结合模式性能和独立性加权的全球增暖1.5/2℃下中国区域气候的未来预估

Climate projection over China under global warming of 1.5 and 2℃considering model performance and independence

基于耦合模式比较计划第6阶段(CMIP6)中的全球气候模式的模拟结果,采用考虑模式性能和独立性结合(Climate model Weighting by Independence and Performance,ClimWIP)的加权方案进行中国区域气候的多模式集合预估及不确定性研究。结果表明,ClimWIP方案在历史阶段的模拟优于等权重方案,降低了多模式模拟的气候态偏差。温度指数的未来预估不确定性较大的区域主要集中在中国北方和青藏高原,而降水指数主要集中在华北和西北地区。ClimWIP方案的预估不确定性与等权重方案相比有所降低。ClimWIP方案预估的温度指数的增温大值区主要集中在中国北方和青藏高原;降水指数在西北和青藏高原增加最为显著。全球额外0.5℃增暖时,中国区域平均的温度指数变化更强,平均高于全球0.2℃,最低温在东北部分地区的额外增温甚至是全球平均的3倍;总降水额外增加5.2%;强降水额外增加10.5%。全球增暖2℃下,中国大部分区域温度指数较当前气候态增加可能超过1.5℃(概率>50%),在中国北方和青藏高原的部分地区增温超过1.5℃的可能性更大(概率>90%);总降水,强降水和连续干日在西北和华北增加幅度有可能超过10%、25%和-5 d(概率>50%)。

Under global warming,China is more vulnerable to the threat of extreme climate events.Studying the future climate change in China and providing more accurate future projections are of great significance for disaster prevention and mitigation,as well as policy-making in response to climate change.Based on the simulations of the global climate models(GCMs)from the Coupled Model Intercomparison Project Phase 6(CMIP6),we adopt a weighted scheme of Climate Model Weighting by Independence and Performance(ClimWIP)to carry out the multi-model ensemble constrain of mean and extreme temperature and precipitation over China region.Based on the performance evaluations of the constrained ensembles,the projected changes at the 1.5 and 2℃global warming under the SSP5-8.5 scenario are studied.Results show that the ClimWIP scheme has better performance when compared to the unweighted scheme,which reduce the climatology bias of ensemble.The spatial correlation coefficient between temperature indices and observations exceeds 0.98,and standard deviation ratios are close to 1.The spatial correlation coefficient between total precipitation(PRCPTOT)and heavy precipitation(R95P)with observations exceeds 0.92 and standard deviation ratios between 0.8 and 1.0.The regions with higher projection uncertainty are mainly in Northern China and Tibetan Plateau for the temperature indices,and in North China and Northwest China for the precipitation indices.The projection uncertainty by the ClimWIP scheme is reduced when compared with the unweighted scheme.The reduction is greater for temperature indices in Southern China and Tibetan Plateau,while precipitation indices show a significant decreased in uncertainty in Northeast China and northwest Xinjiang.Under 2℃global warming,the uncertainty of annual mean temperature(Tas),maximum temperature(TXx),and minimum temperature(TNn)in China is reduced by 19.2%,22.1%,and 17.8%,respectively,while PRCPTOT and R95P is reduced 3.3%and 4.7%,respectively.Regarding to the geographic distribution,ClimWIP scheme would see larger warming in Northern China and Tibetan Plateau for the temperature indices.More intense precipitation concentrate in Northwest China and Tibetan Plateau.Under an additional 0.5℃global warming,the temperature response in China region is stronger than that of global response,with an average higher warming about 0.2℃.The response of TNn in parts of Northeast China even more than three times additional warming.And there would be an additional increase about 5.2%and 10.5%for PRCPTOT and R95P,respectively.From the perspective of probability projection,at the 2℃global warming,the warming magnitude in most regions of China would be likely larger than 1.5℃compared to the current climate state(probability value>50%),and the probability in parts of Northern China and Tibetan Plateau would be much higher(probability value>90%).For the precipitation indices,the probability of wetter condition in Northwest China and North China would be larger,with a likely response magnitude exceeding 10%,25%and-5 days for PRCPTOT and R95P and continuous dry days(CDD)(probability value>50%).The ClimWIP scheme can reduce the uncertainty of future projections and provide more accurate future projections,and more model evaluation metrics such as trends and key physical processes can be considered in ClimWIP scheme in the future.Alternatively,multi-modal large ensembles and high resolution models can be used to improve the reliability of future projections.