Global turbulence response to increased greenhouse gases and aerosols
Concentrations of both BC and SO4 are by far highest in the regions of India and East China (Fig. 1a). This means that in our model experiments, perturbations to aerosol emissions will provide particularly strong impacts in these areas. The experiment involving a doubling of CO2 has a globally uniform concentration perturbation. Figure 1b shows how the three different perturbations (a tenfold increase in BC, fivefold increase in SO4 and a doubling of CO2) influence near-surface-temperature, boundary layer stability, boundary layer turbulence and boundary layer height. As both the spatial and temporal resolution of CESM2-CAM6 is too coarse to resolve PBL turbulence, the Richardson number (Ri) is used as a proxy for this parameter (see Methods). Higher values of Ri indicate more non-turbulent conditions, while higher values of the Brunt-Väisälä frequency indicate more stable conditions.
For the BC perturbation, we see a particularly strong increase in stability and reduction in turbulence and PBL height in the regions of high aerosol emissions. For the SO4 perturbation there is also a general (but weaker) reduction in turbulence over land regions, while the CO2 doubling causes an opposite reduction in stability and increase in turbulence and PBL height over land.
While our focus will be on changes over land, we still note that the BC perturbations causes similar changes to turbulence and PBL height over land and ocean, while changes to SO4 emissions and CO2 concentrations cause opposite changes over land and ocean. The reduction in PBL height over oceans for the CO2 perturbation is established early and is visible even in an additional set of simulations where we held sea-surface temperature fixed (Supplementary Fig. 1). With fixed sea-surface temperatures, the lower-tropospheric stability (as quantified by the Brunt-Väisälä frequency) is reduced over land, where surface temperatures are allowed to increase, but increases over ocean. This rapid shallowing of the marine PBL following a CO2 perturbation follows from suppressed surface fluxes and reduced buoyancy (Kamae and Watanabe, 2013; Watanabe et al., 2012), but here we show that the land/ocean pattern is opposite and is preserved even after both land and ocean temperatures are allowed to evolve for 70 years. For all three climate drivers investigated here, the long-term response in PBL height and turbulence over land is similar both in sign and magnitude to averages over the first 20 years in which sea surface temperatures are held fixed. This indicates that even for CO2, the observed response over land occurs on a relatively fast time scale.
Statistically significant correlations are found between changes in turbulence and changes in PBL height in Fig. 2a, where average monthly mean changes for each grid cell over land are shown in scatter plots for each experiment. As there have been suggestions that BC poses a greater health risk per unit mass than other aerosol species (Janssen et al., 2011), we use BC concentration as a proxy for near-surface air pollution in this analysis. In Fig. 2 darker colors indicate grid cells with larger difference in near-surface air pollution (BC concentration) between the given perturbation run and the baseline run (see Methods). For the BC experiment in particular, regions with large near-surface aerosol changes are also associated with strong reductions in turbulence and PBL height. Note here that for the simulation involving a tenfold increase in BC, we have divided near-surface BC concentrations by 10, so that any increase in BC reflects a stronger increase than the emission perturbation itself should involve. The grid cells with the 95th percentile highest near-surface aerosol changes had a nine times higher reduction in turbulence than the average while the PBLH reduction was 4 times as strong. For the SO4 and CO2 perturbations the link between areas of turbulence changes and near-surface aerosol changes is not as strong
While the perturbations in BC and SO4 in general reduces the boundary layer lapse rate, turbulence and boundary layer height, CO2 tends to increase them over land. The next question is whether these changes are also associated with a change in the number of hours with elevated near-surface aerosol numbers. Some have suggested that this positive feedback mechanism only emerges under conditions that are highly polluted to begin with (Ding et al., 2016; Liu et al., 2019; Petäjä et al., 2016). In the next section we therefore focus on the Eastern China region marked in Fig. 1a, and on the months from November through February as the most severe haze events typically occur in the winter season.
Regional response in the high-emission region Eastern China
Evidence of the boundary layer feedback in China has typically been linked to absorbing aerosols (e.g., Li et al., 2021; Li et al., 2017; Lou et al., 2019; Petäjä et al., 2016; Wang et al., 2020). For instance, based on aerosol retrievals Dong et al. (2017) found strong evidence of a boundary layer feedback in northern parts of China, where aerosols were typically mostly absorbing, while no such evidence was found in southern parts of China, where aerosols were generally less absorbing. Averaged over the East China region indicated in Fig. 1a, we find that a tenfold increase in BC causes a statistically significant 7.8 % reduction in annual mean PBL height. However, as we have seen above, scattering aerosols and greenhouse gases also influence the PBL height: the fivefold increase in SO4 causes a 1.4 % reduction, while the doubling of CO2 causes an increase of 2.7 %.
In Fig. 2b we look for evidence of changes in turbulent conditions over the East China region. As mentioned, we focus on the winter (including here the months of November through February). PDFs of hourly values of the Richardson’s number show that the simulation with added BC is clearly skewed towards the right, toward more non-turbulent conditions. For the simulation with added SO4, there is no significant change in turbulence, while the doubled CO2 simulation is skewed towards the left and thus towards more turbulence. Do these changes to turbulence and PBL height lead to changes in near-surface aerosol concentrations? Again, using BC concentration as a measure of health hazardous near-surface aerosol concentrations, we present in Fig. 2c hourly wintertime values of different levels of near-surface BC concentrations for the baseline simulation as well as the three perturbed simulations. For the most severe cases (high near-surface BC levels) the doubling of CO2 causes a clear reduction, while perturbations in both BC and SO4 cause an increase in near-surface pollution.
Thus, emissions in greenhouse gases as well as emissions of both absorbing and scattering aerosols seem to influence turbulence, PBL height and near-surface aerosol concentration. To understand more of the processes and to reveal whether it is the boundary layer feedback that causes these links, we next take a look at the regional mean diurnal cycle over East China.
Crucial to the development of turbulence and the growth of the boundary layer is the vertical temperature profile. In Fig. 3a we show the wintertime change in the diurnal temperature profile in the lower troposphere, for the three different perturbations. The effect of BC on the temperature profile over China is striking – a strong warming around 800hPa is triggered by the absorbing aerosols which cause a strong enhancement in solar absorption at these levels. The added absorption also reduces the amount of solar radiation reaching the surface, causing a decrease in near-surface temperatures. The upper-level warming is by far strongest in late afternoon local time, around 15.00 hours. This is consistent with Ding et al. (2016), who found for megacities in China that upper-PBL heating by BC was strongest during late afternoon. These aerosol-radiation interactions cause an increase in atmospheric stability by reducing the temperature difference between the surface and higher atmospheric levels. Through reduced surface fluxes, this leads to an increase in the Richardson number, strongest around 15.00 hours (Fig. 3b) but statistically significant by Student’s t-test for all hours of the day. The diurnal variation in the BC-induced temperature increases and turbulence reduction connects closely to a change in PBL height (black line in Fig. 2c), where hours with significant change are marked with star symbols. In the BC perturbation, the PBL height change is significant for all hours of the day but strongest at 15.00 hours. Finally, the increase in near-surface BC concentrations (yellow line) also shows a similar but opposite diurnal cycle, indicating a link between the processes driving the changes to turbulence and PBL height and the exacerbated near-surface aerosol conditions.
In the case of the simulation with perturbed SO4, we also see an increase in atmospheric stability, this time driven by a reduction in surface temperatures as the scattering aerosols efficiently reflect a portion of the incoming solar radiation back to space. However, neither the change in turbulence (Fig. 3b) nor the change in PBL height (Fig. 3c) is statistically significant for any of the 24 hours. While there is a significant increase in near-surface BC, shown also in Fig. 2c, in this case this is clearly not driven by the boundary layer feedback. Instead, by looking at changes in the model’s different aerosol modes (the primary carbon mode and the BC accumulation mode), we find that the added SO4 causes a very strong reduction in primary mode and increase in accumulation mode BC. In the CAM6 MAM4 aerosol scheme, BC particles “grow” through coating by SO4 or SOA, and this constitutes the strong removal of primary mode and addition of accumulation mode BC aerosols. This model response, although approximating a similar effect in nature, makes it difficult to draw conclusions as to the boundary layer feedback effect of SO4 from these simulations. In a future study we will use a regional model to isolate the link more properly between the stabilizing effect of SO4 and surface pollution, comparing the effects of SO4 and BC.
The doubling of CO2 has an influence on the atmospheric temperature profile that is opposite to the effects of BC and SO4; the increase in surface temperature (Fig. 3a) makes the atmosphere less stable, and as a result we see a strong mid-day reduction in the Richardson number in Fig. 3b. The increased turbulence causes an increase in PBL height that is statistically significant at the same time of day that we see a reduction in near-surface BC particles (Fig. 3c). Although these changes are not as strong as for a tenfold increase in BC, we see clear indications of a boundary layer feedback also for this change in CO2.
Application to historical and future trends
Historical changes in aerosol and greenhouse gas emissions have been substantial, but finding the link between emission changes and observed historical trends in PBL height (Huo et al., 2021; Guo et al., 2019; Dong et al., 2017; Díaz et al., 2019) is challenging – not only because so many other factor influence PBL height, but also because the observations are notoriously difficult to compare due to differences in data sources and PBL height approximation methods. There are several indications, however, that changes in observed PBL height is connected to emission changes. For instance, Guo et al. (2019) investigated a larger observational data set for China in general, and found an increase in PBL height from 1979–2004, but a robust reversal of the trend from 2004 to 2016. They found that an increase in urbanization and BC emissions since the turn of the century are strong contributors to this reversal in the PBL height trend. Huo et al. (2021) analyzed observed trends in PBL height and also found an increase in major cities in China, reporting trends in aerosol emissions as an important co-driver of PBLH evolution. Recently, Zhou et al. (2021) analyzed global PBLH-trends between 1979 and 2019 using ERA5 and MERRA2 reanalysis data, and found relatively strong discrepancies regionally between the two. A comparison to historical simulations in the latest Sixth Coupled Model Intercomparison Project (CMIP6) ensemble found stronger agreement to MERRA2 than to ERA5, but all three data sets displayed declining PBL height over India.
In Fig. 4a we have calculated the CMIP6 ensemble mean (see Table S2 for included models) linear trend in PBL height over the entire historical period (1850 to 2014), for which both greenhouse gases and aerosol emissions have gone up. We find declining PBL height coincident with regions of strong aerosol emission increases, suggesting that the aerosol influence on PBL height dominates the response. Based on population data from the Socioeconomic Data and Applications Center (SEDAC) (Jones and O’Neill, 2020) we find that 83% of the world’s population (in year 2010) live in areas that have seen a reduction in the PBL height over the historical period. The response in the CMIP6 ensemble is comparable to that of CESM2-CAM6, as seen in Fig. S2.
Aerosol emission changes are estimated to continue into the future, with particularly large and uncertain trends in Asia (Samset et al., 2019). The Shared Socioeconomic Pathways (SSP) used by the CMIP6 community (e.g., Rao et al., 2017; Lund et al., 2019) span a range of narratives of possible futures – from a collaborative world with strong international cooperation and rapid climate mitigations (e.g., SSP126) to more conflict-ridden societies with where mitigation is not a high priority (e.g. SSP370). The long lifetime of CO2 precludes any substantial reductions in the atmospheric CO2 concentration by year 2100 regardless of emission pathway (Arias et al., 2021). Aerosol emissions, however, are set to go down in all versions of the future, although some scenarios (in particular SSP370) have a delayed onset of the emission reductions. Given the tendency of BC to suppress turbulence and CO2 to enhance turbulence, as found in our idealized model simulations, both an increase in CO2 concentration and a reduction in BC concentrations will contribute to enhanced turbulence, with the potential positive consequence of an alleviation in the number of haze events. We now investigate the link between emissions and PBL height in future scenarios, to get insight into whether the boundary-layer feedback could have detrimental or advantageous impacts on near-surface pollution in the future.
In Fig. 4b, we show future BC emission changes for two selected SSPs; in SSP370 aerosol emissions keep increasing until around 2040-50 and decline thenceforth, while in SSP585 emissions decline from the beginning. A version of Fig. 4b showing also SSP119 and SSP245 is shown in supplementary Fig. S3. Figure 4c shows corresponding near-term (2015 to 2045) and long-term (2015 to 2100) trends in PBL height for the same two scenarios (see Fig. S4 for all SSPs). The effect of continued increases in BC emissions in the South Asian (predominantly India) region in SSP370 is clearly reflected in a continued regional reduction in PBL height in both near- and long-term. In contrast, the immediate emission reductions in SSP585 can be seen as an increase in PBL height which is strong over the Asian region but in general increases over all land regions. In SSP119, which has the strongest near-term emission reductions in BC (Fig. S3), even the near-term PBL height increases substantially over land (Fig. S4).
Depending on the exact emission pathway of both greenhouse gases and aerosols, the future is likely to see an increase in PBL height. If we follow the “middle of the road” pathway of SSP245 around 60% of the world’s population at the end of the century will be living in regions where the PBL height has increased since present-day. Further, around 6% (more than 400 million people) will be living in regions where present-day BC concentrations were very high (above the 95th percentile ) and where the increase in PBL height is set to be particularly strong (above the 90th percentile). In SSP370 this number is around 3%, in SSP585 it is around 13%, and in SSP119 it is more than 30% of the world’s population.
Recent evidence points to beginning reductions in aerosol emissions in the East China region. For instance, observations of BC concentrations in Beijing show a 67% reduction from 2012 to 2020 (Sun et al., 2021). Emission changes in the following century is therefore likely to involve a continued increase (or very weak reduction) in CO2 concentration in combination with reduced BC concentrations. As we see here, both of these changes have the potential to spur an increase in boundary layer turbulence and boundary layer height. Consequently, the number of severe pollution events, and associated detrimental effects on human health, are likely to be reduced in the future.