大气科学学报2026，Vol.49Issue(1)：94-108,15.DOI:10.13878/j.cnki.dqkxxb.20251129001
近百年中国东部持续性冷暖事件的检测归因与约束预估
Detection,attribution,and constrained projections of persistent cold and warm events over eastern China during the past century
摘要
Abstract
In recent decades,eastern China has experienced increasingly frequent persistent extreme temperature events characterized by long duration and widespread impacts,posing substantial risks to socioeconomic systems,public health,and energy security.While previous studies have primarily focused on the frequency and intensity of temperature extremes,detection and attribution analyses of persistent cold and warm events remain limited,partic-ularly at the centennial scale,owing to the scarcity of long-term,high-quality daily observational records.To ad-dress this gap,this study conducts a comprehensive detection,attribution,and constrained projection analysis of persistent cold and warm events over eastern China during 1901-2020,quantified using the cold spell duration index(CSDI)and warm spell duration index(WSDI).The analysis is based on multiple observational datasets,with the newly developed century-long China homogenized daily surface air temperature dataset(CUG-CMA)serving as the primary observational source,complemented by verification using the high-density OBS-1951a sta-tion dataset and three century-scale reanalysis/reconstruction products:the ECMWF Reanalysis of the 20th Century(ERA-20C),the NOAA 20th-Century Reanalysis(NOAA-20CR),and the Berkeley Earth surface tem-perature dataset(BEST).Multi-model simulations from the Coupled Model Intercomparison Project Phase 6(CMIP6),including historical all-forcing(ALL),single-forcing(greenhouse gases,GHG;anthropogenic aerosols,AA;and natural forcing,NAT),and future SSP2-4.5 scenario experiments,are employed.The robustness of attribution results across different periods is further assessed using a"perfect model"framework based on large-ensemble simulations from CanESM5 and MIROC6,followed by observationally constrained projection ana-lyses. The results show pronounced asymmetric changes in persistent cold and warm events over eastern China dur-ing 1901-2020.CSDI exhibits a significant decreasing trend of-13.50 days per century,while WSDI shows an increasing trend of 5.18 days per century.Both indices display a coherent temporal evolution characterized by three distinct phases:early fluctuation(1901—late 1940s),mid-term stabilization(1950s—1970s),and recent acceleration(post-1970s).The post-1970s period is marked by a pronounced asymmetry,with rapid increases in WSDI accompanied by persistently low CSDI values.The CMIP6 multi-model ensemble successfully captures the overall warming-related trends,with relatively good agreement for the long-term increase in WSDI,but systemati-cally underestimates the magnitude of the CSDI decline,particularly prior to the 1950s.Multi-signal detection and attribution analyses based on optimal fingerprinting indicate that anthropogenic forcing is the dominant driver of the observed changes,with greenhouse gas forcing playing a leading role for both indices.Specifically,GHG forc-ing accounts for 91.1%of the centennial-scale decrease in CSDI and 119.2%of the increase in WSDI,while nat-ural forcing contributes only marginally.During the more recent period of 1951-2020,the observed trends are-4.80 days per century for CSDI and 17.10 days per century for WSDI,reflecting a weakening of the CSDI de-cline and a marked acceleration of WSDI increases.The ALL simulations closely reproduce these trends,further confirming the dominant influence of human activities.In contrast,the impact of anthropogenic aerosol forcing is not detectable for either index over both the full and recent periods. Large-ensemble evaluations suggest that centennial-scale attribution is theoretically most robust;however,un-certainties in early observational data reduce the reliability of attribution results based on the full 1901-2020 re-cord.After jointly considering data quality,signal strength,and model performance,scaling factors derived from the 1951-2020 period are selected to constrain future projections.Under the SSP2-4.5 scenario,constrained pro-jections for 2081-2100 relative to the 1961-1990 baseline indicate changes of-4.25 days for CSDI and+95.27 days for WSDI.Compared with raw projections,the constrained CSDI change is amplified by 18.72%,while the WSDI increase is reduced by 8.59%,with the uncertainty range for WSDI reduced by more than 50%.This study provides robust evidence for the dominant role of anthropogenic forcing in shaping persistent tempera-ture extremes over eastern China and highlights the importance of period selection in attribution-constrained pro-jections,offering a scientifically grounded framework for high-confidence regional climate risk assessment.关键词
持续性冷暖事件/持续性寒流日数(CSDI)/持续性热浪日数(WSDI)/检测归因/约束预估/百年尺度/不对称变化/中国东部/CMIP6Key words
persistent cold and warm events/cold spell duration index(CSDI)/warm spell duration index(WS-DI)/detection and attribution/constrained projections/century scale/asymmetric change/eastern China/CMIP6引用本文复制引用
胡婷,孙颖..近百年中国东部持续性冷暖事件的检测归因与约束预估[J].大气科学学报,2026,49(1):94-108,15.基金项目
国家自然科学基金项目(U2342228) （U2342228）
国家重点研发计划项目(2025YFF0812004) （2025YFF0812004）
中国气象局创新团队项目(CMA2022ZD03) （CMA2022ZD03）
