| 英文摘要: | Poor air quality causes an estimated 2.6–4.4 million premature deaths per year1, 2, 3. Hazardous conditions form when meteorological components allow the accumulation of pollutants in the near-surface atmosphere4, 5, 6, 7, 8. Global-warming-driven changes to atmospheric circulation and the hydrological cycle9, 10, 11, 12, 13 are expected to alter the meteorological components that control pollutant build-up and dispersal5, 6, 7, 8, 14, but the magnitude, direction, geographic footprint and public health impact of this alteration remain unclear7, 8. We used an air stagnation index and an ensemble of bias-corrected climate model simulations to quantify the response of stagnation occurrence and persistence to global warming. Our analysis projects increases in stagnation occurrence that cover 55% of the current global population, with areas of increase affecting ten times more people than areas of decrease. By the late twenty-first century, robust increases of up to 40 days per year are projected throughout the majority of the tropics and subtropics, as well as within isolated mid-latitude regions. Potential impacts over India, Mexico and the western US are particularly acute owing to the intersection of large populations and increases in the persistence of stagnation events, including those of extreme duration. These results indicate that anthropogenic climate change is likely to alter the level of pollutant management required to meet future air quality targets. Strategies to improve air quality typically focus on the reduction of emitted pollutants such as particulate matter (PM) and the precursors of tropospheric ozone (O3). However, changing climate dynamics are also likely to play a role in determining future air quality, although the magnitude and direction of this role is uncertain7, 8. A recent assessment8 of meteorological influences found more frequent air stagnation to be the only meteorological condition that consistently resulted in higher near-surface concentrations of both PM and O3. Given strong negative correlation between cyclone frequency and observed stagnation and pollution events5, 6, investigations of the response of air stagnation to enhanced radiative forcing have primarily focused on changes in cyclone frequency over regional (particularly US) domains (for example, refs 5, 6, 15, 16). However, understanding of the response of future air stagnation events to elevated global warming has been found deficient as a result of inaccuracies in the simulation of meteorological variables relevant to air quality (‘model bias’)14, 17, uncertainties in the spatial pattern of projected changes in those variables due to internal climate variability and/or model formulation8, 13, 14, 15, 18, and lack of investigation of changes in stagnation event duration14, 18. We examine air stagnation directly by applying a modified version of the Air Stagnation Index (ASI) to the CMIP5 global climate model ensemble. The ASI follows the ingredients-based approach of weather forecasting, wherein fundamental components of a meteorological phenomenon are identified and analysed using numerical models and/or observational datasets19. The ASI uses thresholds of daily precipitation and upper- and lower-atmospheric winds to determine when the atmosphere is likely to lack contaminant scavenging, horizontal dispersion and vertical escape capabilities4 (Methods). The daily co-occurrences of these meteorological conditions show a robust correlation with observed PM and O3 pollution days5, 6, underpin operational air quality forecasts and, when persistent, are associated with extreme air pollution episodes8, 18. We use historical and high-emission scenario (RCP8.5) CMIP5 simulations to create a multi-model ensemble projection of air stagnation occurrence (Supplementary Discussion). Our analysis examines changes in stagnation event duration, corrects model biases with six unique observational and reanalysis dataset combinations, and applies objective statistical analyses in conjunction with multi-model agreement criteria to quantify robustness of air stagnation change. Grid cell changes are considered robust if 66% of bias-corrected members pass a non-parametric permutation test at the 95% confidence level20 and 66% agree on the change direction (Methods). Stagnant conditions are most frequent in the current climate over the tropics and sub-tropics, with areas of the western US, north Africa, central Asia and Siberian Russia exhibiting relatively high occurrence in the mid-latitudes (Fig. 1a). Regions that experience frequent stagnation but infrequent hazardous air quality, such as Siberian Russia21, confirm that the ASI measures potential impact: in the absence of human inhabitants or natural and/or anthropogenic pollutants, even ideal pollutant-accumulating meteorological conditions do not pose an air quality risk.
|
