Contrasting Influences of Recent Aerosol Changes on Clouds and Precipitation in Europe and East Asia

Camilla W. Stjern Department of Geosciences, University of Oslo, Oslo, Norway

Search for other papers by Camilla W. Stjern in
Current site
Google Scholar
PubMed
Close
and
Jón Egill Kristjánsson Department of Geosciences, University of Oslo, Oslo, Norway

Search for other papers by Jón Egill Kristjánsson in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Over the last few decades, aerosol loadings have increased greatly over Southeast Asia, while Europe and North America have experienced huge reductions. Previous studies have suggested that these changes may have influenced the temperature trends as well as precipitation patterns due to the direct and semidirect aerosol effects. Here, an Earth system model with parameterized aerosol–radiation and aerosol–cloud interactions is used to investigate changes in cloud properties and precipitation between 1975 and 2005. This is done globally as well as for the two focus areas Europe and East Asia. Despite systematic changes in cloud droplet number concentration and cloud droplet size, changes in stratiform precipitation are less clear. In both regions there is a dominance of autoconversion over liquid water accretion as the primary precipitation release mechanism, which alone should imply a strong sensitivity to changes in cloud droplet size. However, in these areas liquid water paths are relatively low and background concentrations are high, which produce low simulated precipitation susceptibilities. High susceptibilities are instead found over remote ocean regions, in agreement with expectations. For convective precipitation, both regions show statistically significant changes that are consistent with oppositely signed changes in direct aerosol forcing over Europe and East Asia, respectively.

Denotes Open Access content.

Corresponding author address: Camilla W. Stjern, Department of Geosciences, University of Oslo, P.O. Box 1022 Blindern, 0315 Oslo, Norway. E-mail: c.w.stjern@geo.uio.no

Abstract

Over the last few decades, aerosol loadings have increased greatly over Southeast Asia, while Europe and North America have experienced huge reductions. Previous studies have suggested that these changes may have influenced the temperature trends as well as precipitation patterns due to the direct and semidirect aerosol effects. Here, an Earth system model with parameterized aerosol–radiation and aerosol–cloud interactions is used to investigate changes in cloud properties and precipitation between 1975 and 2005. This is done globally as well as for the two focus areas Europe and East Asia. Despite systematic changes in cloud droplet number concentration and cloud droplet size, changes in stratiform precipitation are less clear. In both regions there is a dominance of autoconversion over liquid water accretion as the primary precipitation release mechanism, which alone should imply a strong sensitivity to changes in cloud droplet size. However, in these areas liquid water paths are relatively low and background concentrations are high, which produce low simulated precipitation susceptibilities. High susceptibilities are instead found over remote ocean regions, in agreement with expectations. For convective precipitation, both regions show statistically significant changes that are consistent with oppositely signed changes in direct aerosol forcing over Europe and East Asia, respectively.

Denotes Open Access content.

Corresponding author address: Camilla W. Stjern, Department of Geosciences, University of Oslo, P.O. Box 1022 Blindern, 0315 Oslo, Norway. E-mail: c.w.stjern@geo.uio.no

1. Introduction

The indirect effect of aerosols on climate incorporates several types of interactions. In the “Twomey effect” an addition of aerosols tends to lower cloud droplet sizes and increase cloud reflectivity for constant liquid water path (Twomey 1977). “Cloud adjustments” concern the aerosol impact on cloud cover, cloud lifetime, and precipitation (Albrecht 1989). The latter has proven a particularly complex part of the indirect aerosol effects, and observational and model-derived evidence of how aerosols influence precipitation remains elusive (McComiskey and Feingold 2012; Ayers and Levin 2009). One explanation for the divergent results of aerosol–precipitation studies is the dependence on the meteorological context of the cloud receiving the aerosol perturbation. Specifically, the precipitation response has been shown to depend on cloud regime (e.g., Lebsock et al. 2008; Ruckstuhl et al. 2010; Stevens and Feingold 2009), the relative humidity of entrained overlying air (Ackerman et al. 2004), the wind shear (Khain 2009), and liquid water path (Feingold et al. 2013). Another confounder lies in the fact that changing aerosol levels can also affect precipitation through pathways other than the cloud microphysical properties, for instance by affecting large-scale circulation patterns (e.g., Gu et al. 2006) or by modifying the local atmospheric stability through absorption of solar radiation (Lohmann and Feichter 2001). Moreover, precipitation will act to diminish the aerosol concentration (Zhao et al. 2006), adding complexity to an interactive system of forcings and feedbacks where it is difficult to separate cause from effect. In the present study, we seek to quantify interactions between aerosols and precipitation from warm clouds using a global climate model (NorESM; Bentsen et al. 2013) with state-of-the-art parameterizations for aerosol–cloud–climate interactions (Kirkevåg et al. 2013). Special attention will be given to regions of Europe and East Asia that in recent decades have experienced large changes in aerosol concentrations of particulate sulfate (SO4; Smith et al. 2011), black carbon (BC; Bond et al. 2013), and organic matter (OM) (Kanakidou et al. 2005). While SO4 is well known for its ability to act as cloud condensation nuclei (CCN; Aitken 1880), BC aerosols have the potential to affect precipitation through their direct and semidirect effects (Lohmann and Feichter 2001; Zhuang et al. 2013). Secondary organic aerosols (SOA) are receiving increasing attention because of their potential to serve as CCN (e.g., Lambe et al. 2011).

We run the NorESM for two time slices, year 2005 and year 1975, keeping greenhouse gas concentrations constant, and investigate the model’s response to differences in aerosol loading. In particular, we investigate how the aerosol changes influence precipitation and relate our findings to recent observational studies (e.g., Stjern et al. 2011; Fan et al. 2012).We also calculate the model’s precipitation susceptibility (Sorooshian et al. 2009), that is, the sensitivity of its release of precipitation to changes in cloud droplet number concentration, and investigate the dependence of the susceptibility on liquid water path and SO4. The results are compared to published estimates of this quantity based on observations and cloud-resolving model simulations.

In the next section, we present the observed changes in air pollution and precipitation in our regions of interest, followed by an account of data and methods in section 3. Section 4 provides a brief evaluation of the model’s representation of cloud and aerosol specifics. In section 5, we give the results of our investigation. We there present the modeled 1975–2005 changes of aerosols, cloud microphysical properties, and precipitation and go on to explore precipitation susceptibility and the ratio between the two mechanisms of precipitation release in warm clouds: collisions and coalescence of cloud droplets, often termed autoconversion (Kessler 1969) and collection, also termed warm rain accretion. Results are summarized in section 6.

2. Earlier findings for Europe and East Asia

Europe and East Asia are regions with differences in both climatology and aerosol background concentrations. The precipitation pattern in Europe is dominated by westerly midtropospheric flow with frequent low pressure systems over the North Atlantic and associated frontal systems entering the European continent, particularly in the winter (e.g., Bárdossy and Caspary 1990). In the summer, a large portion of the precipitation is convective, especially in central Europe. East Asia, on the other hand, has a climate that is strongly influenced by the Asian monsoon, with most of the precipitation coming in the summer and often associated with the quasi-stationary mei-yu front (e.g., Ding et al. 2008). In both regions, the annual-mean cloud cover of stratocumulus clouds has been found to be around 15%–25% (Wood 2012).

Similarities between the regions can be found in the relative concentration of sulfate in the total aerosol mass; Lanz et al. (2010) investigated results from field campaigns in central Europe for the period 2002 to 2009 and found that sulfate made up 3%–26% of the total mass. Based on observations of 16 sites in China between 2006 and 2007, Zhang et al. (2012) estimated that sulfate constituted about 16% of the total aerosol there. The studies showed, however, that Europe had a higher percentage of organic matter and nitrates, while East Asia had higher concentrations of dust.

In Europe, the ambient concentration of sulfur species has decreased by as much as 90% since 1980 (Tørseth et al. 2012). Largest of all were the emission reductions in a region of central Europe commonly referred to as the Black Triangle (BT), where numerous coal plants contributed to extensive sulfur emissions that culminated in the late 1980s. In Asia, on the other hand, sulfur dioxide emissions increased by 199% between 1980 and 2003 (Ohara et al. 2007), and Larssen et al. (2006) reported that in some regions of China the deposition of sulfur is now even larger than it was in the BT in the early 1980s.

For Europe, Krüger and Graßl (2002), analyzing satellite retrievals of cloud reflectance and radiative transfer calculations, found a reduction in cloud albedo between the late 1980s and the late 1990s, yielding a +1.5 W m−2 change in top of atmosphere shortwave (SW) radiative forcing. Using collocated global solar irradiance measurements and synoptic cloud observations, Ruckstuhl et al. (2010) calculated transmitted shortwave radiation for northern Germany and found that the indirect aerosol effect for low-level stratiform cloud types changed by +0.6 W m−2 between 1981 and 2005, while the direct radiative effect increased by +3.53 W m−2. Norris and Wild (2007) studied the effect of aerosols on clouds and radiation over Europe for the 1965–2004 period using monthly gridded data from the Global Energy Balance Archive (GEBA), synoptic cloud reports, and the International Satellite Cloud Climatology Project (ISCCP). They argued against a strong impact from cloud adjustments, as increasing cloud cover was observed at the same time as decreasing sulfate loads. Rulfová and Kyselý (2014), who investigated surface observations for the Czech Republic between 1982 and 2010, found significant rises in convective precipitation, although no trend was detected for stratiform precipitation. Stjern et al. (2011) found from surface observations a 23% statistically significant increase between 1983 and 2008 in the number of days with rain showers in the Black Triangle region, and a 56% increase in light precipitation frequency in a coastal region of western Europe with low background concentrations. They found no significant trends in precipitation amounts for either region.

As a result of the large Asian emission changes, trends are observed in both clear-sky and cloudy radiation fluxes there. For instance, based on surface solar radiation measurements Che et al. (2005) found a 6.6 W m−2 decade−1 decrease in direct solar radiation over China between 1961 and 2000, in spite of decreasing cloud cover. Precipitation changes associated with the increasing emissions are reported in numerous studies. Zhao et al. (2006) analyzed surface precipitation measurements, the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data and meteorological sounding data and found a significant precipitation reduction over eastern China that was strongly correlated with the increasing aerosol levels. They linked the decrease in precipitation to increased atmospheric stability due to absorbing aerosols. Similarly, climate model simulations by Menon et al. (2002) suggested that recent contrasting precipitation anomalies between northern and southern parts of China might be attributed to increasing aerosol loadings over Southeast Asia. Fan et al. (2012) combined measurements and model simulations for eastern China and found that increasing CCN concentrations led to significant changes in the spatial and temporal distribution of precipitation as well as to delayed precipitation onset. The large changes in aerosol concentrations combined with the contrasting effects of these changes on precipitation make Europe and East Asia particularly interesting areas of study.

In the present analysis, we define East Asia as the region between 20° and 40°N and 100° and 130°E, Europe as the region between 39° and 60°N and 12.5°W and 32.5°E, and BT as the region between 48° and 52°N and 10° and 17.5°E. All three regions are marked in Fig. 1a.

Fig. 1.
Fig. 1.

(a) Location of the regions of interest. Europe: blue, Black Triangle: green, and East Asia: red. (b) Location of the EMEP stations (dots) used in the comparison.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

3. Data and methods

a. EMEP and CRU data

Monthly mean data from 90 stations in the European Monitoring and Evaluation Programme (EMEP) were downloaded (http://www.nilu.no/projects/ccc/emepdata.html) for the years 2003–08 and utilized to perform an evaluation of the NorESM model concentrations of SO4. The locations of these stations are shown in Fig. 1b. At the EMEP sites, SO4 is measured daily by 24-h exposure to sampling filters, which are submitted to subsequent laboratory analyses to get a measure of the SO4 concentrations in units of μg S m−3. The type of filters used varies between the stations, but all are subject to an accuracy demand of 10% or better (see Berg et al. 2002).

For evaluation of NorESM surface precipitation amounts, we use gridded precipitation data from the Climatic Research Unit (CRU) for the years 2003–08 based on meteorological surface observations and downloaded online (http://www.cru.uea.ac.uk). Here, monthly station anomalies (from 1961 to 1990 means) are interpolated into a 0.5° latitude–longitude grid and combined with an existing climatology to obtain absolute monthly values (Harris et al. 2014). Compared to precipitation from the Global Precipitation Climatology Project, CRU has been shown to have a negative bias of 27.6 mm yr−1 for global annual land surface precipitation amounts (Fekete et al. 2004).

b. NorESM

The model tool in this study is the Norwegian Earth System Model (Bentsen et al. 2013), which is based on the National Center for Atmospheric Research (NCAR) Community Climate System Model, version 4 (CCSM4), but with its own aerosol life cycle scheme as well as parameterizations of aerosol–radiation and aerosol–cloud interactions (Kirkevåg et al. 2013). The aerosol module accounts for prognostic sea salt (lifetime τ of 0.28 days), sulfate (SO4; τ = 3.84 days), particulate organic matter (OM = primary organic aerosols + SOA; τ = 7.63 days), black carbon (τ = 8.10 days), and mineral dust (τ = 2.55 days). NorESM uses its own ocean model, which is a modified version (Bentsen et al. 2004) of the Miami Isopycnic Coordinate Ocean Model (MICOM; Bleck et al. 1992), but in our particular model setup, the atmospheric model component is instead coupled to a slab ocean model as well as with fully interactive sea ice [Community Ice Code, version 4.0 (CICE)] and land [Community Land Model, version 4 (CLM)] models. The horizontal resolution is 1.9° × 2.5°, with 26 vertical layers that follow the terrain in the lower troposphere and gradually transition to pressure levels when entering the lower stratosphere. The top-of-atmosphere (TOA) aerosol direct and indirect forcings in NorESM—both defined as the changes from the year 1850 to 2000 using CMIP5 emissions—are respectively −0.10 and −0.91 W m−2 (Kirkevåg et al. 2013). The former is weaker (i.e., less negative) than the CMIP5 intermodel average of effective radiative forcing from 1850 to 2000 (−0.45 ± 0.50 W m−2; Myhre et al. 2013a), mainly due to a strong positive forcing from BC (Myhre et al. 2013b). The negative indirect forcing of −0.91 Wm−2 is rather strong compared to the effective forcing from aerosol–cloud interactions in other CMIP5 models (ranging from 0 to −1.2 W m−2 with a best estimate of −0.45 W m−2; Myhre et al. 2013a).

c. Microphysical parameterizations for aerosols and clouds

The aerosol scheme in NorESM is described in detail in Kirkevåg et al. (2013), so only a brief overview is given here: The scheme calculates mass concentrations of aerosol species, which are tagged according to production mechanisms in both clear and cloudy air. For each tagged mass concentration, there are four size modes (nucleation, Aitken, accumulation, and coarse mode). The aerosol types predicted by the scheme are SO4, BC, OM, sea salt, and mineral dust. In addition, the precursor gases dimethyl sulfide and sulfur dioxide are predicted, while oxidant concentrations for the sulfate chemistry are prescribed. Particle number concentrations and sizes are calculated by combining assumed size distributions for emitted or produced primary particles with calculations of subsequent growth by condensation, coagulation, or aqueous chemistry. To limit computational cost, physical properties of the aerosols, including the optical properties, are estimated by interpolating between precalculated values in lookup tables, using process-tagged aerosol mass concentrations, and ambient relative humidity as input. The lookup tables provide (i) spectrally resolved optical properties that are then used to estimate the direct effect of aerosols in the model and (ii) aerosol modal size parameters that are used as input in the calculation of aerosol activation and thereby aerosol indirect effects.

Compared to more traditional modal or sectional schemes, an advantage of the NorESM aerosol scheme is that the degree of internal versus external mixing can be estimated based on physicochemical processes instead of through explicit assumptions. A disadvantage is that there is no explicit information about the size and mixing state of the aerosol masses after growth.

The treatment of cloud microphysics in NorESM is based on the bulk microphysics scheme of Rasch and Kristjánsson (1998), with the following extensions that account for aerosol–cloud interactions:

  1. cloud droplet number concentration (CDNC) is predicted (Storelvmo et al. 2006) using the Abdul-Razzak and Ghan (2000) activation scheme along with parameterized updraft velocities following Morrison and Gettelman (2008);

  2. aerosol concentrations, sizes, and hygroscopic properties are obtained from the NorESM aerosol scheme and used in the activation scheme (Kirkevåg et al. 2013); and

  3. the size of detraining cloud droplets from convection depends on the number of activated aerosols at cloud base (Hoose et al. 2009).

Cloud droplet size, here expressed by the effective radius (REFF), is determined by CDNC as well as by cloud liquid water content (LWC):
e1
Here, β is the spectral shape factor of the cloud droplet size distribution, expressed as an increasing function of CDNC following Rotstayn and Liu (2009), and ρl is the density of bulk water (1000 kg m−3). A quantity related to LWC is the liquid water path (LWP), which is the vertical integral of LWC.
As described by Rasch and Kristjánsson (1998), stratiform precipitation including detrained cloud water from convection is calculated by the parameterizations of autoconversion
e2
and collection (liquid water accretion)
e3
In Eqs. (2) and (3), Cl,aut and Cracw are empirical coefficients; ρa is the density of dry air; denotes the in-cloud mixing ratio of cloud water = LWC/ρa; H is the Heaviside function (=1 for a positive argument, = 0 for a negative argument); rυ is the mean volume radius = reff/β; rυc is the threshold value of rυ for onset of precipitation, set to 14 μm in NorESM (Kirkevåg et al. 2013); and qr denotes the rainwater mixing ratio.

Convective precipitation, which is obtained from the Zhang and McFarlane (1995) deep convection scheme and the Hack (1994) shallow convection scheme, is not based on any elaborate cloud microphysics parameterization, except for detrained cloud water, as described above.

While the use of diagnostic precipitation, as in NorESM, is still the most common practice in global climate models, models with microphysics schemes featuring prognostic precipitation are now starting to appear (e.g., Gettelman et al. 2015). Posselt and Lohmann (2009) performed experiments comparing model simulations with prognostic versus diagnostic rain schemes and found that the diagnostic scheme tended to put too much emphasis on rain production from autoconversion. The potential significance of this to the present study is commented further in section 5f. The version of NorESM used here does not have any link between aerosols and ice nucleation, so this study will focus on liquid clouds.

d. Experiment setup

We have performed three pairs of simulations, each pair consisting of simulations for two different time slices: one using year 1975 global CMIP5 emissions (van Vuuren et al. 2011) and another with year 2005 global CMIP5 emissions. All simulations use year 2000 greenhouse gas concentrations, solar constant, and initial state. We carried out two sets of simulations:

  1. The first pair of simulations was run in a “fully interactive” model setup, in which the prognostic NorESM aerosols influence the model evolution, thereby enabling the climate system to respond to the aerosol forcing.

  2. The second pair of simulations was run in a “noninteractive” model setup, in which the prognostic aerosols do not influence the model evolution, but additional diagnostic calls to the cloud microphysics and radiation schemes are used to obtain indirect forcing as well as changes in cloud properties (Kristjánsson 2002).

That is, the noninteractive runs for the years 1975 and 2005 have the same “meteorology.” There is a prescribed aerosol (Neale et al. 2010) with an associated direct effect, which is unchanged between the two time slices. The cloud microphysics treatment is the same as described in the preceding subsection, except that CDNC is now prescribed (Rasch and Kristjánsson 1998; Neale et al. 2010). The noninteractive runs enable us to study how cloud properties and precipitation release are altered by a change in emission levels, without including feedbacks (e.g., changes in surface or atmospheric temperatures, changes in liquid water path, or changes in cloud cover) that normally complicate the detection of aerosol signals in precipitation. Note that while we do have a noninteractive variable for LWP, cloud cover changes are only enabled in the fully interactive mode, and thus our detection of the “cloud lifetime” part of the cloud adjustments is somewhat incomplete.

The fully interactive and noninteractive simulations were run for 25 yr after a 20-yr spinup, producing monthly mean output. This means that for each simulation, there are essentially 25 samples (ensemble members) of the annual cycle, enabling a robust separation of the aerosol signal from “meteorological noise.” In addition, we performed a third pair of simulations, with the model setup in a “fully interactive” mode producing “instantaneous output” (at each 20-min time step). These simulations were initiated by the 20-yr spinup but run for only 2 yr.

In the following sections, numbers and figures are from the fully interactive runs with monthly output unless otherwise stated.

4. Model evaluation

a. Model representation of aerosols

The ability of NorESM to simulate aerosol concentrations was evaluated by Kirkevåg et al. (2013), who found that averaged over all AERONET sites, NorESM clear-sky aerosol optical depth (AOD) and absorption aerosol optical depth (ABS) have negative biases of 8% and 32%, respectively, compared to AERONET measurements. In Europe, AOD is slightly overestimated due to a combination of a slight underestimation in winter and a larger overestimation in summer. This is consistent with the findings for central European sites in Makkonen et al. (2014), who compared NorESM aerosol number concentrations to observational data from various sources. For East Asia, Kirkevåg et al. (2013) found a negative bias in AOD. They found ABS to be underestimated for both Europe and East Asia.

We now proceed to evaluate how well NorESM represents observed values of sulfate for Europe. In this context, it is worth noting that in the model, anthropogenic SO2 emissions are assumed constant throughout the year. This is obviously an oversimplification. However, model simulations by Rasch et al. (2000) indicate that seasonal variation in SO2 emissions has a smaller impact on SO4 concentrations than the seasonal variations in, for example, cloud amount, vertical transport, or oxidant amount. Similar conclusions have been obtained in observational studies. For instance, Lövblad et al. (2004) analyzed SO2 emissions for Europe as well as sulfate measurements at a selection of EMEP sites and pointed to the importance of local weather conditions to the seasonal sulfate cycles. In a combined model–observation study of China, Zhang et al. (2010) concluded that the wet deposition induced by the Asian monsoon is a strong driver of the seasonal aerosol cycles and found similar seasonal variations in sulfate in simulations with constant and seasonally varying SO2 emissions. The model’s representation of clouds and precipitation will be presented in sections 4b and 4c, respectively.

Below, monthly averages of sulfate (measured in aerosols by means of sampling filters) for a given EMEP station are compared to the SO4 values of the corresponding NorESM grid box. High-altitude stations located above 2000 m above mean sea level are matched to the model layer that closest matches the station’s altitude. EMEP data from the years 2003–07 are compared to the last 5 yr of the model run using year 2005 emissions. We find an overall model bias of +7% and a statistically significant correlation coefficient of 0.52 (see Fig. 2). The average seasonal cycle of SO4 for Europe is more pronounced in NorESM than for the EMEP data (Fig. 3). NorESM has maximum simulated values in the summer and lower values in the winter, whereas the observed values have maxima around February/March. As we will see in section 4c, precipitation measurements show that February through April are the driest months in this region, which allows for higher atmospheric sulfate concentrations due to inefficient wet removal. Combined with increasing hours of sunlight (and therefore increasing oxidation from SO2 to SO4) at this time of year, the February/March maximum in observed SO4 concentration is not unexpected even in the absence of a seasonal emission cycle. Although oxidation is highest in the summer, convective precipitation during this season provides rapid wet deposition and thereby lowers the sulfate concentrations again after the springtime maximum. In NorESM, however, the driest months of the year over Europe are in the summertime (section 4c), which combined with the efficient summertime oxidation from SO2 to SO4 means that the simulated sulfate cycle peaks in the summer instead of in the spring. For an earlier version of the model Iversen and Seland (2002) noted that the overestimate for SO2 in summer in Europe may also be due to neglected vertical convective transport and inefficient scavenging in convective clouds. As sulfate is the largest contributor to AOD over the European continent (Kirkevåg et al. 2013), the positive summertime bias in simulated sulfate is probably the main cause of the positive summertime AOD bias compared to AERONET data as mentioned above.

Fig. 2.
Fig. 2.

Scatterplot for EMEP and NorESM SO4 values. For each EMEP station, the observed value is compared to the value of the grid box that the station coordinates fall within. Colors mark the density of observation pairs within each sulfate concentration bin.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

Fig. 3.
Fig. 3.

Seasonal cycles of SO4 in Europe for NorESM (blue) and EMEP (red).

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

Finally, NorESM does not include ammonium and nitrate, which according to analyses of field studies by Lanz et al. (2010) for central Europe between 2006 and 2009 constitute 5%–15% and 8%–36% of the total mass, respectively. These constituents have been shown to have a potential to contribute to the concentration of cloud condensation nuclei (Petters and Kreidenweis 2007). How the lack of these constituents in our model influences the total aerosol–precipitation effect is not clear, but it would most likely increase the simulated AOD, thereby worsening the positive summer bias, while improving the negative wintertime bias. Still, as the trends of ammonia and nitrate from 1975 to 2005 have been much weaker than for sulfate (Fagerli et al. 2007), we do not expect it to affect the overall results strongly.

b. Model representation of clouds

NorESM tends to underestimate cloudiness, and Bentsen et al. (2013) found a bias of −13% to −24% compared to ISCCP or CloudSat satellite retrievals, respectively. At the same time, they found that the liquid water content in clouds is generally exaggerated (particularly in the extratropical storm track regions), and while Jiang et al. (2012) obtained a globally averaged LWP of 30–50 g m−2 (15 to 102 g m−2 uncertainty limit) based on NASA A-Train satellite retrievals, NorESM produces a global-mean LWP of 125 g m−2 for the period 1976–2005 (Bentsen et al. 2013). In terms of radiative impact, the biases in cloud cover and LWP tend to cancel, and therefore the simulated cloud radiative effect is realistic. The underestimated cloudiness may, however, contribute to an underestimation of aerosol–cloud interactions, particularly if NorESM underestimates marine stratiform clouds that generally are quite susceptible to aerosol influence. Too high liquid water content, on the other hand, will reduce the clouds’ susceptibility to aerosol influence.

Additionally, Alterskjær et al. (2012) found that NorESM has lower CDNC than an observational study by Quaas et al. (2006), who calculated CDNC based on MODIS retrievals. This would tend to increase the sensitivity of clouds to aerosol changes. Consistently with that, Alterskjær et al. (2012) showed that the cloud albedo from NorESM simulations had a higher susceptibility to aerosol changes than calculations based on MODIS data, which means that the effects of the negative CDNC bias may be dominating over the effects of the positive LWP bias in terms of influence on cloud microphysics. However, it is not given that this would also imply a higher susceptibility of precipitation to aerosol perturbations in NorESM. Precipitation susceptibility will be investigated in more detail in section 5e below.

c. Model representation of precipitation

Here, we take a closer look at the model’s ability to reproduce observed precipitation amounts in Europe and in East Asia. CRU surface observations of precipitation from 2003 to 2007 are compared to the last 5 yr of the fully interactive NorESM model run using 2005 aerosol emissions (see Fig. 4). CRU grid values are interpolated to fit the NorESM grid, and we compare the precipitation values grid cell by grid cell.

Fig. 4.
Fig. 4.

Scatterplot for CRU and NorESM daily mean precipitation amounts for (a) Europe and (b) East Asia. Colors mark the density of observation pairs within each precipitation amount bin.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

On average, when compared to CRU, NorESM overestimates daily precipitation values by 17% for Europe and by 29% for East Asia. The positive bias in the East Asian region is particularly large over the Himalayas and similarly the overestimated European precipitation seems to be focused around the Alps (not shown), both of which seem to be related to the model’s inability to resolve orographic effects because of its coarse spatial resolution. Correlations between CRU and NorESM precipitation values are 0.60 and 0.37 (both significant at the 99% level) for Europe and East Asia, respectively, when based on annual means. However, when calculating correlations based on seasonal (month by month) averages, the correlation drops to 0.40 for Europe and rises to 0.69 for East Asia (both significant at the 99% level). This indicates that NorESM does not capture the seasonal variations over Europe very well but does a better job at this for East Asia (Fig. 5). The erroneous seasonal precipitation cycle over Europe may be related to a tendency of NorESM to underestimate blockings over the North Atlantic during wintertime (Iversen et al. 2013), which would lead to an overestimation of cyclone activity and hence precipitation for this season.

Fig. 5.
Fig. 5.

Seasonal cycles of precipitation for (a) Europe and (b) East Asia (NorESM: blue; CRU: red).

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

5. Results and discussion

a. Precipitation and aerosol characterization for Europe and East Asia

Our model simulations show that the average cloud cover over Europe and East Asia is highest in wintertime and summertime, respectively (not shown), consistent with the dominant circulation patterns of the two regions mentioned in section 2. The vertical distribution of cloud amount is similar between the regions with the highest cloud amounts around 900 hPa and a secondary maximum around 300 hPa (not shown).

The seasonal variation in total precipitation for Europe (Fig. 5a) is dominated by the seasonal variation in stratiform precipitation, with maxima in winter. Convective precipitation in Europe has a weaker cycle with a maximum in summertime. In East Asia, both convective and stratiform precipitation have maxima in the summertime. While stratiform precipitation is associated with similar daily precipitation sums for both regions, a typical (averaged over all seasons) convective precipitation event is associated with 2.5 mm day−1 for China but below 1.5 mm day−1 for Europe.

In both Europe and East Asia, the concentrations of SO4 and BC are highest in the second lowest vertical layer of the model, corresponding to around 700-m altitude, decreasing gradually with height (not shown). As mentioned in section 2, observations indicate that the relative amount of sulfate around year 2005 is similar between the regions in spite of much higher absolute concentrations in East Asia than in Europe. This is reproduced in the model: for the 2005 simulations, SO4 constitutes 31% of the sum of SO4, BC, and OM for Europe and 26% for East Asia. For the 1975 simulations, however, the numbers were 49% and 21%, respectively. Consistent with the observations by Zhang et al. (2012), mineral dust and OM are the aerosol constituents that make up the largest part of the total simulated aerosol mass in both Europe and East Asia.

b. Changes in SO4, BC, and OM column burdens

The difference in the SO4 column burden between the 1975 and 2005 simulation is shown in Fig. 6a. There was a global-mean decrease of 6% between these years but embedded in this number is a great deal of cancellation due to European and North American emission reductions and concomitant Asian emission increases (Table 1). We find that (relative to 1975 values) simulated European SO4 levels decreased by 59% between 1975 and 2005, while East Asian levels increased by 71%. Both changes were statistically significant at the 99% level (Student’s t test produced a p value below 0.01).

Fig. 6.
Fig. 6.

The 2005 minus 1975 difference (red colors indicating increase and blue colors indicating decrease) in column burdens of (a) of SO4, (b) BC, (c) OM, and (d) the sum of SO4, BC, and OM.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

Table 1.

2005 minus 1975 dual difference in meteorological variables, based on the fully interactive runs. Percentages are calculated relative to 1975 values, and Student’s t test (with 300 as the sample size as both simulations are run for 25 yr and have 12 months yr−1) is used to calculate if the differences were statistically significant. A square denotes a p value below 0.1 (90 % significant), a triangle denotes a p value below 0.05 (95 % significant), and an asterisk denotes that the p value was below 0.01 (99 % significant). Calculation of stratiform (PRECS) and convective (PRECC) precipitation is explained in section 3c, and PRECT is the sum of these two.

Table 1.

The 2005 minus 1975 difference in BC column burden has a geographical pattern very similar to that of the SO4 change (see Fig. 6b) but is even more localized around the European/Asian sources. The BC burden increased globally by 34%, largely due to the 83% increase in East Asia (see Table 1). At the same time, European BC burdens decreased by 13%.

The column burden of OM shows a similar pattern of decrease over Europe and increase over East Asia, but high emissions from biomass burning in 2005 are responsible for particularly large increases over Indonesia and eastern Russia/northeast Asia (Fig. 6c), the latter of which influences the 1975–2005 difference over the East Asian region; while Europe has a difference of merely −2% in OM between 1975 and 2005, East Asia has a significant difference of 27% (see Table 1).

The sum of the three constituents above decreased by 27% (2.8 μgm−2) over Europe and increased by 40% (4.2 μgm−2) over East Asia (Fig. 6d).

c. Changes in cloud microphysics

In response to the changing emission levels between 1975 and 2005, the simulated cloud droplet number concentration declined by 35% over Europe and increased by 32% over East Asia (see Fig. 7a). The transition to lower/higher concentrations of cloud droplets caused the mean effective droplet radius in Europe and Asia to increase by 3% and decrease by 3%, respectively (see Fig. 7c). Increased droplet radius over Europe is consistent with satellite observations of reduced cloud albedo between the 1980s and 1990s (Krüger and Graßl 2002). The relatively strong increases in CDNC and decreases in REFF that can be seen over eastern Russia/northeast Asia (45°–65°N, 100°–150°W) are related to the before mentioned strong increase in OM (Fig. 6c) over this region. Although SO4 aerosols in the model have a hygroscopicity parameter (kappa value) that is about 4 times as high as for OM and therefore are considerably more efficient CCN, the OM increase in this region is large enough to have a significant influence on the cloud properties. OM also shows a decrease over central Europe, probably contributing to the cloud microphysical changes seen here (reduction in CDNC and LWP).

Fig. 7.
Fig. 7.

The 2005 minus 1975 difference (red colors indicating increase and blue colors indicating decrease) in (a) cloud droplet number concentration for the vertical model level approximately corresponding to 870 hPa for the fully interactive runs; (b) as in (a), but for the noninteractive runs; (c) cloud droplet effective radius for the vertical model level approximately corresponding to 870 hPa for the fully interactive runs; (d) as in (c), but for the noninteractive run; (e) gridbox cloud liquid water path for the fully interactive runs; and (f) as in (e), but for the noninteractive runs.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

While the globally averaged LWP shows essentially no difference between 2005 and 1975 values (Table 1), Fig. 7e indicates large but opposing trends in different regions. The Europe and Black Triangle areas have particularly large LWP changes of −6% and −11%, respectively. Such reductions (regardless of the changes in CDNC) will tend to weaken the aerosol indirect effect because of the sensitivity of REFF to LWP as seen from Eq. (1). Since the onset of precipitation through collisions and coalescence depends on both cloud droplet size and LWC as seen in Eqs. (2) and (3), the large LWP changes must also be kept in mind when interpreting the precipitation changes in the next subsection.

Cloud cover shows no significant changes in Europe, but for East Asia there was a 90% significant change of −1% (Table 1). The fact that we see large changes in cloud properties such as CDNC, REFF, and LWP but little change in cloud cover has to do with how cloud cover is parameterized in the model. In NorESM, cloud cover is independent of cloud water content and is calculated based on relative humidity, updraft mass fluxes, and tropospheric stability (Neale et al. 2010).

To separate the microphysical effects of the aerosols from the climatic response, we compare results from fully interactive and noninteractive model runs. If the change in a variable is similar between the two model setups, it indicates that the change is dominated by the “pure” aerosol influence. Dissimilar changes, on the other hand, indicate that feedback processes play a major role. The fact that the cloud climatologies are slightly different between the fully interactive and noninteractive runs (section 3d) is likely to have only a minor effect, as we are focusing on the difference between 2005 aerosols and 1975 aerosols. We found a strong agreement between the two simulation setups for CDNC (cf. Figs. 7a and 7b), with a highly significant correlation coefficient of 0.86 between the noninteractive and fully interactive change in CDNC. For REFF (Figs. 7c,d), the agreement was weaker (correlation coefficient of 0.51) because, as seen in Eq. (1), the cloud droplet size is not only determined by CDNC but also by LWP, which in the fully interactive runs is sensitive to the climate response to aerosol forcing. For LWP (Figs. 7e,f), which is more linked to cloud adjustments (Boucher et al. 2013), feedback processes greatly diminish the similarity between the noninteractive and fully interactive fields, giving a correlation coefficient of only 0.34. This is in agreement with the recent insight that cloud adjustments (e.g., Stevens and Feingold 2009) cannot simply be regarded as a “lifetime effect” that always corresponds to increased cloud cover and LWP with increasing aerosol burden. Rather, they represent complex cloud dynamics (unresolved in ESMs) that can lead to either an increase or decrease in LWP.

d. Changes in precipitation

To get a preliminary indication of how sulfate aerosols relate to precipitation in the model, we show total precipitation amounts as a function of column-integrated SO4 in Fig. 8. We separate into different latitude bands as well as LWP regimes to avoid contributions from the effects other than aerosol–precipitation interactions. For instance, at high latitudes we typically find low SO4 values because of the few pollution sources and low precipitation amounts because of the little atmospheric moisture, while in the midlatitudes emissions are higher and the climate is wetter. This would contribute to an aerosol–precipitation correlation but would have nothing to do with the direct or indirect aerosol effects. With the exception of low LWP regimes at high latitudes, there is a consistent sensitivity of precipitation amount to SO4 column burden, in such a way that higher SO4 levels are associated with lower precipitation amounts. Specifically, we note that

  1. for sulfate column burdens up to about 1 μg S m−2, higher sulfate levels are associated with lower precipitation amounts, indicating an influence of SO4 on precipitation;

  2. for sulfate levels above 1 μg S m−2, precipitation no longer changes with increasing concentrations, indicating a saturation effect; and

  3. low LWP regimes are less sensitive to changes in aerosol concentrations than high LWP conditions.

Fig. 8.
Fig. 8.

Total precipitation amount, plotted as a function of sulfate column burden (μg S m−2) and sorted by LWP regimes as well as into three latitude bands: (a) 60° to 90°N, (b) 20° to 60°N, and (c) 20°S to 20°N. Based on values from the fully interactive instantaneous output runs.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

We now investigate whether this link is still apparent when comparing monthly mean precipitation values between 1975 and 2005. In the global average there is virtually no difference in precipitation (Table 1), but changes emerge when we take a closer look at Europe and East Asia, focusing on stratiform and convective precipitation separately.

1) Precipitation change in Europe

In Europe, the change in stratiform precipitation was statistically significant only for the smaller BT region (−5%, significant at the 90% level) (see Fig. 9a and Table 1). The change in convective precipitation (Fig. 9b) was statistically significant both for the entirety of Europe and the smaller BT region and was associated with significantly increasing surface temperatures (see Table 1). This increase was concurrent with a 4.4 W m−2 increase in clear-sky surface solar irradiance for Europe (9.4 W m−2 increase for BT), which is somewhat higher than the observed 2.55 W m−2 (1.7 W m−2 decade−1) trend found by Norris and Wild (2007) for the 1987–2002 period. Increased convective precipitation in the BT area is also consistent with observations [see Rulfová and Kyselý (2014) as well as Stjern et al. (2011)]. There was no significant change in simulated tropospheric stability for either Europe or the BT.

Fig. 9.
Fig. 9.

Simulated 1975 to 2005 absolute changes (mm day−1) in (a) stratiform precipitation in Europe and (b) convective precipitation in Europe, (c) stratiform precipitation in East Asia, and (d) convective precipitation in East Asia.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

In the model, stratiform precipitation is the precipitation type most sensitive to aerosol-induced influence through the cloud microphysical pathway. Droplet sizes increased substantially over Europe between 1975 and 2005, and for the BT simulated LWP decreased by 11%, which is consistent with a weakened lifetime effect due to the larger cloud droplets more often exceeding the autoconversion threshold [rυc in Eq. (2)]. Even so, stratiform precipitation decreases. This seemingly conflicting precipitation response must be related to the microphysical or dynamical feedbacks operating in the fully interactive runs. To confirm this we look at how stratiform precipitation (PRECS) changed between 1975 and 2005 for the noninteractive runs. In these runs, the meteorology that drives the model forward is not influenced by aerosol changes (see section 3d for details). We find that the “expected” aerosol–precipitation link, in which decreased aerosol levels cause precipitation to increase, is sustained; for the BT we now find a statistically significant increase of 2% in PRECS (cf. the changes in Europe between Figs. 10a and 10b).

Fig. 10.
Fig. 10.

The 2005 minus 1975 difference (red colors indicating increase and blue colors indicating decrease) in stratiform precipitation for (a) the noninteractive runs and (b) the fully interactive runs.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

In addition, it is likely that the strong increase in downwelling shortwave radiation at the surface has led to an increase in the frequency of convective situations. In the model, precipitation is defined as convective when atmospheric conditions are sufficiently unstable and stratiform otherwise, which means that to some extent an increase in convective situations will imply a decrease in the frequency of stratiform situations (the exception is that stratiform precipitation is allowed from the anvil of deep convective clouds). Haerter and Berg (2009) commented on the fact that a shift in rainfall amounts from stratiform to convective regimes can occur in regions of warming because of the dominance of convective precipitation at higher temperatures and the dominance of stratiform precipitation at cooler temperatures. Berg et al. (2013) confirmed for observations in Germany that convective precipitation responds much more sensitively to temperature increases than stratiform precipitation does, and in observations from the Czech Republic, Rulfová and Kyselý (2014) found smaller changes in stratiform than in convective precipitation over the 1982–2010 period. Keeping in mind that LWP is—apart from detrainment from deep convection—mainly a measure of stratiform condensate, this would then contribute to a further reduction in simulated LWP in this region.

2) Precipitation change in East Asia

As accounted for in section 2, reduced precipitation has been observed over the past decades in East Asia (e.g., Zhao et al. 2006). The present model simulations agree with such findings, showing reductions in both stratiform (PRECS; −3%) and convective (PRECC; −4%) precipitation in this region (see Figs. 9c and 9d, respectively). The change in stratiform precipitation may be a sign of precipitation suppression by the increasing aerosol loads, which is consistent with the significant decrease in cloud droplet size over East Asia. Meanwhile, the change in convective precipitation is more likely tied to the direct radiative effects of the scattering and absorbing aerosols, which increased dramatically in numbers between 1975 and 2005. In response to this increase, the downwelling clear-sky surface solar flux decreased by 6.0 W m−2, instigating a decrease in surface temperatures of 0.3 K (1.4%). This reduction in surface heating is associated with a statistically significant change in the average vertical motion (Fig. 11a); for the model layer closest to the surface, the average vertical velocity changed from −0.08 hPa s−1 (ascending motion) in 1975 to +0.03 hPa s−1 (descending motion) in 2005. In addition, because of the significant surface cooling, the average temperature difference between 780 hPa and the surface has decreased by 3% or 0.31 K (780-hPa temperatures increased by 0.01 K) (see Fig. 11b). This signifies an increase in tropospheric stability in the East Asia region, which is in contrast to Bollasina et al. (2011), who found decreased precipitation in aerosol forced model experiments for South Asia but no appreciable trend in stability. However, Zhao et al. (2006), using a more refined definition of atmospheric stability calculated from sounding data, observed a 50% reduction in the “unstable day frequency” over east-central China concurrent with observed decreases in precipitation between 1986 and 2003. The increased lower-tropospheric stability may have contributed to the reduced uplift and ultimately to the decrease in precipitation, a process known as “the surface energy budget effect” (Lohmann and Feichter 2005). This is qualitatively consistent with the small reduction of 1% in cloud cover over the region. In addition, we note that the latent heat flux is reduced by 3.2% (Table 1), probably due to the suppression of downwelling solar radiation at the surface. This reduction is indicative of a weakening of the hydrological cycle, which would be expected to reduce convective precipitation.

Fig. 11.
Fig. 11.

(a) Vertical profile of the vertical velocity for 1975 and 2005 averaged over the East Asia region, and (b) vertical profile of 1975–2005 change in temperature for the same region.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

Figure 12 shows changes in SO4, surface temperature, latent heat flux, and convective precipitation averaged over longitudes 105° to 125°E (the most polluted part of East Asia) and plotted by latitude. The north–south agreement in the change in convective precipitation and the other variables is prominent, supporting the above reasoning concerning the hydrological cycle. The correlations between the (zonally averaged) latitudinal changes in PRECC and the latitudinal changes in surface temperature, SO4, and latent heat flux as seen in the figure are 0.41 (p < 0.05), −0.38 (p < 0.05), and 0.30 (p < 0.1), respectively.

Fig. 12.
Fig. 12.

Simulated changes between 1975 and 2005 for the average of longitudes 105° to 125°E, plotted by latitude, for (a) SO4 burden, (b) surface temperature, (c) latent heat flux, and (d) convective precipitation.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

As stated above, NorESM overestimates AOD over Europe, which will contribute to a positive bias in the direct radiative effect (and hence in the surface temperature change and thereby the change in convective activity) between 1975 and 2005. We have also mentioned that Myhre et al. (2013b) find that NorESM probably overestimates the positive radiative forcing from BC globally averaged and that Kirkevåg et al. (2013) conclude that over Europe, ABS (to which BC is the main contributor) is underestimated. Given the somewhat conflicting nature of these model evaluations, it is therefore not certain how these model biases would influence the simulated change in convective precipitation over Europe. In East Asia, on the other hand, both AOD and ABS are underestimated. This would contribute to a negative bias in the resulting change in convective precipitation, balancing parts of or all of the positive bias from the overestimated BC forcing. In this region, therefore, the change in convective precipitation following the change in aerosol concentrations may in reality be higher than simulated.

e. Precipitation susceptibility

It is of great interest to identify those cloud regimes or regions in which precipitation is most likely to be affected by aerosols. To quantify how well precipitation responds to a given change in aerosol concentrations, we make use of the precipitation susceptibility S0 similar to that defined by Sorooshian et al. (2009) but focusing on stratiform precipitation:
e4
Here, S0 describes the relative change in PRECS per change in CDNC. By focusing on the stratiform precipitation only, we attempt to capture the microphysical aerosol effects on precipitation. Calculations are based on the model setup with year 1975 aerosol emissions, as this is the baseline from which we want to know if the precipitation is susceptible to the transition to 2005 emissions. A global map of S0 in Fig. 13 shows high susceptibilities over the remote ocean areas of the tropics and subtropics, in line with the dependency of indirect aerosol effects on background aerosol concentrations (Hoose et al. 2009). While global mean S0 is 0.37, the average susceptibility between 20°S and 20°N is 0.60. The global mean correlation coefficient between S0 and the magnitude (the absolute value) of the 2005–1975 difference in stratiform precipitation is 0.42 and highly significant, which confirms that there is a tendency for more susceptible areas to have larger precipitation responses to the change from 1975 to 2005 aerosol emissions, at least in regions where the 1975 to 2005 change in aerosols was notable. In the tropics, for instance, low sulfate changes (Fig. 6a) yield low precipitation changes (not shown) in spite of high susceptibilities. In addition, Fig. 8 above indicated that the potential change in precipitation with aerosol concentration depends also on LWP. If we plot S0 as a function of LWP (Fig. 14a), we can see the three regimes identified by Sorooshian et al. (2009): (i) low LWP corresponding to low S0, as clouds do not precipitate because of the limited available water; (ii) intermediate LWP corresponding to increasing susceptibility of precipitation to cloud droplet number concentration changes; and (iii) high LWP, when clouds are dense enough to precipitate regardless of high droplet numbers, causing a decrease in precipitation susceptibility. A similar LWP dependency was found in a model study by Gettelman et al. (2013), while Terai et al. (2012) found for observations of precipitating and nonprecipitating thin clouds a slight decrease in S0 with LWP. The highest LWP levels in that study, however, were around 300 g m−2, below which our simulated S0 in Fig. 14a changes only very little. It should be noted that since NorESM overestimates precipitation release (and LWP) over the Himalayas (see section 4c), the high S0 in this area as seen in Fig. 13 is most likely overestimated.
Fig. 13.
Fig. 13.

Global map of precipitation susceptibility of PRECS (for values of PRECS above 0.1 mm day−1) to column integral CDNC, based on instantaneous values from fully interactive runs using 1975 aerosol emissions. Units are cm−3 mm day−1.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

Fig. 14.
Fig. 14.

Precipitation susceptibility of PRECS as a function of cloud liquid water path for (a) all precipitation events above 0.1 mm day−1 and (b) as in (a), but as a function of SO4 concentrations and sorted into various LWP bins (LWP units are g m−2). The figure is based on instantaneous values from the shorter fully interactive runs as described in section 2c.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

The simulated values of S0 in the present work are on the low side compared to observations and high-resolution simulations (e.g., Sorooshian et al. 2009; Feingold et al. 2013; Mann et al. 2014). One explanation for this could be that our S0 values are based on data with substantially coarser spatial (horizontal as well as vertical) resolution than the cloud-resolving LES studies by, for example, Sorooshian et al. (2009) and Feingold et al. (2013). McComiskey and Feingold (2012) showed that the temporal or spatial resolution of a dataset had a strong influence on the quantification of aerosol–cloud interactions. They found, for instance, that satellite data with low spatial resolution yielded lower variability in cloud properties than satellite data with higher resolution, while the variability of the aerosol properties stayed the same. Looking at Eq. (4), we see that such a decrease in the variability of the cloud-related properties (here, precipitation) and no change in the variability of the aerosol-related properties (here, CDNC) could lead to a decrease in S0. Another thing that could conceivably lead to underestimations of S0 is the tendency of NorESM to overestimate LWP values, as mentioned in section 4b. This could mean that too often clouds are so thick that fluctuations in aerosol concentrations have little impact (see the LWP regime to the extreme right in Fig. 14a).

The dependency of S0 on aerosol concentration as indicated by the link between SO4 and precipitation in Fig. 8 is explored further in Fig. 14b, where we show S0 as a function of SO4 concentrations for three LWP regimes. In the intermediate LWP regime (100–1000 g m−2, solid curve), there is a general decrease in S0 with increasing SO4, as expected. However, for clouds in low LWP environments (<100 g m−2, dotted curve) S0 is always small, possibly because many of these clouds (e.g., over land) are too thin to precipitate, regardless of changes in aerosol amount. Interestingly, in this regime S0 increases with increasing sulfate concentrations from a weak negative value for low SO4 to a weak positive value for high SO4 levels. The negative values may indicate that in extremely clean environments with weak updrafts (hence low LWP) there is a CCN limitation, as shown by Mauritsen et al. (2011) for the Arctic. This CCN limitation disappears as sulfate concentrations increase. In high LWP regimes (>1000 g m−2, dashed curve), S0 increases up to a certain limit and then decreases. As was demonstrated in Fig. 14a, clouds containing very high amounts of water have reduced susceptibilities; these clouds are typically characterized by strong updrafts, which at low sulfate concentrations render the precipitation mechanisms efficient enough to be largely unaffected by variations in aerosol concentrations. As the sulfate levels increase, the clouds become more susceptible, and for intermediate sulfate levels, the high LWP regime is associated with much larger precipitation susceptibilities than lower LWP regimes, consistent with the dependency of S0 on LWP noted above. Above a certain sulfate level, however, the susceptibility drops off again, conceivably because the aerosol influence eventually reaches a saturation effect even in these dense clouds. Together the results of Fig. 14b indicate that the susceptibility of precipitation to aerosol perturbations may be higher than expected from Fig. 14a for low LWP regimes in areas of high background concentrations and lower than expected for high LWP regimes in areas of low background concentrations.

For the 1975 as well as the 2005 simulation, both Europe and East Asia have average liquid water paths in the vicinity of 200 g m−2, corresponding to the intermediate LWP regime marked by the solid line in Fig. 14b. Meanwhile, the average SO4 level changes between the simulations; for Europe, the sulfate levels decrease between 1975 and 2005, while East Asia experiences an increase. According to the figure, a decrease in sulfate concentrations as seen in Europe should lead to an increase in the precipitation susceptibility as we move from right to left along the solid line in Fig. 14b. Indeed, we find that S0 for Europe is 0.15 in the “polluted” 1975 simulation and 0.24 in the “cleaner” 2005 simulation. Conversely, the increase in sulfate concentrations between the two simulations for East Asia (moving from left to right along the solid line in Fig. 14b) should imply a decreased susceptibility, and as expected we find that S0 decreases from 0.20 in the cleaner 1975 simulation to 0.17 in the polluted 2005 simulation.

f. Liquid accretion versus autoconversion

To explain why precipitation susceptibility has such a high dependency on LWP in NorESM, we finally take a closer look at the two main mechanisms for warm rain precipitation formation in the model, namely, autoconversion [Eq. (2), above] and liquid accretion [Eq. (3), above]. As we see, release of precipitation by autoconversion is proportional to the 7/3 power of and thereby LWC and hence LWP. Release of precipitation by liquid accretion is proportional to and furthermore to qr, which is a measure of the liquid and ice condensates in the column above. Hence, the formulations of both accretion and even more so autoconversion have a strong dependency on LWP.

When a cloud’s liquid water content increases above a certain level (about 1000 g m−2), Fig. 14a shows decreasing susceptibility of precipitation to aerosol changes. This is because in high LWP regimes the abundant availability of water vapor no longer poses a limitation to droplet growth, and accretion processes form precipitation regardless of high droplet numbers. The same feature can be seen in Fig. 15, which shows the ratio of accretion to autoconversion as a function of LWP. In low LWP regimes, the clouds are highly sensitive to changes in aerosol concentrations, and autoconversion dominates over accretion, yielding low ratios. As the cloud’s liquid water content increases, liquid accretion becomes increasingly important (and autoconversion correspondingly less important) for rain production. For the average liquid water paths in Europe and East Asia, autoconversion is by far the more important of the two mechanisms in our model.

Fig. 15.
Fig. 15.

Accretion/autoconversion ratio as a function of cloud liquid water path. The figure is based on fully interactive runs yielding instantaneous values.

Citation: Journal of Climate 28, 22; 10.1175/JCLI-D-14-00837.1

Note that the accretion/autoconversion ratio is always below one. Feingold et al. (2013) found a strong relationship between precipitation susceptibility and cloud contact time (time available for collision coalescence) and noted that if clouds have a short lifetime and/or the air parcels’ in-cloud residence time is brief, autoconversion will contribute most strongly to rain production. If, on the other hand, there is ample time available for collision–coalescence, accretion will dominate. In NorESM rain mass and number mixing ratios are not carried from time step to time step so that rain formation can only depend on the cloud quantities valid for the current time step (20 min). Accretion is then caused only by rain that is created through autoconversion or ice phase processes at each time step. In other words, autoconversion will tend to be overemphasized in diagnostic rain schemes, which means that the sensitivity of precipitation to changes in aerosol burden may be positively biased. This was also pointed out by Gettelman et al. (2013, 2014), who performed similar studies with the NCAR Community Atmosphere Model, version 5.2 (CAM5), with the Morrison–Gettelman microphysics scheme and found autoconversion rates to be highly overestimated compared to Variability of American Monsoon Systems (VAMOS) Ocean–Cloud–Atmosphere–Land Study (VOCALS) aircraft observations. It is a typical limitation of most global circulation models that potentially induces an overestimation of the aerosol-induced cloud lifetime effects (e.g., Posselt and Lohmann 2009; Quaas et al. 2009; Gettelman et al. 2013; Feingold et al. 2013). An alleviation of this problem is obtained by treating precipitating water as a prognostic variable, which leads to a weakening of the simulated aerosol indirect effect (Posselt and Lohmann 2009).

6. Summary and conclusions

We have investigated, through a series of Earth system model simulations, how a change from 1975 to 2005 global aerosol emissions may have impacted clouds and precipitation over Europe and East Asia. We find that the changing aerosol loads have had a statistically significant impact on cloud properties such as cloud droplet number concentration, cloud droplet size, and liquid water path in both regions. The effects on precipitation, however, are smaller and more elusive. A similar finding was presented by Seifert et al. (2012), who investigated the effect of aerosols on clouds and precipitation over Germany using a combination of radar data and a high-resolution numerical model.

For Europe as well as for the smaller BT area, we find significant increases in convective precipitation, but a decrease in stratiform precipitation, in spite of decreased cloud droplet numbers and increased cloud droplet sizes. We hypothesize that this discrepancy is mainly attributable to the effect of an increasing ratio of convective to stratiform precipitation because of the increased surface warming.

Over East Asia the increased concentrations of sulfate and black carbon between 1975 and 2005 lead to a spindown of the hydrological cycle by reducing the solar heating of the surface. A decline in stratiform precipitation may conceivably be attributable to cloud adjustments, which is consistent with increasing cloud droplet number concentrations and reduced droplet sizes. Meanwhile, surface temperatures drop as a result of the enhanced clear-sky radiative forcing by aerosols, and consequently the simulated lower troposphere atmospheric stability increases. This is likely the cause of the significant decrease in convective precipitation in the region. The model’s strong BC forcing compared to other AeroCom models (Myhre et al. 2013b) suggests that this effect may be exaggerated. However, since the largest contribution to the decrease in convective precipitation originates not from the semidirect effect but from reduced solar heating of the surface, it will be as affected by sulfate and other aerosols as by BC. The fact that NorESM has a negative AOD bias over East Asia in the summer as well as annually averaged (compared to AERONET measurements; Kirkevåg et al. 2013) would tend to underestimate the effect.

The simulated precipitation susceptibility S0 may be biased because of the model’s tendency to overestimate LWP and underestimate CDNC. Indeed, our susceptibility values are lower than what is found through observations and cloud-resolving model studies (e.g., Feingold et al. 2013; Sorooshian et al. 2009). Even so, the variation of S0 with liquid water path agrees remarkably well with the same studies. As expected, S0 is highest over the remote ocean regions with low background aerosol concentrations, and we find a correlation between areas of high precipitation susceptibilities and high 1975–2005 changes in precipitation amounts. For the average liquid water paths of Europe and East Asia, we find that autoconversion is the most important precipitation-forming mechanism, which by itself would imply a sensitivity of precipitation to changes in aerosol concentrations. However, background concentrations of SO4 in both regions are high, rendering precipitation susceptibilities low. This is probably the main reason why the microphysical link between aerosols and warm rain stratiform precipitation is weak in spite of the substantial changes in aerosol concentrations.

In addition, clouds can be seen as a “buffered” system in the sense that significant changes in aerosol concentrations may have little effect on the resulting surface precipitation as the microphysical cloud response can be dampened by a countering (microphysical or dynamical) response (Stevens and Feingold 2009). Indeed, while stratiform precipitation showed no clear response to aerosol perturbations in the fully interactive model runs, we found a significant increase in stratiform precipitation in line with cloud adjustments when running the model in a noninteractive mode, in which feedback processes that conceivably buffer the original signal are greatly reduced.

As pointed out by Seifert et al. (2012), the radiative effects on cloud systems are not buffered in the same sense as the microphysical effects. As a consequence, the direct aerosol effects on precipitation may be more pronounced than the indirect effects on precipitation, in line with the findings of Ruckstuhl et al. (2010), but in contrast to Lohmann and Feichter (2001). This is exactly what we see in this model experiment: The direct aerosol effect seems to be a strong driver of precipitation changes in both Europe and East Asia, but in spite of clear signals in cloud properties the microphysical or indirect aerosol effects on precipitation are confounded by dynamical or radiative feedbacks. This is consistent with observations finding stronger trends in convective than in stratiform precipitation for periods of large emission changes (Rulfová and Kyselý 2014; Stjern et al. 2011; Zhao et al. 2006).

Acknowledgments

The authors are thankful to Jens Debernard for his work on the slab ocean model setup, to Øyvind Seland for useful discussions, and to Helene Muri for repeated help with the model. The simulations were performed on the Hexagon Cluster of the Norwegian metacenter for High Performance Computing (NOTUR). The study was partly funded by the EXPECT project (Norwegian Research Council Grant 229760/E10). We thank the three anonymous reviewers for constructive comments that have led to significant improvements in the paper.

REFERENCES

  • Abdul-Razzak, H., and S. J. Ghan, 2000: A parameterization of aerosol activation: 2. Multiple aerosol types. J. Geophys. Res., 105, 68376844, doi:10.1029/1999JD901161.

    • Search Google Scholar
    • Export Citation
  • Ackerman, A. S., M. P. Kirkpatrick, D. E. Stevens, and O. B. Toon, 2004: The impact of humidity above stratiform clouds on indirect aerosol climate forcing. Nature, 432, 10141017, doi:10.1038/nature03174.

    • Search Google Scholar
    • Export Citation
  • Aitken, J., 1880: On dust, fogs, and clouds. Nature, 23, 195197, doi:10.1038/023195d0.

  • Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245, 12271230, doi:10.1126/science.245.4923.1227.

    • Search Google Scholar
    • Export Citation
  • Alterskjær, K., J. E. Kristjánsson, and Ø. Seland, 2012: Sensitivity to deliberate sea salt seeding of marine clouds—Observations and model simulations. Atmos. Chem. Phys., 12, 27952807, doi:10.5194/acp-12-2795-2012.

    • Search Google Scholar
    • Export Citation
  • Ayers, G., and Z. Levin, 2009: Air pollution and precipitation. Clouds in the Perturbed Climate System: Their Relationship to Energy Balance, Atmospheric Dynamics, and Precipitation, J. Heintzenberg and R. J. Charlson, Eds., MIT Press, 369–399.

  • Bárdossy, A., and H. J. Caspary, 1990: Detection of climate change in Europe by analyzing European atmospheric circulation patterns from 1881 to 1989. Theor. Appl. Climatol., 42, 155167, doi:10.1007/BF00866871.

    • Search Google Scholar
    • Export Citation
  • Bentsen, M., H. Drange, T. Furevik, and T. Zhou, 2004: Simulated variability of the Atlantic meridional overturning circulation. Climate Dyn., 22, 701720, doi:10.1007/s00382-004-0397-x.

    • Search Google Scholar
    • Export Citation
  • Bentsen, M., and Coauthors, 2013: The Norwegian Earth System Model, NorESM1-M—Part 1: Description and basic evaluation of the physical climate. Geosci. Model Dev., 6, 687720, doi:10.5194/gmd-6-687-2013.

    • Search Google Scholar
    • Export Citation
  • Berg, P., C. Moseley, and J. O. Haerter, 2013: Strong increase in convective precipitation in response to higher temperatures. Nat. Geosci., 6, 181185, doi:10.1038/ngeo1731.

    • Search Google Scholar
    • Export Citation
  • Berg, T., and Coauthors, 2002: EMEP manual for sampling and chemical analysis. EMEP/CCC-Rep. 1/95, Reference O-7726, 303 pp.

  • Bleck, R., C. Rooth, D. Hu, and L. T. Smith, 1992: Salinity-driven thermocline transients in a wind- and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22, 14861505, doi:10.1175/1520-0485(1992)022<1486:SDTTIA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Bollasina, M. A., Y. Ming, and V. Ramaswamy, 2011: Anthropogenic aerosols and the weakening of the South Asian summer monsoon. Science, 334, 502505, doi:10.1126/science.1204994.

    • Search Google Scholar
    • Export Citation
  • Bond, T. C., and Coauthors, 2013: Bounding the role of black carbon in the climate system: A scientific assessment. J. Geophys. Res. Atmos., 113, 53805552, doi:10.1002/jgrd.50171.

    • Search Google Scholar
    • Export Citation
  • 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.

  • Che, H. Z., G. Y. Shi, X. Y. Zhang, R. Arimoto, J. Q. Zhao, L. Xu, B. Wang, and Z. H. Chen, 2005: Analysis of 40 years of solar radiation data from China, 1961–2000. Geophys. Res. Lett., 32, L06803, doi:10.1029/2004GL022322.

    • Search Google Scholar
    • Export Citation
  • Ding, Y., Z. Wang, and Y. Sun, 2008: Inter-decadal variation of the summer precipitation in East China and its association with decreasing Asian summer monsoon. Part I: Observed evidences. Int. J. Climatol., 28, 11391161, doi:10.1002/joc.1615.

    • Search Google Scholar
    • Export Citation
  • Fagerli, H., M. Legrand, S. Preunkert, V. Vestreng, D. Simpson, and M. Cerqueira, 2007: Modeling historical long-term trends of sulfate, ammonium, and elemental carbon over Europe: A comparison with ice core records in the Alps. J. Geophys. Res., 112, D23S13, doi:10.1029/2006JD008044.

    • Search Google Scholar
    • Export Citation
  • Fan, J., L. R. Leung, Z. Li, H. Morrison, H. Chen, Y. Zhou, Y. Qian, and Y. Wang, 2012: Aerosol impacts on clouds and precipitation in eastern China: Results from bin and bulk microphysics. J. Geophys. Res., 117, D00K36, doi:10.1029/2011JD016537.

    • Search Google Scholar
    • Export Citation
  • Feingold, G., A. McComiskey, D. Rosenfeld, and A. Sorooshian, 2013: On the relationship between cloud contact time and precipitation susceptibility to aerosol. J. Geophys. Res. Atmos., 118, 10 54410 554, doi:10.1002/jgrd.50819.

    • Search Google Scholar
    • Export Citation
  • Fekete, B. M., C. J. Vörösmarty, J. O. Roads, and C. J. Willmott, 2004: Uncertainties in precipitation and their impacts on runoff estimates. J. Climate, 17, 294304, doi:10.1175/1520-0442(2004)017<0294:UIPATI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gettelman, A., H. Morrison, C. R. Terai, and R. Wood, 2013: Microphysical process rates and global aerosol–cloud interactions. Atmos. Chem. Phys., 13, 98559867, doi:10.5194/acp-13-9855-2013.

    • Search Google Scholar
    • Export Citation
  • Gettelman, A., H. Morrison, C. R. Terai, and R. Wood, 2014: Corrigendum to “Microphysical process rates and global aerosol-cloud interactions” published in Atmos. Chem. Phys., 13, 9855–9867, 2013. Atmos. Chem. Phys., 14, 90999103, doi:10.5194/acp-14-9099-2014.

    • Search Google Scholar
    • Export Citation
  • Gettelman, A., H. Morrison, S. Santos, P. Bogenschutz, and P. M. Caldwell, 2015: Advanced two-moment bulk microphysics for global models. Part II: Global model solutions and aerosol–cloud interactions. J. Climate, 28, 12881307, doi:10.1175/JCLI-D-14-00103.1.

    • Search Google Scholar
    • Export Citation
  • Gu, Y., K. N. Liou, Y. Xue, C. R. Mechoso, W. Li, and Y. Luo, 2006: Climatic effects of different aerosol types in China simulated by the UCLA general circulation model. J. Geophys. Res., 111, D15201, doi:10.1029/2005JD006312.

    • Search Google Scholar
    • Export Citation
  • Hack, J. J., 1994: Parameterization of moist convection in the National Center for Atmospheric Research Community Climate Model (CCM2). J. Geophys. Res., 99, 55515568, doi:10.1029/93JD03478.

    • Search Google Scholar
    • Export Citation
  • Haerter, J. O., and P. Berg, 2009: Unexpected rise in extreme precipitation caused by a shift in rain type? Nat. Geosci., 2, 372373, doi:10.1038/ngeo523.

    • Search Google Scholar
    • Export Citation
  • 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, 623642, doi:10.1002/joc.3711.

    • Search Google Scholar
    • Export Citation
  • Hoose, C., J. E. Kristjánsson, T. Iversen, A. Kirkevåg, Ø. Seland, and A. Gettelman, 2009: Constraining cloud droplet number concentration in GCMs suppresses the aerosol indirect effect. Geophys. Res. Lett., 36, L12807, doi:10.1029/2009GL038568.

    • Search Google Scholar
    • Export Citation
  • Iversen, T., and Ø. Seland, 2002: A scheme for process-tagged SO4 and BC aerosols in NCAR CCM3: Validation and sensitivity to cloud processes. J. Geophys. Res., 107, 4751, doi:10.1029/2001JD000885.

    • Search Google Scholar
    • Export Citation
  • Iversen, T., and Coauthors, 2013: The Norwegian Earth System Model, NorESM1-M—Part 2: Climate response and scenario projections. Geosci. Model Dev., 6, 389415, doi:10.5194/gmd-6-389-2013.

    • Search Google Scholar
    • Export Citation
  • Jiang, J. H., and Coauthors, 2012: Evaluation of cloud and water vapor simulations in CMIP5 climate models using NASA “A-Train” satellite observations. J. Geophys. Res., 117, D14105, doi:10.1029/2011JD017237.

    • Search Google Scholar
    • Export Citation
  • Kanakidou, M., and Coauthors, 2005: Organic aerosol and global climate modelling: A review. Atmos. Chem. Phys., 5, 10531123, doi:10.5194/acp-5-1053-2005.

    • Search Google Scholar
    • Export Citation
  • Kessler, E., 1969: On the Distribution and Continuity of Water Substance in Atmospheric Circulation. Meteor. Monogr., No. 32, Amer. Meteor. Soc., 84 pp.

  • Khain, A. P., 2009: Notes on state-of-the-art investigations of aerosol effects on precipitation: A critical review. Environ. Res. Lett., 4, 015004, doi:10.1088/1748-9326/4/1/015004.

    • Search Google Scholar
    • Export Citation
  • Kirkevåg, A., and Coauthors, 2013: Aerosol–climate interactions in the Norwegian Earth System Model—NorESM1-M. Geosci. Model Dev., 6, 207244, doi:10.5194/gmd-6-207-2013.

    • Search Google Scholar
    • Export Citation
  • Kristjánsson, J. E., 2002: Studies of the aerosol indirect effect from sulfate and black carbon aerosols. J. Geophys. Res., 107 (D15), doi:10.1029/2001JD000887.

    • Search Google Scholar
    • Export Citation
  • Krüger, O., and H. Graßl, 2002: The indirect aerosol effect over Europe. Geophys. Res. Lett., 29, 1925, doi:10.1029/2001GL014081.

  • Lambe, A. T., and Coauthors, 2011: Laboratory studies of the chemical composition and cloud condensation nuclei (CCN) activity of secondary organic aerosol (SOA) and oxidized primary organic aerosol (OPOA). Atmos. Chem. Phys., 11, 89138928, doi:10.5194/acp-11-8913-2011.

    • Search Google Scholar
    • Export Citation
  • Lanz, V. A., and Coauthors, 2010: Characterization of aerosol chemical composition with aerosol mass spectrometry in central Europe: An overview. Atmos. Chem. Phys., 10, 10 45310 471, doi:10.5194/acp-10-10453-2010.

    • Search Google Scholar
    • Export Citation
  • Larssen, T., and Coauthors, 2006: Acid rain in China. Environ. Sci. Technol., 40, 418425, doi:10.1021/es0626133.

  • Lebsock, M. D., G. L. Stephens, and C. Kummerow, 2008: Multisensor satellite observations of aerosol effects on warm clouds. J. Geophys. Res., 113, D15205, doi:10.1029/2008JD009876.

    • Search Google Scholar
    • Export Citation
  • Lohmann, U., and J. Feichter, 2001: Can the direct and semi-direct aerosol effect compete with the indirect effect on a global scale? Geophys. Res. Lett., 28, 159161, doi:10.1029/2000GL012051.

    • Search Google Scholar
    • Export Citation
  • Lohmann, U., and J. Feichter, 2005: Global indirect aerosol effects: A review. Atmos. Chem. Phys., 5, 715737, doi:10.5194/acp-5-715-2005.

    • Search Google Scholar
    • Export Citation
  • Lövblad, G., L. Tarrasón, K. Tørseth, and S. Dutchak, 2004: EMEP Assessment Part I—European Perspective. EMEP Assessment Rep. 2004, 177 pp.

  • Makkonen, R., Ø. Seland, A. Kirkevåg, T. Iversen, and J. E. Kristjánsson, 2014: Evaluation of aerosol number concentrations in NorESM with improved nucleation parameterization. Atmos. Chem. Phys., 14, 51275152, doi:10.5194/acp-14-5127-2014.

    • Search Google Scholar
    • Export Citation
  • Mann, J. A. L., J. C. Chiu, R. J. Hogan, E. J. O’Connor, T. S. L’Ecuyer, T. H. M. Stein, and A. Jefferson, 2014: Aerosol impacts on drizzle properties in warm clouds from ARM Mobile Facility maritime and continental deployments. J. Geophys. Res. Atmos., 119, 41364148, doi:10.1002/2013JD021339.

    • Search Google Scholar
    • Export Citation
  • Mauritsen, T., and Coauthors, 2011: An Arctic CCN-limited cloud-aerosol regime. Atmos. Chem. Phys., 11, 165173, doi:10.5194/acp-11-165-2011.

    • Search Google Scholar
    • Export Citation
  • McComiskey, A., and G. Feingold, 2012: The scale problem in quantifying aerosol indirect effects. Atmos. Chem. Phys., 12, 10311049, doi:10.5194/acp-12-1031-2012.

    • Search Google Scholar
    • Export Citation
  • Menon, S., J. Hansen, L. Nazarenko, and Y. Luo, 2002: Climate effects of black carbon aerosols in China and India. Science, 297, 22502253, doi:10.1126/science.1075159.

    • Search Google Scholar
    • Export Citation
  • Morrison, H., and A. Gettelman, 2008: A new two-moment bulk stratiform cloud microphysics scheme in the Community Atmosphere Model, version 3 (CAM3). Part I: Description and numerical tests. J. Climate, 21, 36423659, doi:10.1175/2008JCLI2105.1.

    • Search Google Scholar
    • Export Citation
  • Myhre, G., and Coauthors, 2013a: Anthropogenic and natural radiative forcing. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 659–740.

  • Myhre, G., and Coauthors, 2013b: Radiative forcing of the direct aerosol effect from AeroCom Phase II simulations. Atmos. Chem. Phys., 13, 18531877, doi:10.5194/acp-13-1853-2013.

    • Search Google Scholar
    • Export Citation
  • Neale, R. B., and Coauthors, 2010: Description of the NCAR Community Atmosphere Model (CAM 4.0). National Center for Atmospheric Research NCAR/TN-485+STR Tech. Rep., 212 pp.

  • Norris, J. R., and M. Wild, 2007: Trends in aerosol radiative effects over Europe inferred from observed cloud cover, solar “dimming,” and solar “brightening.” J. Geophys. Res., 112, D08214, doi:10.1029/2006JD007794.

    • Search Google Scholar
    • Export Citation
  • Ohara, T., H. Akimoto, J. Kurokawa, N. Horii, K. Yamaji, X. Yan, and T. Hayasaka, 2007: An Asian emission inventory of anthropogenic emission sources for the period 1980–2020. Atmos. Chem. Phys., 7, 44194444, doi:10.5194/acp-7-4419-2007.

    • Search Google Scholar
    • Export Citation
  • Petters, M. D., and S. M. Kreidenweis, 2007: A single parameter representation of hygroscopic growth and cloud condensation nucleus activity. Atmos. Chem. Phys., 7, 19611971, doi:10.5194/acp-7-1961-2007.

    • Search Google Scholar
    • Export Citation
  • Posselt, R., and U. Lohmann, 2009: Sensitivity of the total anthropogenic aerosol effect to the treatment of rain in a global climate model. Geophys. Res. Lett., 36, L02805, doi:10.1029/2008GL035796.

    • Search Google Scholar
    • Export Citation
  • Quaas, J., O. Boucher, and U. Lohmann, 2006: Constraining the total aerosol indirect effect in the LMDZ and ECHAM4 GCMs using MODIS satellite data. Atmos. Chem. Phys., 6, 947955, doi:10.5194/acp-6-947-2006.

    • Search Google Scholar
    • Export Citation
  • Quaas, J., and Coauthors, 2009: Aerosol indirect effects—General circulation model intercomparison and evaluation with satellite data. Atmos. Chem. Phys., 9, 86978717, doi:10.5194/acp-9-8697-2009.

    • Search Google Scholar
    • Export Citation
  • Rasch, P. J., and J. E. Kristjánsson, 1998: A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations. J. Climate, 11, 15871614, doi:10.1175/1520-0442(1998)011<1587:ACOTCM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Rasch, P. J., M. C. Barth, J. T. Kiehl, S. E. Schwartz, and C. M. Benkovitz, 2000: A description of the global sulfur cycle and its controlling processes in the National Center for Atmospheric Research Community Climate Model, version 3. J. Geophys. Res., 105, 1367–1385, doi:10.1029/1999JD900777.

    • Search Google Scholar
    • Export Citation
  • Rotstayn, L. D., and Y. Liu, 2009: Cloud droplet spectral dispersion and the indirect aerosol effect: Comparison of two treatments in a GCM. Geophys. Res. Lett., 36, L10801, doi:10.1029/2009GL038216.

    • Search Google Scholar
    • Export Citation
  • Ruckstuhl, C., J. R. Norris, and R. Philipona, 2010: Is there evidence for an aerosol indirect effect during the recent aerosol optical depth decline in Europe? J. Geophys. Res., 115, D04204, doi:10.1029/2009JD012867.

    • Search Google Scholar
    • Export Citation
  • Rulfová, Z., and J. Kyselý, 2014: Trends of convective and stratiform precipitation in the Czech Republic, 1982–2010. Adv. Meteor., 2014, 647938, doi:10.1155/2014/647938.

    • Search Google Scholar
    • Export Citation
  • Seifert, A., C. Köhler, and K. D. Beheng, 2012: Aerosol-cloud-precipitation effects over Germany as simulated by a convective-scale numerical weather prediction model. Atmos. Chem. Phys., 12, 709725, doi:10.5194/acp-12-709-2012.

    • Search Google Scholar
    • Export Citation
  • Smith, S. J., J. van Aardenne, Z. Klimont, R. J. Andres, A. Volke, and S. D. Arias, 2011: Anthropogenic sulfur dioxide emissions: 1850–2005. Atmos. Chem. Phys., 11, 11011116, doi:10.5194/acp-11-1101-2011.

    • Search Google Scholar
    • Export Citation
  • Sorooshian, A., G. Feingold, M. D. Lebsock, H. Jiang, and G. L. Stephens, 2009: On the precipitation susceptibility of clouds to aerosol perturbations. Geophys. Res. Lett., 36, L13803, doi:10.1029/2009GL038993.

    • Search Google Scholar
    • Export Citation
  • Stevens, B., and G. Feingold, 2009: Untangling aerosol effects on clouds and precipitation in a buffered system. Nature, 461, 607613, doi:10.1038/nature08281.

    • Search Google Scholar
    • Export Citation
  • 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, doi:10.1029/2010JD014603.

    • Search Google Scholar
    • Export Citation
  • Storelvmo, T., J. E. Kristjánsson, S. J. Ghan, A. Kirkevåg, Ø. Seland, and T. Iversen, 2006: Predicting cloud droplet number concentration in Community Atmosphere Model (CAM)-Oslo. J. Geophys. Res., 111, D24208, doi:10.1029/2005JD006300.

    • Search Google Scholar
    • Export Citation
  • Terai, C. R., R. Wood, D. C. Leon, and P. Zuidema, 2012: Does precipitation susceptibility vary with increasing cloud thickness in marine stratocumulus? Atmos. Chem. Phys., 12, 45674583, doi:10.5194/acp-12-4567-2012.

    • Search Google Scholar
    • Export Citation
  • Tørseth, K., and Coauthors, 2012: Introduction to the European Monitoring and Evaluation Programme (EMEP) and observed atmospheric composition change during 1972–2009. Atmos. Chem. Phys., 12, 54475481, doi:10.5194/acp-12-5447-2012.

    • Search Google Scholar
    • Export Citation
  • Twomey, S., 1977: The influence of pollution on the shortwave albedo of clouds. J. Atmos. Sci., 34, 11491152, doi:10.1175/1520-0469(1977)034<1149:TIOPOT>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • van Vuuren, D., and Coauthors, 2011: The representative concentration pathways: An overview. Climatic Change, 109, 531, doi:10.1007/s10584-011-0148-z.

    • Search Google Scholar
    • Export Citation
  • Wood, R., 2012: Stratocumulus clouds. Mon. Wea. Rev., 140, 23732423, doi:10.1175/MWR-D-11-00121.1.

  • Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model. Atmos.–Ocean, 33, 407446, doi:10.1080/07055900.1995.9649539.

    • Search Google Scholar
    • Export Citation
  • Zhang, L., H. Liao, and J. Li, 2010: Impacts of Asian summer monsoon on seasonal and interannual variations of aerosols over eastern China. J. Geophys. Res., 115, D00K05, doi:10.1029/2009JD012299.

    • Search Google Scholar
    • Export Citation
  • Zhang, X. Y., Y. Q. Wang, T. Niu, X. C. Zhang, S. L. Gong, Y. M. Zhang, and J. Y. Sun, 2012: Atmospheric aerosol compositions in China: Spatial/temporal variability, chemical signature, regional haze distribution and comparisons with global aerosols. Atmos. Chem. Phys., 12, 779799, doi:10.5194/acp-12-779-2012.

    • Search Google Scholar
    • Export Citation
  • Zhao, C., X. Tie, and Y. Lin, 2006: A possible positive feedback of reduction of precipitation and increase in aerosols over eastern central China. Geophys. Res. Lett., 33, L11814, doi:10.1029/2006GL025959.

    • Search Google Scholar
    • Export Citation
  • Zhuang, B., Q. Liu, T. Wang, C. Yin, S. Li, M. Xie, F. Jiang, and H. Mao, 2013: Investigation on semi-direct and indirect climate effects of fossil fuel black carbon aerosol over China. Theor. Appl. Climatol., 114, 651–672, doi:10.1007/s00704-013-0862-8.

    • Search Google Scholar
    • Export Citation
Save
  • Abdul-Razzak, H., and S. J. Ghan, 2000: A parameterization of aerosol activation: 2. Multiple aerosol types. J. Geophys. Res., 105, 68376844, doi:10.1029/1999JD901161.

    • Search Google Scholar
    • Export Citation
  • Ackerman, A. S., M. P. Kirkpatrick, D. E. Stevens, and O. B. Toon, 2004: The impact of humidity above stratiform clouds on indirect aerosol climate forcing. Nature, 432, 10141017, doi:10.1038/nature03174.

    • Search Google Scholar
    • Export Citation
  • Aitken, J., 1880: On dust, fogs, and clouds. Nature, 23, 195197, doi:10.1038/023195d0.

  • Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional cloudiness. Science, 245, 12271230, doi:10.1126/science.245.4923.1227.

    • Search Google Scholar
    • Export Citation
  • Alterskjær, K., J. E. Kristjánsson, and Ø. Seland, 2012: Sensitivity to deliberate sea salt seeding of marine clouds—Observations and model simulations. Atmos. Chem. Phys., 12, 27952807, doi:10.5194/acp-12-2795-2012.

    • Search Google Scholar
    • Export Citation
  • Ayers, G., and Z. Levin, 2009: Air pollution and precipitation. Clouds in the Perturbed Climate System: Their Relationship to Energy Balance, Atmospheric Dynamics, and Precipitation, J. Heintzenberg and R. J. Charlson, Eds., MIT Press, 369–399.

  • Bárdossy, A., and H. J. Caspary, 1990: Detection of climate change in Europe by analyzing European atmospheric circulation patterns from 1881 to 1989. Theor. Appl. Climatol., 42, 155167, doi:10.1007/BF00866871.

    • Search Google Scholar
    • Export Citation
  • Bentsen, M., H. Drange, T. Furevik, and T. Zhou, 2004: Simulated variability of the Atlantic meridional overturning circulation. Climate Dyn., 22, 701720, doi:10.1007/s00382-004-0397-x.

    • Search Google Scholar
    • Export Citation
  • Bentsen, M., and Coauthors, 2013: The Norwegian Earth System Model, NorESM1-M—Part 1: Description and basic evaluation of the physical climate. Geosci. Model Dev., 6, 687720, doi:10.5194/gmd-6-687-2013.

    • Search Google Scholar
    • Export Citation
  • Berg, P., C. Moseley, and J. O. Haerter, 2013: Strong increase in convective precipitation in response to higher temperatures. Nat. Geosci., 6, 181185, doi:10.1038/ngeo1731.

    • Search Google Scholar
    • Export Citation
  • Berg, T., and Coauthors, 2002: EMEP manual for sampling and chemical analysis. EMEP/CCC-Rep. 1/95, Reference O-7726, 303 pp.

  • Bleck, R., C. Rooth</