMIP5)的21个气候模式在4种典型浓度路径情景(RCP8.5,RCP6.0, RCP4.5, RCP2.6)下增暖空间型相对于历史时期(1951—2005年)的变化,对观测的空间型进行订正,并考虑非线性因素的影响。结果表明,全球平均温度继续上升0.5 ℃,达到1.5 ℃时,4种情景下预估的地表气温变化的空间型和增暖幅度接近。大部分陆地将升温0.6 ℃以上,北半球比南半球高约0.2 ℃,陆地比海洋高约0.3 ℃。预估中国区域升温0.7 ℃以上。RCP2.6下中国北部和中部升温明显高于其他情景。若不考虑订正方法的影响,在全球和区域尺度上,基于观测资料的空间型标度法预估结果的不确定性均远小于气候模式。

The global mean temperature during the recent decade (2007-2016) has increased above 1 ℃ relative to the pre-industrial period (1861-1890). The climate change and impact under 1.5 ℃ warming in the future have become a great concern in global society. Temperature projections, especially in regional scale, show great uncertainty depending on used climate models. Taking advantage of pattern scaling technique and observed temperature changes during 1951-2005, we tried to project the temperature changes globally under 1.5 ℃ threshold relative to current climate state, i.e. about 1 ℃ warming around 2007-2016. The projections of 21 climate models from the Coupled Model Intercomparison Project - Phase 5 under four Representative Concentration Pathways (RCP2.6, RC4.5, RCP6.0 and RCP8.5) were used to correct the assumptions in pattern scaling. Results showed that the geographical distribution and warming amplitude of surface air temperature changes under 1.5 ℃ threshold are similar in the four scenarios. Warming over most of the land would be above 0.6 ℃, 0.3 ℃ warmer than ocean. The Northern Hemisphere would be 0.2 ℃ warmer than the Southern Hemisphere. The temperature over China region will increase by 0.7 ℃. The warming in the Northern and Central China under RCP2.6 was obviously higher than that in the other scenarios. Ignoring the impact of correction method, uncertainty in temperature projection based on pattern scaling was much smaller than that in climate models, both in global and regional scales.