全球变化知识资源中心(globalchange): The missing aerosol response in twentieth-century mid-latitude precipitation observations

Regional temperature change over the twentieth century has been strongly influenced by aerosol forcing^{1, 2}. The aerosol effect is also expected to be pronounced on regional precipitation change³. Changes in historical precipitation—for the global mean and land mean of certain regions—should be more sensitive to spatially heterogeneous aerosol forcing than greenhouse gas forcing^{4, 5, 6, 7}. Here, we investigate whether regional precipitation and temperature respond predictably to a significant strengthening in mid-twentieth-century Northern Hemisphere mid-latitude (NHML) aerosol forcing. Using the latest climate model experiments, we find that observed regional temperature changes and observed Northern Hemisphere tropical land precipitation changes are consistent with the IPCC Fifth Assessment Report⁸ aerosol forcing estimate, but observed NHML land precipitation changes show little evidence of an aerosol response. This may be a result of changes in precipitation measurement practice that increased observed precipitation totals at the same time that aerosol forcing was expected to reduce them⁹. Investigating this inconsistency, we calculate the required increase in early-twentieth-century observed NHML land precipitation to bring this result in line with aerosol forcing. Biases greater than this calculated correction have been identified in countries within the NHML region previously, notably the former Soviet Union^{9, 10}. These observations are frequently used as a metric for the quality of model-simulated precipitation. More homogeneity studies would be of huge benefit.

Twentieth-century climate change has been dominated by greenhouse gas (GHG)-driven warming¹¹, interrupted by a mid-twentieth-century period of slight cooling probably driven by aerosols, both globally and in the Northern Hemisphere mid-latitude (NHML) region. The NHML land region has been the source for a large proportion of global emissions originating from human activity¹², including short-lived forcing agents such as aerosols. Our longest, most comprehensive temperature and precipitation observations also exist here. These have allowed for the identification of a temperature response to aerosol forcing in observations and climate models¹. However, no such link between precipitation and local aerosol forcing in the NHML land region has been reported. Here, we consider whether one should be expected and if it is found in observations and models.

In the global mean, energetic constraints dictate that precipitation change is more sensitive to aerosol forcing than GHG forcing per unit temperature change^{4, 5}. The direct effect of GHG forcing counteracts surface temperature-dependent precipitation change, whereas the direct effect of sulphate aerosol forcing is negligible^{13, 14}. Fig. 1a shows the five-year global mean precipitation–temperature relationship for three twentieth-century experiments driven with different forcings using the CanESM2 climate model. Both the experiment forced only by GHGs and the anthropogenic aerosol alone experiment have a linear precipitation–temperature relationship, but the change in precipitation per unit change in temperature is greater in the latter. The all-forcings experiment reflects the temporal evolution of twentieth-century GHG and anthropogenic aerosol forcing. As GHG forcing and temperature increase in proportion in the early twentieth century, precipitation also increases. Aerosol concentrations increase markedly in the mid twentieth century¹², initiating a slight temperature decrease and a larger decrease in precipitation. Precipitation at the end of the twentieth century increases in line with GHG-driven warming. Hence, the twentieth-century precipitation–temperature relationship looks like two GHG-driven straight lines ‘offset’ by mid-twentieth-century aerosol-driven cooling (Methods).