1. Introduction
Increasing awareness and concern about the risks associated with the unfolding of climate change, including changes in droughts and floods associated with shifts in precipitation patterns, have posed considerable demand for credible climate projections at regional scales. Robust information on regional impacts is crucial to be better prepared to manage and mitigate the impacts on our society and the environment, including agriculture, energy, and water resources. Despite their importance, reliable projections of regional climate change still remain a major challenge (IPCC 2013).
Anthropogenic aerosols can drive changes in global and regional climate by impacting both the energy and water cycles; yet, despite progress in the last decade, aerosols have remained the dominant contributor to the uncertainty in total anthropogenic forcing over the industrial era in the last three IPCC reports (IPCC 2013). Uncertainties in quantifying the aerosol impact on climate also hinder our ability to estimate the greenhouse gas (GHG) contribution to the recent global temperature change (Hegerl et al. 1997; Stott and Jones 2012; Ribes and Terray 2013; Gillett et al. 2013; Schurer et al. 2018). Although the IPCC (2013) assessed it as “extremely likely” that more than half of the observed warming since the mid-twentieth century was due to human influences, the uncertainties in the separate effects of greenhouse gases and aerosols are much larger. As emissions of GHGs and aerosols follow different pathways in the future, separation of their respective effects is essential for improving confidence in climate projections for the twenty-first century.
At the global scale, aerosols are estimated to exert a net negative effective radiative forcing of −0.9 W m−2 (from −1.9 to −0.1 W m−2 at the 90% confidence level) on present-day climate, offsetting part of the GHG-induced warming (Myhre et al. 2013). This negative radiative forcing results from perturbation of the climate system through radiative and cloud microphysical effects (aerosol–radiation and aerosol–cloud interactions, respectively; Boucher et al. 2013), the relative magnitudes of which vary across aerosol species. In the direct effect (Charlson et al. 1992), aerosols scatter and absorb mainly solar radiation and thus modify the surface and atmospheric temperature distributions. The atmospheric heating thereby caused by strongly absorbing aerosols like black carbon can cause cloud changes by altering the surface energy budget and static stability, as well as by cloud evaporation (burnoff); these changes are often referred to as the semidirect effect (Hansen et al. 1997). Sulfate aerosols, on the other hand, scatter solar radiation and are additionally efficient cloud condensation nuclei (CCN), which in turn affects cloud microphysical properties and precipitation processes—the cloud albedo (first indirect; Twomey 1977) and the cloud lifetime (second indirect) effects (Albrecht 1989). Nonlinearities in these indirect aerosol effects, caused by saturation effects in the cloud–aerosol interactions, can cause aerosols to be more important in a more pristine atmosphere (e.g., Carslaw et al. 2013).
Being emitted mainly by fossil fuel and biomass burning sources, present-day global atmospheric levels of sulfate, organic carbon, and black carbon are substantially higher than those at preindustrial times. Historical changes, however, have neither been linear nor spatially homogeneous. They result mainly from a combination of changes over two macroregions: emissions from North America and Europe (NAEU) increased steadily from preindustrial times to the peak in the 1970s, when air quality legislation led to their reduction. Asian emissions, in contrast, started to rapidly increase in the 1950s as a result of rapid economic development and have been increasing ever since (Lamarque et al. 2010; Hoesly et al. 2018). These geographical differences in the emission time series, combined with the short lifetime of aerosols in the atmosphere, result in spatially and temporally heterogeneous distributions and higher concentrations around the source regions. The global mean impact of aerosols is therefore not necessarily representative of regional climate responses, which can be substantial (e.g., Ramanathan et al. 2001), even when the magnitude of global forcing remains relatively small.
Numerous studies have suggested that anthropogenic aerosols had an important impact on the twentieth-century climate over various regions (Boucher et al. 2013; Bindoff et al. 2013). Emissions of aerosols from human activities predominantly located in the Northern Hemisphere (NH) have, for instance, been linked to a southward shift of the intertropical convergence zone (ITCZ) since the early 1900s by muting the warming of the Northern Hemisphere relative to the Southern Hemisphere (Chiang and Friedman 2012; Rotstayn and Lohmann 2002; Hwang et al. 2013). Aerosols have been shown to have played a predominant role in the decrease of Northern Hemisphere monsoon precipitation (Polson et al. 2014), as well in weakening its regional components over East Asia (Li et al. 2016; Guo et al. 2013), South Asia (Bollasina et al. 2013; Undorf et al. 2018), and Northwest Australia (Rotstayn et al. 2012). Furthermore, aerosols were found to have contributed to the decrease of precipitation in West Africa (Ackerley et al. 2011; Dong et al. 2014), to cause tropical contraction (Allen and Ajoku 2016), and to have modulated the Atlantic multidecadal circulation and North Atlantic storms (Booth et al. 2012; Dunstone et al. 2013). Aerosol reductions in recent decades might have amplified Arctic temperature increase and sea ice loss (Acosta Navarro et al. 2016). Yet, many of these findings are associated with large uncertainties and are, at times, controversial, resulting in vivid discussions across the scientific community. IPCC (2013) noted that the “lack of agreement across studies prevents generalization of findings to project aerosol-induced changes in regional atmospheric circulation or precipitation in the near term.”
This is also true for studies focused on European climate, where changes in surface temperature and diurnal temperature range (DTR) have been related to an aerosol-induced shift in the surface solar radiation regime from dimming to brightening (Wild et al. 2007; Makowski et al. 2008, 2009). DTR is a measure of high-frequency temperature variability and is not only important for climate impacts as diverse as crop yields (Lobell et al. 2007) and human mortality and morbidity (Kim et al. 2016), but has also been suggested to be a useful indicator to separate aerosol and GHG forcing (Schnur and Hasselmann 2005; Wild et al. 2007): Through reducing surface solar radiation, aerosols are expected to reduce maximum (Tmax) more than minimum (Tmin) temperature, causing a decrease in DTR, while GHGs interact with outgoing longwave radiation and are thus expected to impact Tmax and Tmin equally (although a preferred response of Tmin to GHG warming has also been suggested; Rohde et al. 2013a). Other studies found the recent reduction in aerosols to have increased the frequency of light precipitation events, despite no changes in total precipitation in observations (Stjern et al. 2011) and an increase in the ratio between convective and stratiform precipitation in a GCM (Stjern and Kristjánsson 2015). Complex interactions also underpin the relationship between European aerosol loading and the North Atlantic Oscillation (Chiacchio et al. 2011; Pausata et al. 2014).
Located upstream of the large Eurasian land mass and just north of Africa, European aerosols in particular thus have a potential impact on huge inhabited areas, both through local impacts and through remote impacts mediated by atmospheric circulation adjustments. Their impact throughout the twentieth century, however, has not been consistently investigated so far in a multimodel framework, particularly for the earlier part of the century, when aerosol effects might have been disproportionately large in a cleaner atmosphere (Carslaw et al. 2013). This is a gap this study aims to address. We make use of experiments carried out as part of phase 5 of the Coupled Model Intercomparison Project (CMIP5) initiative (Taylor et al. 2012). These comprise multiple realizations of experiments of all historical forcings as well as individual forcings with a range of coupled global climate models, which allows us to analyze the effect of anthropogenic aerosols and compare it to those of other forcing factors, both when acting in isolation and when combined. The availability of a large set of simulations allows the identification of robust patterns of forced climate response by 1) averaging over multiple members, thus reducing the effect of internal variability and 2) sampling across various model formulations.
To provide a basic characterization of the climate response to NAEU aerosol emissions expected from CMIP5 simulations, we focus on the boreal summer season, when the aerosol impact on temperature and other variables mediated by temperature differences, such as the position of the ITCZ and the jet streams, is expected to be largest due to the insolation maximum in the Northern Hemisphere [see discussion in Hegerl et al. (1997)]. The analysis examines changes in near-surface temperature (TAS), precipitation, DTR, sea level pressure (SLP), and zonal wind at 300 hPa. This subset, although limited, allows us to identify the aerosol impact on relevant surface climate features, some of which are long-term observed, and investigate the potential of aerosols to bring about changes in the atmospheric circulation.
The remainder of the manuscript is organized as follows. After a short description of the datasets and the analysis methods used (section 2), we illustrate the spatiotemporal changes in anthropogenic aerosol emissions and the resulting atmospheric aerosol loading and radiative effect (section 3). The linear trends in the all-forcing and single-forcing CMIP5 ensembles during the historical period of increasing NAEU sulfate emissions are then shown, contrasted by those during the later period of decreasing NAEU emissions (section 4). While this captures multidecadal to near-centennial changes, higher-frequency variability of European near-surface temperature, sea level pressure, and DTR is analyzed by means of area-mean time series also from observations, and the contribution of each single-forcing factor to the total simulated forced interdecadal variability is estimated (section 5). The study is completed by a summary and conclusions (section 6).
2. Data and methods
We use data from the CMIP5 twentieth-century historical experiments and select models that provide simulations forced only with anthropogenic aerosols (Table 1). In addition to this ensemble of aerosol-only simulations (AA), we also analyze simulations with forcing from well-mixed greenhouse gases only (GHG), natural forcings only (solar radiation and volcanoes; NAT), and all these forcings combined (ALLF) for the same set of models. Note that ALLF are the standard historical simulations, which are expected to reproduce the observations. For each model, we choose ensembles of GHG, NAT, and ALLF with the same size as the ensemble available for AA, so that the results from the different ensembles are comparable. Results from the trend analysis are shown for the multimodel mean (MMM), obtained by first averaging the individual ensemble members for each model and successively averaging these ensemble means. Given the different ensemble sizes—ranging from one to five—this ensures that each model is given equal weight.
CMIP5 models used in this study, including the number of ensemble members available for aerosol-only simulations, as well as whether each model includes any representation of the first and second indirect aerosol effects (marked with an x). An overview of the experiment design is given in Taylor et al. (2012).
While the chosen subset has a more comprehensive representation of aerosol effects, compared to the full suite of CMIP5 models (Taylor et al. 2012), there is still substantial variation among the models (Ekman 2014). In GFDL CM3 and NorESM1-M, aerosol microphysics and chemistry including cloud droplet number concentration (CDNC; first indirect effect) are simulated online, and precipitation formation depends on this CDNC (second indirect effect). CSIRO Mk3.6.0, HadGEM2-ES, and CanESM2 also have online aerosol schemes, but use diagnostic CDNC formulations, on which precipitation does (CSIRO-Mk3.6.0, HadGEM2-ES) or does not (CanESM2) depend. GISS-E2-R and IPSL-CM5A-LR prescribe aerosol fields from offline calculations and use diagnostic CDNC representations, which precipitation does (GISS-E2-R) or does not (IPSL-CM5A-LR) depend on. On top of these differences, the chemistry models used and/or the way of the diagnostic derivation of CDNC may vary as well.
Given that all models are forced with the same aerosol emission inventories (Lamarque et al. 2010), the use of multiple models thus does not only reduce the contribution from internal variability, but also allows us the identification of features that are robust to these differences in the models’ representation of indirect aerosol effects, as well as their aerosol distribution and background (preindustrial) aerosol concentrations (Wilcox et al. 2015). This is important since these intermodel differences have been found to cause a large spread in the aerosol responses (Boucher et al. 2013; Kasoar et al. 2016). Robustness is assessed by comparing the sign of the response in the MMM with that in the models’ ensemble means.
The models considered span the full range of CMIP5 spread in climate sensitivity (Flato et al. 2013) and represent the diversity in the effective radiative forcing due to direct and indirect aerosol effects combined (Zelinka et al. 2014). Their historical all-forcing MMM reproduces the climatology and observed changes in surface temperature patterns and simulates the climatology of many large-scale circulation features reasonably well (Flato et al. 2013; Kumar et al. 2016). For DTR, too, the all-forcing simulations from chosen CMIP5 models separately, as well as the MMM of all models’ all-forcing simulations, agree well with observations over Europe both in terms of climatology (Lindvall and Svensson 2015; Cattiaux et al. 2015; Liu et al. 2016) and long-term trends (Lewis and Karoly 2013).
The aerosol forcing in all models is based on decadal aerosol emissions with an annual cycle. We compute boreal summer [June–August (JJA)] means from monthly mean data, and all model data are regridded prior to further analysis to match the model with the lowest resolution, which is 3.75° × 2.8°. For the comparison with observations, we instead interpolate the model data to the observational resolution and mask according to the observational coverage. Anomalies from the climatological state are considered, and the climatologies are computed from the models’ preindustrial control simulations.
Simulated temperature and sea level pressure changes are compared to those derived from available observational data. For near-surface temperature anomalies, we use the HadCRUT dataset version 4.5 (HadCRUT; Morice et al. 2012), the CRU Time Series (TS) dataset version 3.24.01 (CRU; Harris et al. 2014), the gridded Berkeley Earth Surface Temperature dataset (BEST; Rohde et al. 2013b), and GISS Surface Temperature Analysis (GISTEMP; GISTEMP Team 2017; Hansen et al. 2010). For DTR, we use CRU and BEST, and for sea level pressure, we use the Met Office Hadley Centre mean sea level pressure dataset HadSLP2 (1850–2004; HadSLP; Allan and Ansell 2006).
We compute DTR from the CMIP5 model data as the difference between monthly mean Tmax and monthly mean Tmin, as provided by the modeling groups (see Table 1), and follow the same procedure for the BEST dataset; for the CRU data, monthly mean DTR is used as provided by Harris et al. (2014). Note that the observational datasets differ in their station input sources, time sampling, quality control, homogenization, and area averaging; for a comprehensive analysis of the relevance of these differences, see Thorne et al. (2016).
Long-term changes in climate are investigated by computing linear trends, calculated as least squares regressions, and displayed as change per decade. The robustness of the simulated trends is measured by the agreement among the various models. Stippled areas are regions where at least all but one of the models’ ensemble means (i.e., five out of six or six out of seven, depending on the availability of some variables for specific experiments) agree on sign. The time period common to all model simulations is 1860–2005, but we limit the analysis to 1900 onward when observations are less sparse (Morice et al. 2012). Trends in the NAT ensemble are not shown for brevity unless otherwise stated, as they are found to be small compared to those in the AA and GHG ensembles as well as barely robust anywhere across models for the variables, time periods, and regions considered in this analysis.
To support the use of linear trends to identify the link between aerosol loading and near-surface temperature, a joint empirical orthogonal functions (EOF) analysis is also performed (Deser and Blackmon 1993; Wang et al. 2016). This method, which is an ordinary EOF analysis of the two variable fields combined in space, decomposes the data into orthogonal modes, each consisting of one pattern per variable and a common time series, ordered by the fraction of combined variance explained. In contrast to a regression analysis, this identifies covarying patterns without the need to presume time series or spatial patterns. Prior to the EOF analysis, the data are temporally smoothed in order to suppress interannual variability. This is done by taking subsequent 11- and 7-yr running means, which improves filter characteristics relative to taking running means only with a single window length (e.g., von Storch and Zwiers 1999). For a sensitivity analysis, see Text S1 of the online supplementary material.
3. Spatiotemporal changes in aerosol emissions and associated near-surface temperature
Sulfur dioxide (SO2) is the main component of anthropogenic aerosol emissions (Lamarque et al. 2010; Hoesly et al. 2018). The time series of total SO2 emissions from various regions illustrate the dominance of NAEU emissions from preindustrial times until the mid-twentieth century and show the peak in the 1970s, with the trend reversing in Europe around 1975, following a slightly earlier reversal in the United States (Fig. 1). While Asian (including eastern Russian) emissions also started to increase before 1970, they do not exhibit strong (Japan and Russia) or any (e.g., China and India) downward trends afterward. This is also visible in the spatial pattern of the linear trends of SO2 emissions during 1900–70 and 1971–2010 (Fig. 2).
Total SO2 emissions (kg s−1) during boreal summer (JJA) from all sectors (anthropogenic, biomass burning, and shipping) from EU (30°–75°N, 15°W–45°E; orange), NA (15°–75°N, 60°–135°W; red), eastern Russia (RU; 45°–75°N, 45°E–180°; light blue), Asia (0°–45°N, 60°–150°E; medium blue), and the remaining parts of the globe (dark blue). Global emissions are also shown (black). The areas are chosen based on their emission time series and do not conform strictly to political or geographical boundaries. NA, EU, and RU are similar to the respective Tier 1 regions from the Hemispheric Transport of Air Pollution 2 experiments (Koffi et al. 2016), but include for convenience other parts with negligible emissions (Arctic regions in NA, EU, and RU; northern Mexico in NA; and Mongolia and Kazakhstan in RU). The Asian region includes mainly India, most of China, Japan, and southern central Asia. The vertical lines indicate the year dividing the first and second time periods for which trends are calculated: 1970 (light gray) for emissions (which are decadal) and 1975 for simulated variables (dark gray). Data from Lamarque et al. (2010).
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
Linear trends of JJA (top) total SO2 emissions (Mt m−2 yr−1 decade−1) from all sectors, (middle) column-integrated sulfate loading (mg m−2 decade−1), and (bottom) AOD (decade−1) during (a),(c),(e) 1900–70 and (b) 1971–2010 and (d),(f) 1971–2005. Emission data are decadal, so we use 1970 here instead of 1975 as in later figures. For sulfate loading and AOD, the MMMs of the aerosol-only simulations from the models listed in Table 1 are shown, except for HadGEM2-ES in (c) and (d) and CanESM2 in (e) and (f) due to data unavailability. Stippling indicates where at least five out of the six models’ ensemble means agree on sign, and the fraction of stippled grid points within 0°–90°N, 30°W–120°E is given in the top-left corner. Trends in the all-forcing ensemble are very similar (not shown). Data from Lamarque et al. (2010) (emissions) and Taylor et al. (2012) (CMIP5). Note that trends during the second period exceed the color bar range, as indicated by triangular color bar ends.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
The corresponding trend patterns in column-integrated sulfate content (sulfate loading) and aerosol optical depth (AOD) at 550 nm in the CMIP5 model ensemble show that aerosols spread farther east of the source region toward central Eurasia as a result of transport by the climatological westerlies (Fig. 2). Although CMIP5 models use the same aerosol emission inventories, the simulated aerosol distribution patterns may vary from model to model due to differences in meteorology, chemical parameterizations, and preindustrial background concentrations (Carslaw et al. 2013; Wilcox et al. 2015; Kasoar et al. 2016). Nonetheless, there is considerable agreement on the sign of the trends over Eurasia (Figs. 2c,d). The emissions of other aerosol species (black carbon and organic carbon) changed over time less homogeneously in space (Figs. S1–S4), but the simulated changes in AOD, which integrate the amount and optical properties of all simulated aerosol species, are very similar to those in sulfate loading, thus reflecting the dominance of sulfate aerosols (Figs. 2e,f).
A joint EOF analysis reveals modes of covariability of sulfate loading and TAS as simulated in the AA ensemble for summer over the NH, excluding the Pacific (Fig. 3). The dominant temperature signature is a large-scale cooling associated with an overall increase in sulfate aerosols until they peak in the 1970s (Fig. 3a). The first mode’s patterns and time series for both sulfate loading and TAS are very similar to those of the first mode from EOFs of the variables separately (Figs. S5a,b), which confirms that these patterns represent a substantial part of variability in both sulfate loading and TAS individually in the CMIP5 ensemble. Separate EOF analyses of other variables show that approximating the aerosol-related impact with linear trends captures a large fraction of their variability in the CMIP5 AA ensemble (e.g., precipitation and DTR; Figs. S5c,d).
Joint EOF analysis for JJA sulfate loading (SO4load) and TAS from the MMM of the aerosol-only CMIP5 simulations. The (top) principal components of the (a) first and (b) second modes are shown along with the respective patterns for (middle) SO4load and (bottom) TAS. For comparison, the total SO4load over the NH in (a) and difference between SO4load over a western (NAEU) region (the sum of 20°–90°N, 130°W–0°; 30°–90°N, 0°–30°E; and 45°–90°N, 30°–60°E) and an eastern region (0°–45°N, 30°–135°E) in (b) are also shown (blue, dashed lines). The models used are those in Table 1, except for HadGEM2-ES [no data available; separate EOFs show the other results are not sensitive to the inclusion of this model (not shown)]. We apply an 11–7-yr filter prior to the analysis.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
The higher modes from the joint EOF analysis reflect differences in the time series of aerosol emissions and associated temperature impacts at smaller spatial scales. The sulfate loading and temperature patterns of the second mode (Fig. 3b), for instance, have opposite signs in North America and Europe compared to Asia. The principal components accordingly resemble the time series of the difference in sulfate loading over these regions. However, since only a very small fraction of the covariance is explained by the higher modes, we focus here on the dominant behavior as reflected in the first mode.
4. Simultaneous long-term changes in NAEU emissions and near-surface climate
Given the strong opposite trends in SO2 emissions from NAEU before and after the peak in the 1970s, it is reasonable to expect their climate signature to also show a trend reversal around the same time. Emissions from elsewhere (e.g., Asia), in contrast, started to increase in the 1950s and might thus have perturbed the climate during the first period, but we expect their continued increase during the second period to result in changes of the same sign in both periods, albeit of different magnitudes. This also holds true for GHG forcing, which has increased during both periods. We thus identify changes during 1900–74 in the ALLF ensemble as dominantly driven by NAEU aerosol forcing if they are robust and have the same sign as those in the AA but different to those in the GHG or NAT ensembles. This approach is further validated by contrasting the trends with those during 1975–2005, when we expect the decreasing NAEU emissions to drive changes of the opposite sign. Note, however, that the impact of emissions from, say, Asia, is not negligible during 1975–2005, so the effect of decreasing NAEU emissions might be offset by Asian emissions.
Despite clear limitations, this approach allows us to disentangle the NAEU aerosol imprint in the presence of other forcings (provided they changed monotonously during the whole record). The advantage over other methods (e.g., EOF analysis) is that it allows us to assess intermodel consistency and to compare with the other ensembles and observations, for which aerosol data are not available.
Based on the timing of the trend reversal in simulated sulfate loading (indicated by a black line in the time series in Fig. 3), we choose 1900–74 and 1975–2005 as the time periods with increasing and decreasing aerosols, respectively. Note that choosing instead the partition year 1970, as done for the trends in emissions because they are decadal only, does not affect the results reported below (not shown). The first time period includes the slow increase in emissions (Fig. 1), so that trends per decade are smaller during the first than during the second period; European area-mean (35°–60°N, 0°–60°E) trends in the aerosol-only MMM amount to +2.0 and −2.6 mg m−2 decade−1 for sulfate loading and +0.028 and −0.046 1 decade−1 for AOD during 1900–75 and 1975–2005, respectively.
Because emissions from both North America and Europe reversed sign around this time, we will attribute the changes to NAEU emissions in the following. The larger trends in European emissions and their closer proximity to the area studied, however, might suggest them to have a larger share of the impacts, at least on Eurasian near-surface temperature.
a. Near-surface temperature and sea level pressure
The CMIP5 AA ensemble shows a widespread decrease in near-surface temperature over the Atlantic–Eurasian region during the period of increasing NAEU emissions (Fig. 4a). This trend is reversed in the later period, with a warming of similar magnitude as the earlier cooling over central Europe and a slightly weaker warming over the Asian midlatitudes (Fig. 4b). Note that over eastern China, temperatures decrease during both periods, consistent with trends in local aerosol emissions (Figs. 2a,c), so NAEU emissions are not the (sole) drivers of the cooling over this region in the CMIP5 models. The larger trends during the second period—shown in Fig. 4, scaled by a factor of ½—compared to those during the first period are consistent with the larger trends in emissions, sulfate loading, and AOD (Fig. 2).
Linear trends in JJA TAS (color shading; K decade−1) and SLP (contour lines; Pa decade−1) during (a),(c),(e) 1900–74 and (b),(d),(f) 1975–2005 for the MMM of the (top) AA, (middle) ALLF, and (bottom) GHG simulations. Stippling indicates where at least six out of the seven models’ ensemble means agree on sign for TAS, and the fraction of stippled grid points within 0°–90°N, 30°W–120°E is given in the top-left corner. Dashed contour lines indicate negative SLP trends; the zero contour line is drawn in turquoise. The contour interval is 1.25 Pa (10 yr)−1. See Fig. S7 for the robustness of the SLP trends. Note that both temperature and SLP trends during the second period are scaled by ½ for easier comparison with those during the first period.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
GHG forcing produces warming during both periods (Figs. 4e,f), and both the AA and the GHG signatures are robust across the models in the whole region. In the ALLF ensemble, which has the two anthropogenic forcings counteract each other in the first period, the models agree on the midlatitude cooling as well as the warming at higher and lower latitudes. The zonally extended temperature anomalies are consistent with Shindell et al. (2010), who found the influence of inhomogeneous radiative forcing to extend 3–4 times farther in the zonal than in the meridional direction. The vertical cross section of temperature trends averaged over 30°–60°N shows moreover that although the strongest cooling is located near the surface over Europe, the temperature signal extends up to the mid- and upper troposphere and as far eastward as central Asia (Figs. S6a,b), suggestive of an eastward propagation from Europe. Natural forcing does not result in appreciable trends in the CMIP5 models (not shown).
Sea level pressure, consistent across the models, shows anomalous anticyclonic trends over the Eurasian midlatitudes in the first period both in the AA and the ALLF ensembles, which is consistent with the atmospheric adjustment to the surface cooling (Figs. 4a,c and S7a,c). The largest aerosol-induced anomalies in TAS and SLP east of the area of highest emission changes (Figs. 4, S6, and S7) are consistent with expectations from aerosol transport as well as temperature advection by climatological westerlies (not shown). While increasing anthropogenic aerosols thus dominate the simulated Eurasian midlatitude temperature changes during 1900–74, simulated tropical temperature changes appear to be more strongly influenced by GHG forcing, with a widespread warming seen both in the GHG and in the ALLF CMIP5 ensembles (Figs. 4c–f).
b. Jet stream strength and position
The zonal character of the aerosol-driven temperature changes as simulated with the CMIP5 models suggests that aerosols might also affect circulation features in the latitudinal direction via changes in the meridional temperature gradients. We find indeed that both the AA and the ALLF ensembles show a strengthening of the equatorward side of the subtropical jet over Asia, as identified from changes in 300-hPa zonal wind strength during the first time period and a weakening during the second period (Figs. 5a–d). GHG (and NAT; not shown), on the other hand, show no large-scale significant or robust trends (Figs. 4e,f and 5e,f). This is consistent with the thermal wind relation, with AA cooling the midlatitudes more than the tropics and thus enhancing the temperature gradient, which results in a strengthened zonal wind (Figs. 4a,c and 5a,c). Conversely, during the second period, the decrease in NAEU emissions warms the midlatitudes more than the tropics, thus reducing the gradient and weakening the zonal wind (Figs. 4b,d and 5b,d). The significant positive wind speed anomalies on the southern flank of the climatological jet position during the first period, and negative ones during the second, can be additionally interpreted as a southward and northward, respectively, displacement of the circulation in response to the changing meridional temperature gradients.
Linear trends in JJA zonal wind at 300 hPa (U300; contour lines) during (a),(c),(e) 1900–74 and (b),(d),(f) 1975–2005 for the MMM of the (top) AA, (middle) ALLF, and (bottom) GHG simulations. Dashed contour lines indicate negative trends. The contour interval is 0.02 m s−1 decade−1. The 0.00 and ±0.02 contours are not displayed for clarity. Gray shading indicates where the climatological zonal wind speed at 300 hPa exceeds 12.5 m s−1 as a rough estimate of the jet stream position. Coloring (red for positive trends, blue for negative ones) indicates where at least six out of the seven models’ ensemble means agree on sign. Note that trends during the second period are scaled by ½ for easier comparison with those during the first period.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
c. Precipitation
Another feature associated with the change in the meridional circulation is the shift of the ITCZ in response to variations in the interhemispheric temperature gradient, driven by the aerosols’ preferential cooling of the Northern versus the Southern Hemisphere (Chiang and Friedman 2012; Hwang et al. 2013). In the CMIP5 models, we also see an aerosol signal on precipitation over western Africa, where rainfall is largely controlled by the ITCZ (Undorf et al. 2018): a drying during the period of increasing NAEU aerosols and a wettening during the period of decreasing aerosols both in the AA and the ALLF ensembles (Figs. 6a–d).
Linear trends in precipitation (mm day−1 decade−1) for JJA during (a),(c),(e) 1900–74 and (b),(d),(f) 1975–2005 for the MMM of the (top) AA, (middle) ALLF, and (bottom) GHG simulations. Stippling indicates where at least six out of the seven models’ ensemble means agree on sign, and the fraction of stippled grid points within 0°–90°N, 30°W–120°E is given in the top-left corner. Note that trends during the second period are scaled by ½ for easier comparison with those during the first period.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
Over most of Eurasia, where climatological precipitation is relatively low and spatially heterogeneous, precipitation trends in the CMIP5 AA ensemble during 1900–74 are small (Fig. 6a). Key robust features are a wettening trend over western Europe as well as a drying over the high latitudes, including Scandinavia and northern Russia, which is also reversed during 1975–2005 (Fig. 6b). The ALLF ensemble shows also that if aerosols are included, the drying over Europe as simulated for GHG forcing only is restricted to the Mediterranean (Figs. 6c–f).
In the subtropical and tropical monsoon regions of South Asia, East Asia, and western Africa, in contrast, we see a robust aerosol signal during both periods (Figs. 6a–d). During the first period, the AA ensemble shows a drying over northern India, which reverses during the second period (Figs. 6a,b). This robust and significant aerosol signal also dominates—although over a smaller domain—in the ALLF ensemble, despite the counteracting effect of GHGs (Figs. 6c–f).
Both the AA and the ALLF ensembles also show a meridional tripole pattern of longitudinally elongated bands featuring a drying over central China and the western Pacific, a wettening over southern China and Indochina, and again a drying over the Maritime Continent during the first period (Figs. 6a,c). This pattern is especially consistent between AA and ALLF over land, whereas the GHG impact appears to emerge in driving the oceanic precipitation increase over the Indian Ocean and the western Pacific (Fig. 6e). While these precipitation changes have thus been clearly driven by anthropogenic aerosols in the CMIP5 models, our results do not allow us to unequivocally separate the roles of remote and local aerosols, owing to the simplicity of the linear-trend approach. In addition to NAEU emissions, local emissions also increased—again, potentially more effective than later in the century—the effects of which are presumably added onto those from NAEU emissions (Guo et al. 2013). The importance of NAEU aerosols is clear, however, from comparison with a shorter, earlier period (1900–50) for which the assumption of negligible Asian emissions change is even more justified (Fig. 1). The precipitation changes during 1900–75 when both NAEU and Asian emissions increased strongly resemble those during 1900–50, when only NAEU emissions increased, so that the response to NAEU forcing must be a substantial part of the response to the combined forcing (Figs. 6 and S10). Impact from NAEU aerosols in our results is plausible, too, since the anomalous anticyclone over midlatitude Asia, resulting from the large-scale near-surface and tropospheric cooling (Figs. 4, S6, and S7), might cause northeasterly winds that oppose the climatological monsoon flow and thus induce a weakening of the East Asian summer monsoon circulation, as visible in the precipitation changes over East Asia (Fig. 6). During 1975–2005, on the other hand, the impact of increasing local emissions could have counteracted that of decreasing NAEU emissions (Guo et al. 2013), and the continued drying suggests that NAEU emissions were not the most important factor in driving the precipitation changes over China and Southeast Asia in the CMIP5 simulations during this later period (Figs. 6a,b).
d. Diurnal temperature range
DTR shows a widespread decrease over much of the European and Asian midlatitudes both in the AA and the ALLF ensembles during the first time period (Figs. 7a,c). Over Europe, the trend is reversed during the second period (Figs. 7b,d). The trends in the GHG ensemble (and NAT; not shown) are less uniform and not consistent with ALLF (Figs. 7e,f), which suggests that aerosols, not greenhouse gases, dominate the twentieth-century DTR variations over Europe, in agreement with expectations. Interestingly, a significant, consistent aerosol signal is also noticeable at low latitudes (from northern Africa to India), but of opposite sign to that in the midlatitudes, with GHG (and NAT) trends again much smaller, less robust, and inconsistent with those in the ALLF ensemble; coinciding precipitation trends (Fig. 6) hint at modulations in diurnal convection by changes in cloudiness as a possible mechanism underlying this aerosol control on DTR over these regions, which is supported by trends in total cloud fraction (Fig. S8).
Linear trends in DTR (K decade−1) for JJA during (a),(c),(e) 1900–74 and (b),(d),(f) 1975–2005 for the MMM of the (top) AA, (middle) ALLF, and (bottom) GHG simulations for the models listed in Table 1. For AA, no HadGEM2-ES data are available; thus, the MMM of the remaining six models is shown. Stippling indicates where at least six out of the seven (ALLF, GHG) or five out of the six (AA) models’ ensemble means agree on sign, and the fraction of stippled grid points within 0°–90°N, 30°W–120°E is given in the top-left corner. Note that trends during the second period are scaled by ½ for easier comparison with those during the first period.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
e. Sensitivity study and early twentieth century
The robustness of the trends during 1900–74 is further ascertained by considering the shorter 1900–50 period, during which Asian emissions increased even less and were at levels comparable to those of combined European and North American emissions in 1850 (Fig. 1). During this period, trends of similar patterns to those for 1900–74 emerge in the CMIP5 simulations over the region considered, but with smaller magnitudes, as expected from the smaller increase in aerosols (and GHGs) that had taken place by then (Figs. S9–S11). For near-surface temperature, the cooling in the AA ensemble during 1900–50 does not dominate the ALLF response, but reduces the warming trends over Europe seen in the GHG ensemble in the midlatitude band that shows aerosol-related cooling during 1900–74. The pattern of change in precipitation over Southeast Asia described above is already recognizable in the AA trends during 1900–50; also, DTR shows a robust narrowing over the midlatitudes in both the AA and the ALLF ensembles.
These findings suggest that aerosols were already a key driver of regional climate anomalies before the peak aerosol increase between the 1950s and 1970s, for some variables even dominant over GHGs. Note that saturation effects might contribute to this by potentially making the aerosol emissions less effective later in the twentieth century (e.g., Carslaw et al. 2013). The similarity of the patterns and their already considerable magnitude during 1900–50 furthermore supports the notion that remote (NAEU) emissions contributed substantially to the simulated near-surface climate variation over Russia and Asia, since the amount of remote aerosol emissions dominated over that of local emissions even more in 1950 than in 1970. In 1950 (1970), there were 8 (4) times more emissions from Europe than from Asia, and from 1900 to 1950 (1900–70), those from Europe increased nearly 6 (3) times faster than those from Asia (Fig. 1).
5. Temporal evolution of European climate
a. Comparison with observations
In this section, the model-based findings on long-term changes associated with aerosols discussed above are complemented by an analysis of area-mean time series for the region with the largest trends in the AA and ALLF ensembles (35°–60°N, 0°–60°E; Fig. 8). We focus on near-surface temperature, sea level pressure, and DTR, which all show a spatially homogeneous response across this region in the CMIP5 ensemble (Figs. 4 and 7). These time series are also compared with observations, the coverage of which is better for this region than most others, especially early in the twentieth century.
Comparison with observations over Europe (35°–60°N, 0°–60°E). Time series of area mean JJA (a) TAS (K), (b) SLP (Pa), and (c) DTR (K). Observations (black lines) are shown as well as the CMIP5 MMM for single-forcing (AA in blue, GHG in red, NAT in yellow) and all-forcing simulations (ALLF in gray) along with the range of 1.96 standard deviations of the individual model simulations (shading). The CMIP5 data for TAS and DTR are masked to the coverage of the respective CRU dataset. All data are temporally smoothed with an 11–7-yr filter, and anomalies with respect to 1901–30 are plotted. For DTR, no HadGEM2-ES data are available; thus, the MMM of the remaining six models is shown.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
The temporal change in near-surface temperature shows the contrast between the warming in the GHG-only experiments and the cooling in the AA ensemble (Fig. 8a). The small gradients in both ensembles in the first decades of the twentieth century are consistent with weak variations in anthropogenic emissions, while clear differences between the gradients in the two ensembles start appearing around 1925, with similar magnitudes of opposite sign until the 1970s. The largest aerosol-induced variations are found during 1940–75, followed by a plateau and a weak recovery, reflecting the peak and subsequent decline in global aerosol emissions. In contrast, the GHG warming increased exponentially during the twentieth century, with the largest trends after 1960. The all-forcing ensemble reflects the modulation due to both drivers and the contribution from natural variability (e.g., the volcanic cooling after the Agung eruption in 1963; e.g., Robock and Mao 1995). After a few decades of negligible anthropogenic forcing at the beginning of the century, temperature variations are dominated by aerosol cooling until at least 1970, when GHG warming starts to dominate instead. Comparison with the observed changes shows that the CMIP5 models are able to capture the observed twentieth-century variations of near-surface temperature over Europe (observations within the 95% range of ALLF). The most notable feature in the observations is the cooling during 1940–80, which is consistent with the prevailing role of anthropogenic aerosols.
For sea level pressure (Fig. 8b) and DTR (Fig. 8c), the single-forcing ensembles diverge less clearly, reflecting higher variability, lesser model agreement, and more complex responses of these variables, compared to temperature to the different forcings. For SLP, both the AA and the ALLF ensembles encompass the observations, but the GHG and NAT ensembles do not, indicating that aerosol forcing is essential to explain the observed variations.
For DTR, the two observed datasets agree on multidecadal variability, showing a decrease until 1930, followed by a steep increase and a subsequent decrease between the mid-1930s and the mid-1980s. They differ in their absolute anomalies, however, due to discrepancies in the first half of the century (Thorne et al. 2016). While all ensembles are compatible with the CRU observations at the 95% level, the GHG and NAT ensembles are incompatible with the BEST dataset in the second half of the twentieth century. The AA and ALLF ensembles are the only ones to show a DTR narrowing over most of the twentieth century, as do the observations. This overall provides further confirmation of aerosol forcing being necessary to explain historical observed variations. However, because of the strong influence of internal climate variability, conclusive attribution of observed changes was not possible.
b. Contribution of individual forcings to simulated European summer climate
The contribution of each single forcing on the forced component of simulated European interdecadal climate variability can be estimated by calculating the gradient of the multimodel mean time series from each ensemble and comparing the magnitude of those from the single-forcing CMIP5 ensembles (AA, GHG, and NAT) to their arithmetic sum as in Wilcox et al. (2013). This approach assumes that the role of internal variability is negligible in the multimodel mean and that the responses to the individual forcings add linearly (Fig. 9). The time series are smoothed prior to the analysis by consecutively taking 11- and 7-yr running means to suppress variability on time scales shorter than interdecadal (section 2).
Estimates of the contribution of each single forcing on the interdecadal variability of (a) TAS (%), (b) SLP (%), and (c) DTR (%) for the European region as in Fig. 8 (here, the data are not masked to the observational coverage). At the top of each panel, the instantaneous gradient of the MMM all-forcing time series (gray) and the sum of the gradients of the MMM single-forcing time series (black) are shown. At the bottom of each panel, the contributions from GHG (red), AA (blue), and NAT (yellow) are shown, derived by taking the gradient of the ensemble-mean time series for each of these forcing experiments and dividing their magnitudes by the sum of all these gradient magnitudes. All time series are temporally smoothed with an 11–7-yr filter prior to the analysis and further smoothed (7–5-yr filter) for plotting.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0850.1
For European near-surface temperature, the sum of the gradients (AA + GHG + NAT) is a reasonable approximation to the gradient from the ALLF ensemble (Fig. 9a, top), and during the twentieth century, the contribution of AA to the total forced simulated variability (36%) is comparable to that of GHG (43%) and larger than that of NAT (21%). The aerosol impact is even dominant during 1940–70, with AA explaining more than 50% of the total forced variability. The aerosol contribution to the total forced interdecadal model variability in sea level pressure is also discernible and amounts to about a third throughout the twentieth century and slightly more again in the period 1940–70 (Fig. 9b). For DTR, AA contributes, on average, a fourth throughout the twentieth century; this contribution grows toward the later decades, but then the sum of the gradients of the single-forcing time series approximates the all-forcing one less well (Fig. 9c).
6. Summary, discussion, and conclusions
The long-term signature of anthropogenic aerosol emissions mainly from North America and Europe (NAEU) on Eurasian summer climate throughout the twentieth century has been identified in an ensemble of coupled climate (CMIP5) models. The analysis was motivated by the need to advance the current understanding of the effects of aerosols on regional climate, which is of utmost importance to more confidently assess and quantify the drivers of past climate variations, as well as to reduce uncertainties in near-future climate projections. North America and Europe were the key aerosol emission regions worldwide for most of the twentieth century, with the potential to influence climate downstream over large inhabited regions. Yet, the topic has not been consistently investigated so far. While earlier studies have either mostly focused on individual variables or used only one climate model, robust information on the aerosol impact on a range of relevant variables across multiple coupled models is still largely unknown.
The aerosol imprint on simulated summer climate was inferred from intermodel agreement on linear trends during a period of increasing NAEU sulfate emissions (1900–74), contrasted with trends during the more recent period of decreasing emissions (1975–2005). This was supported by an analysis of the covariability of near-surface temperature and sulfate loading. The main findings are as follows:
Regionally and during the 1900–74 period, aerosols generated a large-scale anomalous cooling stretching from Europe across most of the Eurasian midlatitudes in the CMIP5 models. The cooling, largest at the surface, extends to the mid- and upper troposphere and is associated with a large-scale anticyclonic sea level pressure anomaly centered over Russia, as well as with a widespread narrowing of DTR. Aerosols also strengthened the Northern Hemisphere subtropical jet on its equatorward side, which is consistent with changes in the meridional temperature gradient, and decreased monsoon precipitation over western Africa, northern India, and eastern China.
During the 1975–2005 period, most of these changes are reversed, which provides support to their association with NAEU sulfate aerosol emissions—only precipitation changes over East Asia, where strong positive trends in local aerosol emissions are expected to superpose the effect of decreasing NAEU emissions, do not reverse.
Analysis of observations averaged across Europe shows clear evidence for aerosols and GHGs both being important for models to reproduce inter- and multidecadal variations in near-surface temperature. Observed variations in sea level pressure are noisy but tend to agree much better with simulations that include anthropogenic aerosols than with those without aerosols. Observed DTR has very high data uncertainty, but shows a decline over much of the century that does not occur in the CMIP5 GHG and NAT ensembles.
Assuming linearity among the single forcing responses, anthropogenic aerosols are estimated to explain, on average, more than a third of the forced simulated interdecadal variability in European near-surface temperature during the twentieth century and more than half during 1940–70.
The time series analysis thus highlights the importance of aerosols in addition to GHGs for explaining temporal variations in observed European near-surface climate, in agreement with Bindoff et al. (2013). For DTR, our results add on to those by Makowski et al. (2008), providing the evidence from an ensemble of CMIP5 experiments. Furthermore, we show that regionally, AA explains a substantial fraction of simulated European climate interdecadal variability from as early on as 1900, which is even larger than that found by Wilcox et al. (2013) for global mean temperature.
Significant changes in strength and latitudinal position of the subtropical jet stream during the twentieth century have been discussed in a number of studies, including those focusing on variations of the tropical belt width, given that the jet stream can be interpreted as the poleward boundary of the tropics. While some studies only looked at annual means (Archer and Caldeira 2008b; Fu and Lin 2011; Moore 2013) or the winter season (Strong and Davis 2007), changes in winter and summer, if any, were generally found to be consistent in sign (Archer and Caldeira 2008a; Hudson 2012; Davis and Birner 2013; Pena-Ortiz et al. 2013; Abish et al. 2015), despite the pronounced climatological seasonality. Evidence suggests a weakening and/or poleward shift of the NH subtropical jet since 1979 from satellite observations (Fu and Lin 2011), reanalyses (Archer and Caldeira 2008a; Hudson 2012), and observations (Davis and Birner 2013; Pena-Ortiz et al. 2013). Together with weaker and/or even opposite trends since the mid-twentieth century (Strong and Davis 2007; Archer and Caldeira 2008b; Pena-Ortiz et al. 2013; Abish et al. 2015) and ice core proxy data (Moore 2013), this indicates a strengthening before 1979. In the framework of tropical expansion/contraction, these changes have been previously linked to sulfate forcing (Allen et al. 2012, 2014; Allen and Ajoku 2016; Ming et al. 2011), especially during the summer (Tao et al. 2016), but the sign reversal associated with the evolution of aerosol forcing throughout the twentieth century has not been previously discussed in the context of jet stream changes, and the appearance of noticeable trends as early as the beginning of the century has not been shown yet.
The remote impact on monsoon precipitation found here in the CMIP5 models is consistent with Polson et al. (2014). For western Africa, observations similarly show a drying during 1950–85 and a recovery thereafter (Nicholson 2013), and the region has been identified as sensitive to aerosol forcing (e.g., Rotstayn and Lohmann 2002; Ackerley et al. 2011). Dong et al. (2014), who found both Asian and European sulfur dioxide emissions relevant for the decreased West African precipitation, and Undorf et al. (2018), who attributed observed precipitation changes over West Africa during 1920–2005 mainly to NAEU emissions, both used a single model. Our results, based on a range of CMIP5 models, also indicate the predominant importance of NAEU emissions on summer monsoon precipitation over western Africa. The overall weakening of the South Asian monsoon during the second half of the twentieth century forced by anthropogenic aerosols has been acknowledged before (e.g., Bollasina et al. 2011). While other studies found regional and remote aerosols responsible in varying ratios (e.g., Bollasina et al. 2014; Guo et al. 2015), we show that considering the periods with increasing and decreasing remote emissions separately indicates remote emissions to be most important. For East Asia, we found remote emissions not dominant during 1975–2005, consistent with strong impact of local aerosols (e.g., Guo et al. 2013), but our results do not exclude substantial impact of remote aerosols, especially before 1975, and are not inconsistent with a mechanism of remote aerosol impact on monsoon precipitation found in earlier studies (Cowan and Cai 2011; Dong et al. 2016).
The models used in this study explore the range of climate sensitivity and aerosol forcing of those CMIP5 models that include a representation of indirect effects, from those with weak (e.g., IPSL-CM5A-LR) to those with strong (e.g., GFDL-CM3) indirect effects (section 2). Previous studies, however, showed substantial differences between the response to historical aerosol emissions in CMIP5 models with and without a representation of aerosol indirect effects (Wilcox et al. 2013; Guo et al. 2015); other models, which neglect indirect aerosol effects, might therefore misrepresent the response to aerosol emissions as identified in this study.
While the use of historical single- and all-forcing simulations allows us to identify features of aerosol impact within the system of interactions and feedbacks—insofar represented in the models—targeted simulations would allow a more rigorous disentanglement of the effects of aerosols from different source regions, including NA and EU. Similarly, the CMIP5 models’ differing aerosol representations increase the robustness of our findings, but also prevent us from exploring the mechanisms of aerosol impact further. A future study with a different focus would thus complement our work and help better understand the physical mechanisms mediating NAEU aerosol impact.
To conclude, the findings add valuable information about the robustness of the aerosol impacts as well as new insights on their climate footprint, such as their modulation of DTR and the jet stream. The symmetry of the simulated impacts during the period of increasing and of decreasing emissions also sheds light on the changes to be expected from future reductions in aerosol emissions from NAEU and elsewhere.
Acknowledgments
We thank Laura Wilcox for discussion and data. The authors acknowledge the use of observational data from the Met Office Hadley Centre, the Climatic Research Unit at the University of East Anglia, Berkeley Earth, and the NASA Goddard Institute for Space Studies. We also thank the World Climate Research Program’s Working Group on Coupled Modeling, the climate modeling groups (Table 1), the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison, and the Global Organization for Earth System Science Portals. This work was supported by the ERC-funded project (EC-320691) TITAN. G. Hegerl is funded by NCAS and the Wolfson Foundation and the Royal Society as Royal Society Wolfson Research Merit Award holder (WM130060).
REFERENCES
Abish, B., P. V. Joseph, and O. M. Johannessen, 2015: Climate change in the subtropical jetstream during 1950–2009. Adv. Atmos. Sci., 32, 140–148, https://doi.org/10.1007/s00376-014-4156-6.
Ackerley, D., B. B. B. Booth, S. H. E. Knight, E. J. Highwood, D. J. Frame, M. R. Allen, and D. P. Rowell, 2011: Sensitivity of twentieth-century Sahel rainfall to sulfate aerosol and CO2 forcing. J. Climate, 24, 4999–5014, https://doi.org/10.1175/JCLI-D-11-00019.1.
Acosta Navarro, J. C., and Coauthors, 2016: Amplification of Arctic warming by past air pollution reductions in Europe. Nat. Geosci., 9, 277–281, https://doi.org/10.1038/ngeo2673.
Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245, 1227–1230, https://doi.org/10.1126/science.245.4923.1227.
Allan, R., and T. Ansell, 2006: A new globally complete monthly historical gridded mean sea level pressure dataset (HadSLP2): 1850–2004. J. Climate, 19, 5816–5842, https://doi.org/10.1175/JCLI3937.1.
Allen, R. J., and O. Ajoku, 2016: Future aerosol reductions and widening of the northern tropical belt. J. Geophys. Res. Atmos., 121, 6765–6786, https://doi.org/10.1002/2016JD024803.
Allen, R. J., S. C. Sherwood, J. R. Norris, and C. S. Zender, 2012: Recent Northern Hemisphere tropical expansion primarily driven by black carbon and tropospheric ozone. Nature, 485, 350–354, https://doi.org/10.1038/nature11097.
Allen, R. J., J. R. Norris, and M. Kovilakam, 2014: Influence of anthropogenic aerosols and the Pacific decadal oscillation on tropical belt width. Nat. Geosci., 7, 270–274, https://doi.org/10.1038/ngeo2091.
Archer, C. L., and K. Caldeira, 2008a: Historical trends in the jet streams. Geophys. Res. Lett., 35, L08803, https://doi.org/10.1029/2008GL033614.
Archer, C. L., and K. Caldeira, 2008b: Reply to comment by Courtenay Strong and Robert E. Davis on “Historical trends in the jet streams.” Geophys. Res. Lett., 35, L24807, https://doi.org/10.1029/2008GL035114.
Bindoff, N. L., and Coauthors, 2013: Detection and attribution of climate change: From global to regional. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 867–952, https://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter10_FINAL.pdf.
Bollasina, M. A., Y. Ming, and V. Ramaswamy, 2011: Anthropogenic aerosols and the weakening of the South Asian summer monsoon. Science, 334, 502–505, https://doi.org/10.1126/science.1204994.
Bollasina, M. A., Y. Ming, and V. Ramaswamy, 2013: Earlier onset of the Indian monsoon in the late twentieth century: The role of anthropogenic aerosols. Geophys. Res. Lett., 40, 3715–3720, https://doi.org/10.1002/grl.50719.
Bollasina, M. A., Y. Ming, V. Ramaswamy, M. D. Schwarzkopf, and V. Naik, 2014: Contribution of local and remote anthropogenic aerosols to the twentieth century weakening of the South Asian monsoon. Geophys. Res. Lett., 41, 680–687, https://doi.org/10.1002/2013GL058183.
Booth, B. B. B., N. J. Dunstone, P. R. Halloran, T. Andrews, and N. Bellouin, 2012: Aerosols implicated as a prime driver of twentieth-century North Atlantic climate variability. Nature, 484, 228–232, https://doi.org/10.1038/nature10946.
Boucher, O., and Coauthors, 2013: Clouds and aerosols. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 571–657, https://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter07_FINAL.pdf.
Carslaw, K. S., and Coauthors, 2013: Large contribution of natural aerosols to uncertainty in indirect forcing. Nature, 503, 67–71, https://doi.org/10.1038/nature12674.
Cattiaux, J., H. Douville, R. Schoetter, S. Parey, and P. Yiou, 2015: Projected increase in diurnal and interdiurnal variations of European summer temperatures. Geophys. Res. Lett., 42, 899–907, https://doi.org/10.1002/2014GL062531.
Charlson, R. J., S. E. Schwartz, J. M. Hales, R. D. Cess, J. A. Coakley Jr., J. E. Hansen, and D. J. Hofmann, 1992: Climate forcing by anthropogenic aerosols. Science, 255, 423–430, https://doi.org/10.1126/science.255.5043.423.
Chiacchio, M., T. Ewen, M. Wild, M. Chin, and T. Diehl, 2011: Decadal variability of aerosol optical depth in Europe and its relationship to the temporal shift of the North Atlantic Oscillation in the realm of dimming and brightening. J. Geophys. Res., 116, D02108, https://doi.org/10.1029/2010JD014471.
Chiang, J. C. H., and A. R. Friedman, 2012: Extratropical cooling, interhemispheric thermal gradients, and tropical climate change. Annu. Rev. Earth Planet. Sci., 40, 383–412, https://doi.org/10.1146/annurev-earth-042711-105545.
Collins, W. J., and Coauthors, 2011: Development and evaluation of an Earth-system model—HadGEM2. Geosci. Model Dev., 4, 1051–1075, https://doi.org/10.5194/gmd-4-1051-2011.
Cowan, T., and W. Cai, 2011: The impact of Asian and non-Asian anthropogenic aerosols on 20th century Asian summer monsoon. Geophys. Res. Lett., 38, L11703,https://doi.org/10.1029/2011GL047268.
Davis, N. A., and T. Birner, 2013: Seasonal to multidecadal variability of the width of the tropical belt. J. Geophys. Res. Atmos., 118, 7773–7787, https://doi.org/10.1002/jgrd.50610.
Deser, C., and M. L. Blackmon, 1993: Surface climate variations over the North Atlantic Ocean during winter: 1900–1989. J. Climate, 6, 1743–1753, https://doi.org/10.1175/1520-0442(1993)006<1743:SCVOTN>2.0.CO;2.
Dong, B., R. T. Sutton, E. J. Highwood, and L. J. Wilcox, 2014: The impacts of European and Asian anthropogenic sulfur dioxide emissions on Sahel rainfall. J. Climate, 27, 7000–7017, https://doi.org/10.1175/JCLI-D-13-00769.1.
Dong, B., R. T. Sutton, E. J. Highwood, and L. J. Wilcox, 2016: Preferred response of the East Asian summer monsoon to local and non-local anthropogenic sulphur dioxide emissions. Climate Dyn., 46, 1733–1751, https://doi.org/10.1007/s00382-015-2671-5.
Donner, L. J., and Coauthors, 2011: The dynamical core, physical parameterizations, and basic simulation characteristics of the atmospheric component AM3 of the GFDL global coupled model CM3. J. Climate, 24, 3484–3519, https://doi.org/10.1175/2011JCLI3955.1.
Dufresne, J.-L., and Coauthors, 2013: Climate change projections using the IPSL-CM5 Earth system model: From CMIP3 to CMIP5. Climate Dyn., 40, 2123–2165, https://doi.org/10.1007/s00382-012-1636-1.
Dunstone, N. J., D. M. Smith, B. B. B. Booth, L. Hermanson, and R. Eade, 2013: Anthropogenic aerosol forcing of Atlantic tropical storms. Nat. Geosci., 6, 534–539, https://doi.org/10.1038/ngeo1854.
Ekman, A. M. L., 2014: Do sophisticated parameterizations of aerosol–cloud interactions in CMIP5 models improve the representation of recent observed temperature trends? J. Geophys. Res. Atmos., 119, 817–832, https://doi.org/10.1002/2013JD020511.
Flato, G., and Coauthors, 2013: Evaluation of climate models. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 741–866, https://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter09_FINAL.pdf.
Fu, Q., and P. Lin, 2011: Poleward shift of subtropical jets inferred from satellite-observed lower-stratospheric temperatures. J. Climate, 24, 5597–5603, https://doi.org/10.1175/JCLI-D-11-00027.1.
Gillett, N. P., J. C. Fyfe, and D. E. Parker, 2013: Attribution of observed sea level pressure trends to greenhouse gas, aerosol, and ozone changes. Geophys. Res. Lett., 40, 2302–2306, https://doi.org/10.1002/grl.50500.
GISTEMP Team, 2017: GISS Surface Temperature Analysis (GISTEMP). NASA Goddard Institute for Space Studies, accessed 17 January 2017, https://data.giss.nasa.gov/gistemp/.
Guo, L., E. J. Highwood, L. C. Shaffrey, and A. G. Turner, 2013: The effect of regional changes in anthropogenic aerosols on rainfall of the East Asian summer monsoon. Atmos. Chem. Phys., 13, 1521–1534, https://doi.org/10.5194/acp-13-1521-2013.
Guo, L., A. G. Turner, and E. J. Highwood, 2015: Impacts of 20th century aerosol emissions on the South Asian monsoon in the CMIP5 models. Atmos. Chem. Phys., 15, 6367–6378, https://doi.org/10.5194/acp-15-6367-2015.
Hansen, J., M. Sato, and R. Ruedy, 1997: Radiative forcing and climate response. J. Geophys. Res., 102, 6831–6864, https://doi.org/10.1029/96JD03436.
Hansen, J., R. Ruedy, M. Sato, and K. Lo, 2010: Global surface temperature change. Rev. Geophys., 48, RG4004, https://doi.org/10.1029/2010RG000345.
Harris, I., P. D. Jones, T. J. Osborn, and D. H. Lister, 2014: Updated high-resolution grids of monthly climatic observations—The CRU TS3.10 dataset. Int. J. Climatol., 34, 623–642, https://doi.org/10.1002/joc.3711.
Hegerl, G. C., K. Hasselmann, U. Cubasch, J. F. B. Mitchell, E. Roeckner, R. Voss, and J. Waszkewitz, 1997: Multi-fingerprint direction and attribution analysis of greenhouse gas, greenhouse gas-plus-aerosol and solar forced climate change. Climate Dyn., 13, 613–634, https://doi.org/10.1007/s003820050186.
Hoesly, R. M., and Coauthors, 2018: Historical (1750–2014) anthropogenic emissions of reactive gases and aerosols from the Community Emission Data System (CEDS). Geosci. Model Dev., 11, 369–408, https://doi.org/10.5194/gmd-11-369-2018.
Hudson, R. D., 2012: Measurements of the movement of the jet streams at mid-latitudes, in the Northern and Southern Hemispheres, 1979 to 2010. Atmos. Chem. Phys., 12, 7797–7808, https://doi.org/10.5194/acp-12-7797-2012.
Hwang, Y.-T., D. M. W. Frierson, and S. M. Kang, 2013: Anthropogenic sulfate aerosol and the southward shift of tropical precipitation in the late 20th century. Geophys. Res. Lett., 40, 2845–2850, https://doi.org/10.1002/grl.50502.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.
Iversen, T., and Coauthors, 2013: The Norwegian Earth System Model, NorESM1-M—Part 2: Climate response and scenario projections. Geosci. Model Dev., 6, 389–415, https://doi.org/10.5194/gmd-6-389-2013.
Jeffrey, S., L. Rotstayn, M. Collier, S. Dravitzki, C. Hamalainen, C. Moeseneder, K. Wong, and J. Syktus, 2013: Australia’s CMIP5 submission using the CSIRO-Mk3.6 model. Aust. Meteor. Oceanogr. J., 63, 1–13, https://doi.org/10.22499/2.6301.001.
Kasoar, M., A. Voulgarakis, J.-F. Lamarque, D. T. Shindell, N. Bellouin, W. J. Collins, G. Faluvegi, and K. Tsigaridis, 2016: Regional and global temperature response to anthropogenic SO2 emissions from China in three climate models. Atmos. Chem. Phys., 16, 9785–9804, https://doi.org/10.5194/acp-16-9785-2016.
Kim, J., and Coauthors, 2016: Comprehensive approach to understand the association between diurnal temperature range and mortality in East Asia. Sci. Total Environ., 539, 313–321, https://doi.org/10.1016/j.scitotenv.2015.08.134.
Koffi, B., F. Dentener, G. Janssens-Maenhout, D. Guizzardi, M. Crippa, T. Diehl, S. Galmarini, and E. Solazzo, 2016: Hemispheric Transport Air Pollution (HTAP): Specification of the HTAP2 experiments. Joint Research Centre Tech. Rep., 60 pp., https://doi.org/10.2788/725244.
Kumar, S., J. L. Kinter, Z. Pan, and J. Sheffield, 2016: Twentieth century temperature trends in CMIP3, CMIP5, and CESM-LE climate simulations: Spatial-temporal uncertainties, differences, and their potential sources. J. Geophys. Res. Atmos., 121, 9561–9575, https://doi.org/10.1002/2015JD024382.
Lamarque, J.-F., and coauthors, 2010: Historical (1850–2000) gridded anthropogenic and biomass burning emissions of reactive gases and aerosols: Methodology and application. Atmos. Chem. Phys., 10, 7017–7039, https://doi.org/10.5194/acp-10-7017-2010.
Lewis, S. C., and D. J. Karoly, 2013: Evaluation of historical diurnal temperature range trends in CMIP5 models. J. Climate, 26, 9077–9089, https://doi.org/10.1175/JCLI-D-13-00032.1.
Li, Z., and Coauthors, 2016: Aerosol and monsoon climate interactions over Asia. Rev. Geophys., 54, 866–929, https://doi.org/10.1002/2015RG000500.
Lindvall, J., and G. Svensson, 2015: The diurnal temperature range in the CMIP5 models. Climate Dyn., 44, 405–421, https://doi.org/10.1007/s00382-014-2144-2.
Liu, L., Z. Li, X. Yang, H. Gong, C. Li, and A. Xiong, 2016: The long-term trend in the diurnal temperature range over Asia and its natural and anthropogenic causes. J. Geophys. Res. Atmos., 121, 3519–3533, https://doi.org/10.1002/2015JD024549.
Lobell, D. B., C. Bonfils, and P. B. Duffy, 2007: Climate change uncertainty for daily minimum and maximum temperatures: A model inter-comparison. Geophys. Res. Lett., 34, L05715, https://doi.org/10.1029/2006GL028726.
Makowski, K., M. Wild, and A. Ohmura, 2008: Diurnal temperature range over Europe between 1950 and 2005. Atmos. Chem. Phys., 8, 6483–6498, http://www.atmos-chem-phys.net/8/6483/2008/.
Makowski, K., E. B. Jaeger, M. Chiacchio, M. Wild, T. Ewen, and A. Ohmura, 2009: On the relationship between diurnal temperature range and surface solar radiation in Europe. J. Geophys. Res., 114, D00D07, https://doi.org/10.1029/2008JD011104.
Ming, Y., V. Ramaswamy, and G. Chen, 2011: A model investigation of aerosol-induced changes in boreal winter extratropical circulation. J. Climate, 24, 6077–6091, https://doi.org/10.1175/2011JCLI4111.1.
Moore, G. W. K., 2013: Tibetan ice core evidence for an intensification of the East Asian jet stream since the 1870s. Atmos. Sci. Lett., 14, 235–242, https://doi.org/10.1002/asl2.445.
Morice, C. P., J. J. Kennedy, N. A. Rayner, and P. D. Jones, 2012: Quantifying uncertainties in global and regional temperature change using an ensemble of observational estimates: The HadCRUT4 data set. J. Geophys. Res., 117, D08101, https://doi.org/10.1029/2011JD017187.
Myhre, G., and Coauthors, 2013: Anthropogenic and natural radiative forcing. Climate Change 2013: The Physical Science Basis, T. Stocker et al., Eds., Cambridge University Press, 659–740, https://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter08_FINAL.pdf.
Nicholson, S. E., 2013: The West African Sahel: A review of recent studies on the rainfall regime and its interannual variability. ISRN Meteor., 2013, 453521, https://doi.org/10.1155/2013/453521.
Pausata, F. S. R., M. Gaetani, G. Messori, S. Kloster, and F. J. Dentener, 2015: The role of aerosol in altering North Atlantic atmospheric circulation in winter and its impact on air quality. Atmos. Chem. Phys., 15, 1725–1743, https://doi.org/10.5194/acp-15-1725-2015.
Pena-Ortiz, C., D. Gallego, P. Ribera, P. Ordonez, and M. Del Carmen Alvarez-Castro, 2013: Observed trends in the global jet stream characteristics during the second half of the 20th century. J. Geophys. Res. Atmos., 118, 2702–2713, https://doi.org/10.1002/jgrd.50305.
Polson, D., M. Bollasina, G. C. Hegerl, and L. J. Wilcox, 2014: Decreased monsoon precipitation in the Northern Hemisphere due to anthropogenic aerosols. Geophys. Res. Lett., 41, 6023–6029, https://doi.org/10.1002/2014GL060811.
Ramanathan, V., P. J. Crutzen, J. T. Kiehl, and D. Rosenfeld, 2001: Aerosols, climate, and the hydrological cycle. Science, 294, 2119–2124, https://doi.org/10.1126/science.1064034.
Ribes, A., and L. Terray, 2013: Application of regularised optimal fingerprinting to attribution. Part II: Application to global near-surface temperature. Climate Dyn., 41, 2837–2853, https://doi.org/10.1007/s00382-013-1736-6.
Robock, A., and J. Mao, 1995: The volcanic signal in surface temperature observations. J. Climate, 8, 1086–1103, https://doi.org/10.1175/1520-0442(1995)008<1086:TVSIST>2.0.CO;2.
Rohde, R., and Coauthors, 2013a: A new estimate of the average Earth surface land temperature spanning 1753 to 2011. Geoinfor. Geostat., 1 (1), 1–7, https://doi.org/10.4172/2327-4581.1000101.
Rohde, R., and Coauthors, 2013b: Berkeley Earth temperature averaging process. Geoinfor. Geostat., 1 (2), 1–13, https://doi.org/10.4172/gigs.1000103.
Rotstayn, L. D., and U. Lohmann, 2002: Tropical rainfall trends and the indirect aerosol effect. J. Climate, 15, 2103–2116, https://doi.org/10.1175/1520-0442(2002)015<2103:TRTATI>2.0.CO;2.
Rotstayn, L. D., S. J. Jeffrey, M. A. Collier, S. M. Dravitzki, C. Hirst, J. I. Syktus, and K. K. Wong, 2012: Aerosol- and greenhouse gas-induced changes in summer rainfall and circulation in the Australasian region: A study using single-forcing climate simulations. Atmos. Chem. Phys., 12, 6377–6404, https://doi.org/10.5194/acp-12-6377-2012.
Schmidt, G. A., and Coauthors, 2014: Configuration and assessment of the GISS ModelE2 contributions to the CMIP5 archive. J. Adv. Model. Earth Syst., 6, 141–184, https://doi.org/10.1002/2013MS000265.
Schnur, R., and K. L. Hasselmann, 2005: Optimal filtering for Bayesian detection and attribution of climate change. Climate Dyn., 24, 45–55, https://doi.org/10.1007/s00382-004-0456-3.
Schurer, A. P., G. C. Hegerl, D. Polson, C. Morice, and S. F. B. Tett, 2018: Estimating the transient climate response from observed warming. J. Climate, https://doi.org/10.1175/JCLI-D-17-0717.1, in press.
Shindell, D., M. Schulz, Y. Ming, T. Takemura, G. Faluvegi, and V. Ramaswamy, 2010: Spatial scales of climate response to inhomogeneous radiative forcing. J. Geophys. Res., 115, D19110, https://doi.org/10.1029/2010JD014108.
Stjern, C. W., and J. E. Kristjánsson, 2015: Contrasting influences of recent aerosol changes on clouds and precipitation in Europe and East Asia. J. Climate, 28, 8770–8790, https://doi.org/10.1175/JCLI-D-14-00837.1.
Stjern, C. W., A. Stohl, and J. E. Kristjánsson, 2011: Have aerosols affected trends in visibility and precipitation in Europe? J. Geophys. Res., 116, D02212, https://doi.org/10.1029/2010JD014603.
Stott, P. A., and G. S. Jones, 2012: Observed 21st century temperatures further constrain likely rates of future warming. Atmos. Sci. Lett., 13, 151–156, https://doi.org/10.1002/asl.383.
Strong, C., and R. E. Davis, 2007: Winter jet stream trends over the Northern Hemisphere. Quart. J. Roy. Meteor. Soc., 133, 2109–2115, https://doi.org/10.1002/qj.171.
Tao, L., Y. Hu, and J. Liu, 2016: Anthropogenic forcing on the Hadley circulation in CMIP5 simulations. Climate Dyn., 46, 3337–3350, https://doi.org/10.1007/s00382-015-2772-1.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, https://doi.org/10.1175/BAMS-D-11-00094.1.
Thorne, P. W., and Coauthors, 2016: Reassessing changes in diurnal temperature range: Intercomparison and evaluation of existing global data set estimates. J. Geophys. Res. Atmos., 121, 5138–5158, https://doi.org/10.1002/2015JD024584.
Twomey, S., 1977: The influence of pollution on the shortwave albedo of clouds. J. Atmos. Sci., 34, 1149–1152, https://doi.org/10.1175/1520-0469(1977)034<1149:TIOPOT>2.0.CO;2.
Undorf, S., D. Polson, M. Bollasina, Y. Ming, A. Schurer, and G. C. Hegerl, 2018: Detectable impact of local and remote anthropogenic aerosols on the 20th century changes of West African and South Asian monsoon precipitation. J. Geophys. Res. Atmos., 123, 4871–4889, https://doi.org/10.1029/2017JD027711.
von Salzen, K., and Coauthors, 2013: The Canadian Fourth Generation Atmospheric Global Climate Model (CanAM4). Part I: Representation of physical processes. Atmos.–Ocean, 51, 104–125, https://doi.org/10.1080/07055900.2012.755610.
von Storch, H., and F. W. Zwiers, 1999: Statistical Analysis in Climate Research. Cambridge University Press, 484 pp.
Wang, H., S. P. Xie, and Q. Liu, 2016: Comparison of climate response to anthropogenic aerosol versus greenhouse gas forcing: Distinct patterns. J. Climate, 29, 5175–5188, https://doi.org/10.1175/JCLI-D-16-0106.1.
Wilcox, L. J., E. J. Highwood, and N. J. Dunstone, 2013: The influence of anthropogenic aerosol on multi-decadal variations of historical global climate. Environ. Res. Lett., 8, 024033, https://doi.org/10.1088/1748-9326/8/2/024033.
Wilcox, L. J., E. J. Highwood, B. B. B. Booth, and K. S. Carslaw, 2015: Quantifying sources of inter-model diversity in the cloud albedo effect. Geophys. Res. Lett., 42, 1568–1575, https://doi.org/10.1002/2015GL063301.
Wild, M., A. Ohmura, and K. Makowski, 2007: Impact of global dimming and brightening on global warming. Geophys. Res. Lett., 34, L04702, https://doi.org/10.1029/2006GL028031.
Zelinka, M. D., T. Andrews, P. M. Forster, and K. E. Taylor, 2014: Quantifying components of aerosol-cloud-radiation interactions in climate models. J. Geophys. Res. Atmos., 119, 7599–7615, https://doi.org/10.1002/2014JD021710.
