气候变化领域集成服务门户

Surface temperature variations across the tropics exhibit different levels of spatial coherence, yet this is poorly characterized. Years of high temperature anomalies occurring simultaneously across large geographical regions have the potential to adversely impact food production and societal well-being. Using cluster analysis of correlations between extensive temperature measurements from the last six decades, we find a major change occurs in the late 1970s. Two spatial clusters merge to a single dominant one, and therefore, warmer years are experienced at the same time across most tropical land regions. Noting this change occurs at the same time as the Pacific Decadal Oscillation shifts a warm phase, we investigate this potential driver by a range of coupled ocean-atmosphere-land climate models. These simulations verify that stronger spatial tropical land temperature coherence tends to occur in Pacific Decadal Oscillation warm phases, although model differences exist in projections of how climate change may modulate this dependence.

Plain Language Summary Recent increasing surface temperature variations significantly enhance weather extremes and exacerbate food production and societal well-being. Adverse impacts of high temperature variations could be further aggravated when fluctuations of annual temperature are spatially coherent. Compared to isolated weather disasters, simultaneous co-occurrences of events may cause much broader damages and impair mitigation capacity. Yet it is unclear how temperature covaries across different regions, and if covariation strength is changing. Using cluster analysis of correlations between gridded temperature anomalies over the last six decades, we demonstrate increasing spatial coherence in temperature interannual variability. At the late 1970s, two spatial clusters merged into one dominant cluster covering tropical land regions. Stronger spatial temperature coherence tends to occur in Pacific Decadal Oscillation warm phases, although climate change may modulate this dependence.