In this study the direct and diffuse solar radiation changes are estimated, and they contribute to the understanding of the observed global dimming and the more recent global brightening during the industrial era. Using a multistream radiative transfer model, the authors calculate the impact of changes in ozone, NO2, water vapor, CH4, CO2, direct and indirect aerosol effects, contrails, and aviation-induced cirrus on solar irradiances at the surface. The results show that dimming is most pronounced in central Africa, Southeast Asia, Europe, and northeast America. Human activity during the industrial era is calculated and accounts for a decrease in direct solar radiation at the surface of up to 30 W m−2 (30%–40%) and an increase in diffuse solar radiation of up to 20 W m−2. The physical processes that lead to the changes in direct and diffuse solar radiation are found to be remarkably different and the authors explain which mechanisms are responsible for the observed changes.
A decline in solar radiation has been observed at the surface (Liepert 2002; Stanhill and Cohen 2001). Called global dimming, this decline is presumed to be a consequence of an increased amount of scattering and absorbing aerosols and gases in the atmosphere from human activity and is likely to be linked to the reduced pan evaporation (Roderick and Farquhar 2002). Alpert et al. (2005) pointed out that the larger the population the stronger the decline in surface solar radiation. They estimated that urban activities between 1964 and 1989 explained the relatively large reduction in larger cities, estimated at a maximum of −0.41 W m−2 per year, compared to the much smaller reduction in rural areas. On a global scale, the decline in surface solar radiation based on ground-based measurements sites is estimated to be 7 W m−2 for the period 1961 to 1990 (Liepert 2002). A decline of 19 W m−2 for the United States in particular is observed in the same study.
The abovementioned studies are based on data from land observations carried out from the mid-1950s until the beginning of the 1990s. However, subsequent measurements of surface solar radiation taken between 1992 and 2002, at surface sites spread throughout the world, provide evidence of increasing insolation at the surface (Wild et al. 2005) called global brightening. This brightening in surface measurements is also supported by satellite measurements (Pinker et al. 2005).
Both global dimming and global brightening are functions of changes in the sum of direct and diffuse (scattered light) surface solar radiation. Therefore, it is important to quantify the changes in the direct and diffuse solar radiation throughout the industrial era up to the present, as well as to better understand the mechanisms behind them.
One factor that might explain change in surface solar radiation is the presence of clouds, due to their large variability and the extent to which they are influenced by anthropogenic aerosols (Kaufman et al. 2002; Ramanathan et al. 2001). Reduction in total cloud cover is normally consistent with increasing surface radiation. Qian et al. (2006), however, show the opposite and explain this by the appearance of haze caused by the pollutants that prevent solar radiation from completely penetrating down to the surface. Anthropogenic aerosols can potentially evaporate and inhibit cloud formation, especially in the case of absorbing aerosols (Ackerman et al. 2000), for example, atmospheric brown clouds (Ramanathan et al. 2005).
In addition to cloud cover, solar surface radiation is also affected by gases, aerosols (directly and indirectly), and contrails, all of which absorb or scatter (or both) solar radiation. Thus changes in these components will alter both diffuse and direct solar radiation. Most of the mechanisms responsible for the global dimming largely go toward masking global warming, although some may also contribute to it. The aim of this paper is to estimate direct and diffuse solar radiation changes and explain the observed changes in the total surface solar radiation over the industrial era.
We use a multistream radiative transfer model to accurately calculate the solar irradiances at the surface for direct and diffuse radiation. The model contains the discrete ordinate radiative transfer (DISORT) algorithm (Myhre et al. 2002; Stamnes et al. 1988) adopted with eight streams. It includes absorption by atmospheric gases, clouds, and Rayleigh scattering and calculates downward solar radiation at the surface for direct and diffuse components. A spectral resolution of four bands is used in the radiative transfer model (Myhre et al. 2002). The surface albedo is spectrally computed with solar zenith angle dependence included for the bands in the model (Myhre et al. 2003). Vegetation data for calculation of the surface albedo are from Ramankutty and Foley (1999). The meteorological data on temperature and cloud cover are from the European Centre for Medium-Range Weather Forecasts (ECMWF) for the year 2000. Radiative transfer calculations are performed for monthly means with 3-h time step (aerosol hygroscopicity for the direct aerosol effects is calculated every third hour). Calculations for the indirect aerosol effects are performed on daily data with 3-h time step. Annual means are calculated based on the monthly mean data (daily data for the indirect aerosol effect). The model resolution is T42 (approximately 3° × 3°) and with 40 vertical layers. Our results are given as the changes in direct and diffuse solar radiation at the surface due to changes in the components that play a role in global dimming and brightening. It uses the same cloud cover for preindustrial times and the present; thus, interannual variation is not taken into account. In the version of DISORT used in this study the delta-M scaling is not included. A large partition of the scattered solar radiation may be in the forward direction of the direct solar radiation. Aerosols and clouds scatter solar radiation mostly in the forward direction. Therefore, part of the direct solar radiation that is scattered into diffuse radiation may be at the surface difficult to distinguish from the direct solar radiation. Note that in the model we calculate the diffuse radiation as solar light that is scattered once or more. For some of the components, the results are also given in terms of radiative forcing at the top of the atmosphere to more clearly indicate to what extent they mask or enhance global warming.
Results for diffuse and direct solar radiation from changes in gases, aerosols (direct and indirect effects), and contrails are first shown separately over the industrial era. We then compile the data to derive the change in the total downward solar surface radiation. As a reference, Fig. 1 shows the distribution of direct and diffuse surface radiation with the current abundance of gases, aerosols, and clouds. All radiative transfer results in this paper are for downward solar irradiances, and their changes in figures and tables are all shown as annual mean values.
Over the last few decades, ozone has decreased in the stratosphere and during the industrial era increased in the troposphere (Gauss et al. 2006; Ramaswamy et al. 2001). In this work, we use data on ozone change since preindustrial times from a chemistry transport model (Oslo CTM2), which is part of a model intercomparison (Gauss et al. 2006). Figures 2a and 2b show the change in direct and diffuse solar radiation at the surface caused by changes in ozone concentrations. Since ozone absorbs solar radiation, direct and diffuse solar radiation will be changed with the opposite sign of the ozone changes. We find that ozone reductions in the stratosphere dominate over ozone increase in the troposphere, except in the Tropics. Loss of ozone in the stratosphere increases both direct and diffuse solar radiation at the surface, most strongly over Antarctica. The values are relatively weak for ozone changes, and globally the change in total ozone cause a brightening at the surface of 0.18 W m−2 (see Table 1).
NO2 is a precursor of ozone and therefore contributes indirectly to changes in solar radiation. However, it also has a direct effect because it absorbs solar radiation itself. Thus an anthropogenic increase in NO2 (Richter et al. 2005) reduces direct and diffuse solar radiation (see Figs. 2c,d). The impact of NO2 on solar surface radiation is more regional than that of ozone due to its shorter lifetime. The NO2 vertical-averaged column is high in urban areas in northern America, Southeast Asia, and Europe. These areas have had a NO2 increase of ∼1016 molecules cm−2 since preindustrial times, with maximum values at present of 1–2 × 1016 molecules cm−2. Based on the changes in NO2 during the industrial era, our simulation shows that the strongest values for the direct and diffuse components are −0.3 W m−2, and it contributes to dimming of around 0.04 W m−2 in global and annual mean. A global mean shortwave radiative forcing is calculated for the first time here for NO2 and is estimated to be 0.04 W m−2 or slightly more than 10% of the radiative forcing from tropospheric ozone (Gauss et al. 2006; Ramaswamy et al. 2001).
c. Water vapor
In addition to water vapor being the most important greenhouse gas, it also absorbs in the solar spectrum and can contribute to global dimming. There has been an observed increase in water vapor during the last 15 yr (1988–2003), which can be seen as a linear trend of 1.3 ± 0.3% decade−1 (Trenberth et al. 2005). This trend has not been homogeneous around the globe and is very likely a result of the increase in global surface temperatures and thus a climate feedback. The trend is also expected to have occurred before 1988, but this is somewhat more difficult to quantify. To investigate how water vapor may have affected current levels of solar radiation, we increased the column water vapor homogeneously with the change representative for the period from 1988 to 2003, according to Trenberth et al. (2005). Figures 2e and 2f show that the largest reduction in direct and diffuse solar radiation is at low latitudes, which has both the largest amount of water vapor and strong incoming solar radiation. The results (see Table 1) show a global reduction in direct and diffuse solar radiation at the surface of −0.15 and −0.14 W m−2, respectively. The local changes can be larger than 0.5 W m−2.
Methane also absorbs solar radiation (Collins et al. 2006). We have performed calculations for methane with an increase in the concentration since preindustrial time from 0.7 to 1.745 ppmv (Ramaswamy et al. 2001). In the calculations the reduced mixing ratio with altitude in the stratosphere is taken into account. Most of the methane absorption in the solar region takes place in the stratosphere; therefore, the diffuse solar radiation is barely changed at the surface (0.01 W m−2). The direct solar radiation is globally reduced by 0.08 W m−2 at the surface. The pattern of the change at the surface for CH4 in Figs. 2g and 2h can be related to the presence of cloud cover and partly to absorption of water vapor. Strongest reduction is present at approximately 30°N and 30°S with a maximum of −0.30 W m−2 in the Saharan Desert. Methane is not one of the major contributors to the surface radiation change even though it is stronger than NO2 on a global scale.
Collins et al. (2006) indicate that CO2 is a slightly weaker contributor to dimming than CH4. We have performed model simulations with a constant concentration in the atmosphere of 278 and 365 ppmv for preindustrial times and the present, respectively. The surface reduction in direct and diffuse solar reduction shows a similar pattern as CH4 due to clouds and partly due to overlapping absorption with water vapor (see Figs. 2i,j). A large fraction of the absorption of CO2 is also in the stratosphere, which explains the small magnitude of the change of diffuse solar radiation. However, the stratospheric absorption by CO2 is smaller than for CH4 and thus the reduction in the diffuse radiation is larger for CO2. Globally the direct radiation is reduced by 0.06 W m−2 and the diffuse part is decreased by 0.02 W m−2.
f. Direct aerosol effect
Anthropogenic activity has increased the aerosol content in the atmosphere. In this respect the most important components are sulfate, black carbon (BC), and organic carbon (OC) from both fossil fuel combustion and biomass burning. Aerosol datasets for both present and preindustrial times are based on simulations from the Oslo CTM2 chemistry transport model, which has been part of a global aerosol comparison study (AEROCOM; http://nansen.ipsl.jussieu.fr/AEROCOM/) (Textor et al. 2006). Emissions of aerosols and their precursors used in the calculations are from Dentener et al. (2006). Aerosol optical properties are described in Myhre et al. (2007). The aerosol simulations are performed with background aerosols, such as mineral dust and sea salt. In this study no anthropogenic dust is included.
Sulfate aerosols scatter solar radiation. Therefore, anthropogenic sulfate aerosols reduce direct solar radiation but increase diffuse radiation (see Figs. 2k,l). The magnitude of the changes is up to 20 W m−2, with somewhat larger changes for direct solar radiation than for diffuse radiation. The sum of these is the total change in solar radiation at the surface, which for scattering aerosol is similar to the radiative forcing. Organic carbon from fossil fuel is also mainly in the form of scattering aerosols, whereas BC from fossil fuel is mostly absorbing. Therefore, BC aerosols—like atmospheric gases and unlike scattering aerosols—reduce both direct and diffuse solar radiation. The fossil fuel and biofuel OC and BC can affect direct and diffuse solar radiation locally up to 10 W m−2 (see Figs. 2m,n). In accordance with other works, we find that BC has a much larger impact on solar radiation at the surface than at the top of the atmosphere. Particles from biomass burning affect solar radiation over particularly large regions in both Africa and southern America (see Figs. 2o,p). These particles are estimated to have reduced direct solar radiation by up to 20 W m−2 during the industrial era. Aerosols from biomass burning are partly absorbent, so in a clear sky they mainly scatter solar radiation, but under cloudy conditions with a high degree of scattered light, absorption is quite strong. Figure 2p shows that the solar diffuse radiation differs in sign over the globe for the biomass burning aerosols. Globally, the total direct aerosol effect reduces the amount of direct radiation by 2.8 W m−2 at the surface, while the amount of diffuse radiation is increased by 1.4 W m−2 (see Table 1). The radiative forcing of the total direct aerosol effect is −0.42 W m−2.
g. Indirect aerosol effect
An increase in aerosols leads to more numerous but smaller cloud droplets that reflect more solar radiation, also known as the cloud albedo effect. To calculate the change in effective radius from anthropogenic aerosols (those included in the direct aerosol section), we follow the approach in Quaas et al. (2006). Their relationship between the concentration of cloud droplets and aerosols is based on Moderate Resolution Imaging Spectroradiometer (MODIS) data. All hydrophilic aerosols, including natural aerosols such as subsize sea salt and secondary organic aerosols, are included in this approach. The radiative forcing due to the cloud albedo effect is slightly stronger in our simulations (−0.66 W m−2) than in Quaas et al. (2006). This is likely to arise from a higher spatial resolution (in particular vertical resolution) since larger variability in cloud liquid water content will strengthen the radiative forcing.
Figures 2q and 2r clearly show how the change in diffuse radiation is much stronger than the direct component, since when clouds are present the direct sunlight is already scattered to diffuse radiation. Southeast Asia is most influenced by the indirect aerosol effect. In this region, there has been an increase in sulfate and carbonaceous aerosols compared to preindustrial times, which reduces the effective radius. Here, the diffuse radiation has its maximum reduction close to 10 W m−2. Areas in Europe and North America are also influenced by the reduction of effective radius with a decline of approximately 5 W m−2 for diffuse solar radiation at the surface. We have not included the currently very uncertain second aerosol indirect effect (Albrecht 1989; Kaufman et al. 2005). By increasing the cloud amount, this effect would reduce the direct solar radiation and enhance the diffuse solar radiation.
h. Contrails and aviation-induced cirrus
Aircraft activity causes condensation trails (contrails), and under certain circumstances line-shaped contrails can evolve into cirrus clouds (Minnis et al. 1998; Schroder et al. 2000). We have used the same contrail cover as in Myhre and Stordal (2001) updated with the increase in air traffic used in Sausen et al. (2005). Stordal et al. (2005) estimated a trend of about 1%–2% decade−1 increase in cirrus cloud cover in regions with high levels of air traffic, and we adopted their total increase in cirrus cloud cover and related that to contrail cover. Figures 2s and 2t show changes in direct and diffuse radiation from a combination of contrail cover and aviation-induced cirrus. The impact on solar radiation shown in the figure is largest in northern America, Europe, and in the flight corridor between these two regions. The maximum reduction in direct solar radiation is in northeast America with a magnitude of −23 W m−2. The diffuse radiation increases in the same region by 19 W m−2. The global mean changes in solar radiation at the surface caused by contrails and cirrus is calculated to be −0.30 and +0.24 W m−2 for the direct and diffuse radiation, respectively.
i. Solar surface radiation change
Figure 3 shows the surface solar radiation change from ozone, NO2, and water vapor, as well as the total direct aerosol effect, cloud albedo effect, and aviation-induced contrails and cirrus. Changes in atmospheric gases have only a relatively small impact on the reduced total solar surface radiation in comparison with the other effects over land areas with the strongest dimming. The atmospheric gases play a significantly larger role to the dimming over the ocean and even become the dominating component in some regions. In the Arctic and Antarctica the gases cause an increase in the downward solar surface radiation and in particular at high southern latitude the gases have a dominating role for the total change in downward solar radiation. The results presented here show that the direct aerosol effect is a major contributor to global dimming, with a reduced solar surface radiation of 10 W m−2 in industrialized and biomass burning areas. Since the cloud albedo effect, only to a very small extent, had an impact on the direct solar radiation, the total solar surface radiation is similar to the change in the diffuse surface radiation with the greatest changes in the industrialized areas. Contrails and aviation-induced cirrus can also contribute to global dimming and reach a maximum reduction in the solar radiation over North America of 4 W m−2.
j. Total changes
Figures 4a and 4b show the total of all anthropogenic impact on direct and diffuse surface solar radiation. Regions in central Africa, Southeast Asia, Europe, and northeast America are most influenced by an anthropogenic reduction in the downward surface solar radiation. The direct solar radiation is reduced by up to 30 W m−2 and the diffuse radiation is increased up to 20 W m−2 in these areas. The reduction in direct solar radiation reaches 30% over certain parts of the eastern United States and 40% over certain parts of China. The diffuse solar radiation is strengthened by 30% in the eastern United States. In the figures of total changes in solar radiation we have summed each of the individual components. Several model sensitivity experiments were performed to investigate nonlinearities between the effects that cause global dimming. The calculations included two or more components causing the dimming in the same experiment (including experiments with all gases and all aerosols). The results showed near linearity between the mechanisms with a few exceptions in biomass regions. The total radiation at the surface has only been changed within 1%–3% compared to the additive approach. This can be explained by the partially differing regional occurrence but also the fact that the spectral behavior among the mechanisms varies.
The global mean changes in direct and diffuse radiation are −3.3 and +0.9 W m−2, respectively. The large change in North America is mostly due to contrails and cirrus from aircraft traffic and the direct aerosol effect, while surface solar radiation in southeast Asia is mostly influenced by direct and indirect aerosol effects. Figure 4c is the sum of direct and diffuse radiation and reveals the total change of solar radiation at the surface. The global dimming is pronounced over the continents, and the maximum reduction in the total solar surface radiation is around 10 W m−2. The reduction in total solar surface radiation of around 10 W m−2 over the United States and China is lower by a factor of 2 compared to that observed in 1961–90 in the United States (Liepert 2002) and also somewhat smaller than that observed in China in 1955–2000 (Qian et al. 2006). There are several explanations for the difference. Over the United States a large part of the increase in emissions of aerosols and their precursors occurred during the period 1961–90, with a reduction after 1990 (Lefohn et al. 1999; Novakov et al. 2003; Streets et al. 2006). In our simulations, changes over the whole industrial era are considered. Therefore larger changes in total surface solar radiation would be expected from the model than in the measurements. On the other hand, reductions since 1990 have occurred. In addition, Alpert et al. (2005) found that the global dimming was strongly linked to the population. In our model, simulations that are performed on a 3° × 3° horizontal resolution fail to represent the maximum values. Based on MODIS aerosol optical depth (AOD) (see Fig. 5), we find that within our model grid, the AOD and thus the global dimming may be higher by a factor of at least 2 than the average in certain regions. Thus the lack of sufficient resolution to resolve small scales in the model to a large degree explains the difference between the model and the measured global dimming because most of the measurements are made in urban areas. For other mechanisms, such as stratospheric ozone and aviation-induced contrails and cirrus, changes have occurred over the period with surface radiation measurements and for these only reduced stratospheric ozone can contribute to the reversal of global dimming.
Global dimming is a phenomenon that to a large degree takes place over certain land areas and the surrounding oceanic regions. We have shown that the spatial variability is large in our model results and that some of the causal mechanisms of global dimming have a spatial resolution that cannot be fully resolved in global models. Brightening has been observed over the last decade (Wild et al. 2005). The brightening is found to be largest in many high latitude measurement sites in accordance with our model results, and it is likely that stratospheric ozone is one of several contributors. This study shows that many mechanisms contribute to global dimming and brightening. Furthermore, the mechanisms that contribute to this have a larger and dissimilar impact on direct and diffuse solar radiation at the surface due to different physical processes. The reduction in the direct solar radiation is as large as 40% in the most industrialized and populated regions during the industrial era. This reduction in direct solar radiation may even be larger in more limited regions, which our model cannot resolve, with spatial inhomogeneities in the AOD, a substantial contributor to this reduction.
In this study it is shown that many atmospheric constituents contribute to the global dimming. The uncertainties vary among these components (Hansen et al. 2005; Ramaswamy et al. 2001). In particular the indirect aerosols effect from reduced cloud droplet size and aviation-induced cirrus are uncertain. Also the magnitude of the radiative effect of the direct aerosol effect (Schulz et al. 2006) and contrails (Sausen et al. 2005) differs between various estimates. We have neglected cloud cover changes due to aerosols or as a result of climate feedback since their sign, spatial pattern, and magnitude are even more uncertain than the components included in this study.
Our results also underscore the main distinction between human-induced and natural dimming as we have shown that dimming resulting from anthropogenic emissions has a distinct land signature. On the other hand, dimming from natural sources—largely volcanic eruptions leading to injection of aerosols in the stratosphere—is spread more evenly over ocean and land areas. After the volcanic eruption at Mt. Pinatubo in 1991, for example, optical depth increased over both land and ocean because of the large enhancement of stratospheric aerosols with potential impact on the vegetation CO2 uptake (Roderick et al. 2001). The restriction of anthropogenic dimming to land areas suggests that it may have a substantial impact on vegetation and agricultural production. While plants respond to both direct and diffuse radiation, they are most sensitive to changes in diffuse radiation because it affects a greater surface area of the plant (Roderick et al. 2001; Stanhill and Cohen 2001). Although our results show an increase in diffuse radiation over the course of the industrial era, the magnitude of the decrease in direct sunlight is greater. Thus the consequences of the total global dimming (both direct and diffuse) for vegetation and agricultural production need to be further investigated (Stanhill and Cohen 2001).
We show that solar absorption in the atmosphere from human activity has contributed to the dimming and is thus also likely to contribute to global warming. This solar absorption is mainly due to absorbing aerosols and gases, such as NO2, tropospheric ozone, water vapor, CO2, and CH4. However, this absorption is unable to explain the observed global dimming in industrialized regions and indicates that scattering components have a major role in global dimming. Thus, it is likely that human-influenced scattering by aerosols and clouds has contributed to a substantial offsetting of the warming from the greenhouse gases.
We thank Lynn P. Nygaard and Frode Stordal for valuable improvements in earlier versions of the manuscript. We appreciate the constructive and useful comments from two anonymous reviewers.
Corresponding author address: Gunnar Myhre, Department of Geosciences, Postbox 1022 Blindern, 0315 Oslo, Norway. Email: firstname.lastname@example.org