Journal of Geophysical Research: Atmospheres

A model analysis of the interactions between East Asian anthropogenic aerosols and North Pacific atmospheric transients in boreal winter

Authors


Corresponding author: R. Zhou, School of Earth and Space Sciences, University of Science and Technology of China, No. 96 Jinzhai Rd., Hefei, Anhui 230026, China. (zrj@ustc.edu.cn), (yi.deng@eas.gatech.edu)

Abstract

[1] Perpetual winter simulations are conducted with the Community Earth System Model under conditions of normal and zero emission of East Asian anthropogenic aerosols. The inclusion of aerosol emission and subsequently the increase of aerosol concentrations over the North Pacific due to the downstream transport induce statistically robust changes to the structure of the time-mean atmospheric circulation across the basin. Specifically, the activity of atmospheric transient eddies significantly weakens in a zone extending from the east of Japan to the Bering Sea, Alaska, and slightly strengthens over the central North Pacific. Further partitioning in the frequency domain reveals that low-frequency eddies with a 10–30 days time scale dictates the overall transient eddy response to aerosols. The amplitude of synoptic-scale, high-frequency eddies (2–6 days), on the other hand, increases from the central North Pacific to the Gulf of Alaska and decreases at the entrance of the North Pacific storm track and near the west coast of North America. The changes in the synoptic eddy field leave a distinct and consistent signal in surface precipitation. An analysis of the local energy budget of transient eddies indicates that changes in baroclinic conversion, more specifically transient eddy heat flux, largely determines the simulated differences in the activity of atmospheric transients between conditions of normal and zero emission of East Asian anthropogenic aerosols.

1 Introduction

[2] Extratropical atmosphere in winter is characterized by activity of transient eddies that operate on time scales ranging from a few days to several weeks [Blackmon, 1976; Blackmon et al., 1977]. These so-called atmospheric transients are responsible for a significant portion of the meridional transport of heat, moisture, and momentum in midlatitudes, and play a critical role in maintaining the winter-mean circulation [Peixoto and Oort, 1992]. For example, studies have shown that the transient eddy forcing has significant impacts on the strength of the zonal mean flow and the structure of stationary waves [e.g., Hoskins and Pearce, 1988; Hurrell, 1995]. In addition, the forcing of synoptic-scale transients is an important factor during the life cycle of atmospheric blocking [Haines and Marshall, 1987; Shutts, 1983]. The North Pacific region in boreal winter is occupied by intense activity of synoptic-scale transients and known as one of the major “storm tracks” in northern extratropics due to the direct connection between synoptic-sale transients and surface cyclones [Blackmon, 1976; Blackmon et al., 1977]. Due to the scale-interaction between synoptic-scale, high-frequency transients and low-frequency eddies, and the interaction between transients and time-mean flow, winter circulation over the North Pacific exhibits variability across a broad time scale [e.g., Branstator, 1995; Chang et al., 2002; Lau, 1988].

[3] What makes the North Pacific basin even more interesting is that it is under the direct influence of significant amounts of atmospheric aerosols of East Asian origin [Husar et al., 2001; Liu et al., 2003]. Aerosols affect the energy budget of the Earth-atmosphere system through absorbing, reflecting, and scattering radiation. Meanwhile, aerosol particles can serve as cloud condensation nuclei or ice nuclei, and modify cloud microphysical and optical properties and subsequently precipitation efficiency, which are also known as the aerosol indirect effect on the weather and climate [Ramanathan et al., 2001]. Atmospheric aerosols have a large number of natural and anthropogenic sources. In the Northern Hemisphere, East Asia is an important source region of dust aerosols and anthropogenic aerosols, including nitrogen oxides, sulfur dioxide, and other pollutants (aerosol precursor gases) [Lamarque et al., 2010; Streets et al., 2001; Zhang and Shi, 2000]. Taking into account the economic development, energy consumption, and population growth in the present time and near future, East Asia will continue to be a main emission region of anthropogenic aerosols. These aerosol emissions have not only affected local air quality, but also had the potential to alter atmospheric circulation over the North Pacific due to the transport of aerosols toward the downstream region [Bao et al., 2009; Ramanathan and Carmichael, 2008; Zhang et al., 2005]. Zhang et al. [2011] investigated the direct effect of aerosols on regional climate using the Community Climate System Model, and found that emissions of anthropogenic aerosols in China and India lead to the deepening of the Aleutian Low in winter. Combining satellite observations with mesoscale model simulations, Zhang et al. [2007] demonstrated that an increasing trend of deep convective clouds over the North Pacific is likely associated with the indirect effect of East Asian anthropogenic aerosols.

[4] At present, there are still a limited number of studies that explore the effects of East Asian anthropogenic aerosols on regional atmospheric circulations in the context of a global climate model that incorporates both direct and indirect effects of aerosols. The goal of our study is to provide a preliminary assessment of the effect of East Asian anthropogenic aerosols on the atmospheric circulation over the North Pacific in boreal winter, with a special emphasis on transient eddies of both high- and low-frequency. Specifically, we adopt the Community Earth System Model Version 1.0.2 (CESM1.0.2) to detect aerosol impacts and conduct a local energy budget analysis of atmospheric transients to understand the mechanism through which aerosols exert their influences on transient eddies.

2 Model and Experiment Design

[5] The CESM1.0.2 is a coupled global climate model, composed of a central coupler and four separate models simultaneously simulating the Earth's atmosphere, ocean, land surface, and sea-ice. The atmosphere module mainly uses CAM5 (Version 5.0 of the Community Atmosphere Model). Compared with CAM4, CAM5's handling of the physical processes has substantially improved: a new moist turbulence scheme explicitly simulates stratus-radiation-turbulence interactions, making it possible to capture full aerosol indirect effects within stratus; stratiform microphysical processes also highlight the impact of aerosol activation on cloud droplets and ice crystals; the three-mode modal aerosol scheme (MAM3) has been implemented with a full inventory of observation-based aerosol emission data supplied. These major improvements allow the CESM to simulate the influence of aerosols on cloud properties and provide a physics-based estimate of the impact of anthropogenic aerosols on the radiative forcing of climate by clouds. More information about CAM5 can be found in the CAM5 scientific description (NCAR Tech. Note TN-486) through http://www.cesm.ucar.edu/models/cesm1.0/cam/. Although the CAM5 model adopted in our study contains relatively simple cloud microphysics compared to a high-resolution, cloud resolving regional model such as WRF, and there is no explicit treatment of aerosol activation in convective clouds and no separate treatment of convective transport and scavenging of aerosols, the model includes key ingredients of direct and indirect aerosol forcing. Most importantly, CAM5 is a global model that satisfies our needs to handle scale-interaction among atmospheric transients and therefore the excitation of internal modes of variability that are limited in a regional model due to the implementation of lateral boundary conditions. As a global model, CAM5 is able to capture qualitatively the geographical and temporal variations of aerosol mass and number concentrations, size distributions, and aerosol optical properties. The aerosol module of CAM5 accounts for all of the important processes that influence the aerosol: nucleation, coagulation, condensational growth, gas- and aqueous-phase chemistry, emission, dry deposition and gravitational settling, water uptake, in-cloud and below-cloud scavenging, and production from evaporated cloud and rain droplets. Hence, CAM5 contains the essential physics to represent aerosol direct, indirect, and semidirect effect. More details about the capability of CAM5 in representing aerosol-related processes can be found in Liu et al. [2012] and Ghan et al. [2012].

[6] Our experiments adopt the MAM3 scheme for aerosol treatment in CAM5, where aerosols are divided into three modes: the Aitken mode, the accumulation mode, and the coarse mode. MAM3 covers the following aerosol processes: emissions, chemistry, nucleation, condensation, coagulation, transport, activation, and deposition of aerosols and precursor gases. Aerosol emission data set includes data of natural source aerosol emissions, anthropogenic aerosol emissions, and precursor gas emissions. To exclude the effects of seasonal cycle and major modes of interannual variability such as ENSO, we conduct the so-called “perpetual winter” runs where relevant model parameters and forcing fields are fixed for constant winter values. In these experiments, the land surface module uses the Community Land Model. The ocean module uses the data ocean component and sea surface temperature (SST) data are read from the input data set, which is the climatological January SST constructed from the Hadley SST data set [Rayner et al., 2003]. The sea-ice module uses an extension of the Los Alamos National Laboratory sea-ice model, and the process of land ice is not taken into account. The spatial resolution of the experiments is 1.9 by 2.5° on 30 isobaric levels.

[7] Results of two experiments are reported in this paper: (1) the control experiment (CTRL), where the emission of anthropogenic aerosols in East Asia (defined as the region 95°E–145°E, 10°N–55°N) is kept at the normal level, and (2) the zero-emission experiment (NOEMIS), where various types of emissions (such as waste treatment emissions, transportation emissions, industry emissions, energy emissions, domestic emissions, agricultural waste burning emissions) of BC, OC, SO2, and SO4 are set to zero in East Asia. In CTRL, data of anthropogenic aerosol emission are adopted from the Intergovernmental Panel on Climate Change (IPCC) AR5 emission data set [Lamarque et al., 2010] and January 2000 values are used. In both CTRL and NOEMIS, volcanic elevated emissions, grass fire and forest fire emissions are regarded as natural sources and are kept the same at the respective normal levels. In both experiments, models are integrated for 11,000 days and outputs starting from Day 300 are analyzed, giving a sample size of 10,200 days (~340 months). We want to emphasize here that our experiments are “idealized” in nature. The purpose of these experiments is not to obtain the best simulation of aerosols in comparison to observation; instead, the goal is to detect the robust dynamical footprints left by aerosols in the North Pacific atmospheric transients under the simplest possible model setting. This is also precisely the reason why we conduct perpetual winter experiments instead of incorporating the full seasonal cycle into the model.

[8] Figure 1 compares the winter-mean 550 nm aerosol optical depth (AOD) in Moderate Resolution Imaging Spectroradiometer (MODIS) and in our simulations. The spatial distribution of AOD is generally consistent between MODIS observation (Figure 1a) and the CTRL simulation (Figure 1b), although the overall magnitude is smaller in the model. Several reasons are behind this model-observation difference: (1) MODIS AOD shown in Figure 1a is the January climatology over an 11 year period (2001–2011), while the simulated AOD comes from a perpetual winter run (CTRL) that adopts the January 2000 anthropogenic aerosol emission data from the IPCC AR5 emission data set. (2) Liu et al. [2012] showed that there is a low bias in modeled AOD on a global scale, especially in the developing countries. They attributed this low bias to an underestimation of anthropogenic aerosol emissions in these regions. Comparing CTRL to NOEMS (Figures 1c–1d), we find that AOD of CTRL is significantly larger in East Asia (up to 80% larger in southeastern China). Although only anthropogenic aerosol emissions in the dashed box of Figure 1d is suppressed, the change in AOD from CTRL to NOEMS is seen over the North Pacific, the Bay of Bengal, and the Indian Peninsula due to the large-scale transport of aerosols. Specifically, AOD of CTRL is 5% higher than that of NOEMIS considering the entire North Pacific, and is up to 20% higher along the east coast of Asia (Figure 1d).

Figure 1.

550 nm aerosol optical depth in (a) MODIS (January average: 2001–2011), (b) CTRL, (c) NOEMIS, and (d) differences between CTRL and NOEMIS. Dots in Figure 1d indicate the 90% level of statistical significance based on the Student's t-test.

[9] Further examination shows that the difference of the North Pacific AOD between the CTRL and NOEMIS experiment is mainly caused by changes in accumulation mode aerosols, followed by changes of coarse mode aerosols. In the CTRL experiment, concentration of accumulation mode aerosols is higher over the entire North Pacific compared to the NOEMIS experiment. Coarse mode aerosols (largely sea salt aerosols), on the other hand, increase from the central North Pacific to the Gulf of Alaska, and decrease from the east of Japan to the Bering Sea in CTRL. Because sea salt aerosol emission is closely tied to temperature and wind speed near the ocean surface, the removal of East Asian anthropogenic aerosols in our experiments clearly leaves fingerprints in these fields through affecting the large-scale circulation and processes of radiation, cloud and precipitation [Martensson et al., 2003; Monahan et al., 1986]. In addition to AOD, we also evaluated the simulated surface precipitation, wind, and temperature fields using the NOAA CMAP (CPC Merged Analysis of Precipitation) and National Centers for Environmental Prediction (NCEP)/National Center for Atmospheric Research reanalysis data. The model-observation comparison indicates that the model is able to capture the overall spatial distributions of the time-mean precipitation, surface temperature, 300 hPa temperature, and 300 hPa/850 hPa winds (figure not shown).

[10] To demonstrate the existence of the various types of aerosol effects in the model, we present in Figure 2 changes in cloud and radiation fields due to the introduction of East Asian anthropogenic aerosols. Figure 1d shows that the emission of East Asian anthropogenic aerosols leads to an obvious increase of AOD over a large portion of the North Pacific. As shown in Figure 2a, cloud droplet number concentration also increases over the North Pacific demonstrating the presence of the aerosol indirect effect in the model. Cloud water path increases accordingly over the North Pacific, especially over the western North Pacific—a region closer to the emission source of aerosols (Figure 2b). Aerosol can change cloud cover by heating the atmosphere where clouds reside, an effect known as the semidirect effect. Figure 2c shows the change of total cloud cover caused by anthropogenic aerosols. Although the cloud condensation nuclei (CCN) number concentration generally increases across the North Pacific (as demonstrated in Figure 2a), cloud cover change is not spatially homogenous with increase over regions such as the northwestern North Pacific and decrease over places such as Eastern China and a zone extending from the central North Pacific to the Gulf of Alaska. These cloud cover changes, especially the reduction of cloud cover over East China, at least partly reflects the aerosol semidirect effect in the model. Figure 2d shows the change of the convection top pressure, which is a proxy for the intensity of convection. With East Asian anthropogenic aerosols, convective activity strengthens significantly (decrease of convection top pressure) over and to the east of Japan where the North Pacific storm track originates. This result is consistent with those reported by Zhang et al. [2007], namely aerosol forcing tends to enhance convective activity within extratropical cyclones constituting the Pacific storm track. It also verifies that despite the use of cumulus parameterization, CAM5 is capable of capturing at least part of the influence of aerosols on atmospheric convection. Figures 2e and 2f plot respectively the change of the net longwave radiative flux at the top of the atmosphere and the change of the net solar flux at the surface. The decrease of the outgoing longwave radiation over and east of Japan is consistent with the strengthening of convection, increase of cloud top height, and decrease of convection top pressure shown in Figure 2d. The direct impact of aerosols on surface solar flux is characterized by maximum cooling effect in a zone extending from East China to the central Pacific. In summary, the CAM5 model is able to capture some key aspects of aerosol direct, indirect, and semidirect effects, and changes of cloud and radiation fields across East Asia-North Pacific are in line with what is expected from an increase of aerosol emission over East Asia.

Figure 2.

Differences of the (a) vertically-integrated cloud droplet number concentration, (b) total grid-box cloud water path, (c) vertically-integrated total cloud fraction, (d) cloud top pressure, (e) net longwave radiative flux at the top of the atmosphere, and (f) net solar flux at the surface between CTRL and NOEMIS run (i.e., CTRL-NOEMIS). Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test.

3 Results

3.1 Response of the Time-Mean Circulation

[11] Changes in fields relevant to the time-mean circulation from NOEMIS to CTRL are shown in Figure 3 where color shading indicates the differences (CTRL-NOEMIS), contours correspond to the values in the CTRL experiment and dots indicate 90% level of statistical significance. Compared to NOEMIS, geopotential height on the 500 hPa level decreases (increases) near the Sea of Okhotsk and the Bering Sea (in the vicinity of Alaska and northern Canada) in the CTRL experiment, suggesting strengthening of both the East Asian Trough and the North American ridge, thus an amplification of the winter planetary wave over the North Pacific (Figure 3a). The Aleutian low in CTRL deepens correspondingly (figure not shown)—a result consistent with the findings of Zhang et al. [2011]. Accompanying the deepening of the East Asian Trough (Figure 3a), westerly winds strengthen slightly north of the jet core over the western Pacific and weaken near the west coast of North America (Figure 3b). Anomalous northerly winds occur over South Japan and over Alaska and Gulf of Alaska while anomalous southerly winds are found over the Sea of Okhotsk and in the central North Pacific (Figure 3c). Following the thermal wind relation, the location and intensity changes of westerly jet reflect changes in the time-mean, meridional temperature gradient, which determines baroclinic instability of the shear flow [e.g., Simmons and Hoskins, 1978]. We calculate and display in Figure 3d the NOEMIS-to-CTRL changes in Eady growth rate [Eady, 1949; Hoskins and Valdes, 1990], which is proportional to vertical shear of westerlies and serves as an effective measure of local baroclinicity. It is clear that local baroclinicity is the strongest in the western North Pacific, where the entrance of the Pacific storm track is located (contours in Figure 3d). Emission of anthropogenic aerosol in East Asia leads to increasing baroclinicity in the western North Pacific and decreasing baroclinicity over the exit region of the storm track (i.e., from Gulf of Alaska to western North America). The changes in local baroclinicity and their potential projections onto the local storm activity are also partly reflected in the surface precipitation field, where significant precipitation increase is found over the western North Pacific and major reduction occurs along the west coast of North America (Figure 3e). The precipitation changes are found to be mainly driven by changes in large-scale, stratiform precipitation with convective precipitation playing a very minor role. This suggests that activity of high-frequency transients (i.e., extratropical cyclones) over the North Pacific are significantly modified by the introduction of East Asian anthropogenic aerosols.

Figure 3.

Differences of the (a) time-mean 500 hPa geopotential height, (b) 300-hPa zonal wind, (c) 300 hPa meridional wind, (d) 700 hPa Eady growth rate, and (e) surface precipitation rate between CTRL and NOEMIS (CTRL-NOEMIS). Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test.

3.2 Response of the Atmospheric Transients

[12] We quantify the transient eddy activity in models simulations through root-mean-square (RMS) of the filtered, 500 hPa geopotential height (Z′) and sea level pressure (SLP′). Z′ (SLP′) is defined as the departure from monthly-mean to represent the total transients, and as 2–6 days and 10–30 days band-pass-filtered height (SLP) values to represent the high-frequency and low-frequency component of the transients, respectively. The differences in the RMS of Z′ and SLP′ between CTRL and NOEMS are shown respectively in Figures 4 and 5. Figures 4a and 5a indicate that the total activity of transient eddies significantly weakens from the east of Japan to the Bering Sea and near the northwest coast of North America, and strengthens in the central North Pacific in response to the East Asian anthropogenic aerosols. The regions of strengthening and weakening are marked respectively by a red and a purple line that are imposed on all subpanels of Figures 2-10.

Figure 4.

Differences of the RMS of the 500 hPa geopotential height between CTRL and NOEMIS (CTRL-NOEMIS). Z′ is defined as (a) departure from the monthly mean, (b) 2–6 days band-pass-filtered height, and (c) 10–30 days band-pass-filtered height. Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test.

Figure 5.

Same as Figure 4, except for the sea level pressure.

Figure 6.

Differences of the (a) EKE, (b) advection of EKE, (c) conversion from EAPE to EKE, (d) convergence of the ageostrophic geopotential flux, (e) conversion caused by Reynolds stress, and (f) mechanical dissipation in the eddy kinetic energy equation between CTRL and NOEMIS (CTRL-NOEMIS). Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test. Unit is m2/s2/day except for m2/s2 in (a).

Figure 7.

Differences of the (a) EAPE, (b) advection of EAPE, (c) baroclinic conversion, and (d) nonconservative sources and sinks of EAPE in the eddy available potential energy equation between CTRL and NOEMIS (CTRL-NOEMIS). Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test. Unit is m2/s2/day except for m2/s2 in (a).

Figure 8.

Differences of the (a) time-mean eddy meridional heat flux and (b) temperature gradient at the 700 hPa level between CTRL and NOEMIS (CTRL-NOEMIS). Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test. Unit is K · m/s in (a) and m/s2/K in (b).

Figure 9.

Differences of (a) the total eddy energy and (b) the convergence of the total eddy energy flux between CTRL and NOEMIS (CTRL-NOEMIS). Color shading corresponds to CTRL-NOEMIS and contours represent values in the CTRL experiment. Dots indicate the 90% level of statistical significance based on the Student's t-test. Unit is m2/s2 in (a) and m2/s2/day in Figure 9b.

Figure 10.

(a, b, and c) The same as Figures 9a, 7c, and 8a except that total atmospheric transients are considered and obtained using the same NOEMIS time-mean state are used in the evaluation of the local eddy energy budget.

[13] The changes in the activity of high-frequency transients, which constitutes the storm track, are consistent with the changes in the activity of total transients (Figures 4b and 5b). The suppression (enhancement) of high-frequency transients near the west coast of North America (over the central North Pacific) explains the reduction (increase) of surface precipitation in the same region as shown in Figure 3e. It is also important to note here that CESM simulates the North Pacific storm track very well (contours in Figures 4b and 5b), except that the intensity of the model storm track is slightly lower than that in the reanalysis data sets.

[14] The spatial distribution of the response in low-frequency transients bears great similarity with that of the total and high-frequency transients (Figures 4c and 5c). However, the magnitude of the response is nearly 4 times larger in low-frequency transients than that in high-frequency transients. The difference in the total transient activity is thus dictated by the low-frequency component. In fact, activity of low-frequency transients shows pronounced weakening from the east of Japan to the Bering Sea and strengthening across the central North Pacific, although the response near the west coast of North America (particularly in the RMS of Z′) is weaker compared to the changes in high-frequency transients.

3.3 Local Energy Budget of Transient Eddies

[15] To better understand the mechanisms through which the introduction of East Asian anthropogenic aerosols affects the activity of atmospheric transients, we analyze the local energy budget of transient eddies. The calculation of various budget terms follows Orlanski and Katzfey [1991], where the eddy kinetic energy (EKE) equation, eddy available potential energy (EAPE) equation, and total eddy energy equation are written respectively as

display math(1)
display math(2)

and

display math(3)

[16] In equations ((1))–((3)), subscript “m” denotes monthly mean and the primes indicate deviation from the monthly mean. Ke, Ae, and Ee are EKE, EAPE, and the total eddy energy respectively. Vm, Θm, αm are the monthly-mean of wind, potential temperature, and specific volume. inline image are the eddy component of wind, ageostrophic wind, geopotential height, vertical velocity in pressure coordinates, specific volume, potential temperature. inline image is the horizontal area average of potential temperature over the Northern Hemisphere. disse is the frictional sink by eddies. F0 can be interpreted as the time average of the local momentum tendency. S represents the nonconservative sources and sinks of eddy potential energy.

[17] In equation ((1)), inline image. The first two terms on the right-hand side of equation ((1)) are advection of eddy kinetic energy by the time-mean flow and eddies. The third term is the convergence of the ageostrophic geopotential flux. The fourth term − ωα′ represents the conversion from EAPE to EKE. The fifth term is the energy conversion caused by the Reynolds stress (i.e., barotropic conversion between the mean flow kinetic energy and the eddy kinetic energy). The sixth term is a net conversion of EKE to the first-order correlation kinetic energy. The final term is the effect of the stationary forcing of the mean flow into eddies. The sixth and final term would be zero in the time mean sense.

[18] In equation ((2)), inline image. The first two terms on the right-hand side of equation ((2)) are advection of EAPE by the mean flow and eddies. The third term ωα′ represents the loss of EAPE due to conversion into EKE, which is exactly opposite to the fourth term on the right-hand side of equation ((1)). The fourth term corresponds to the traditional baroclinic conversion, i.e., conversion from the mean flow available potential energy to EAPE. The fifth term is the correlation between the perturbation temperature and its time-mean advection by eddies. The time mean of this term is also zero.

[19] In equation ((3)), Ee = Ke + Ae. The first two terms on the right-hand side represent the convergence of total energy flux. The third and fourth term correspond, respectively, to barotropic and baroclinic conversion. The fifth term is mechanical dissipation, which is diagnosed as a residual from equation ((1)). The last term represents the generation of total eddy energy due to diabatic processes, and is diagnosed here as a residual from equation ((2)).

[20] Figure 6 shows the NOEMIS-to-CTRL change in each conversion term on the right-hand side of the EKE equation. The differences in EKE (Figure 6a) is marked by significant decreases from Japan to the Bering Sea as aerosol emission is introduced. This decrease is consistent with the overall reduction of transient eddy activity in this region as seen previously in the RMS field (Figures 4a and 5a). The counterpart increase of transient eddy activity over the central North Pacific, on the other hand, does not show up clearly in the EKE field, which is understandable because EKE is closely tied to the square of the gradient changes in the geopotential height field while RMS measures fluctuations in the total height field. Among all the relevant energy conversion terms shown in Figures 6b–6f, conversion from EAPE to EKE (Figure 6c) through “warm air rising and cold air sinking” appears to be the main contributor to the tendency of decreasing EKE from Japan to the Bering Sea. The slight decrease of EKE seen in the subtropical eastern North Pacific is collectively contributed by weakening of EAPE-to-EKE conversion (Figure 6c) and reduced EKE convergence due to mean-flow and eddy advection (Figure 6b) in that area. A significant positive anomaly in the ageostrophic geopotential flux convergence (Figure 6d) is found over the central subtropical North Pacific, but is largely compensated by a negative anomaly of barotropic conversion (Figure 6e) in the same region. The mechanical dissipation weakens significantly over western North America (Figure 6f), which is consistent with an overall decrease of EKE (thus eddy wind speed) (Figure 6a).

[21] Figure 7 shows the counterpart results for various conversion terms in the EAPE equation. The decrease of EAPE from the east of Japan to Alaska and increase from the central North Pacific to the west coast of North America as seen in the EAPE field (Figure 7a) match nicely with the spatial distribution seen previously in the RMS field (Figures 4a and 5a). This similarity is expected given the connection between EAPE and temperature (thickness) perturbations and the link of the thickness of an atmospheric layer to isobaric geopotential height. Comparing the conversion terms shown in Figures 7c–7d, we find that change in baroclinic conversion (Figure 7c) is the most critical factor leading to the observed EAPE change, particularly the decrease of EAPE from the east of Japan to Alaska. Advection of EAPE has some minor contributions to the reduction of EAPE over northwest Canada (Figure 7b). It is interesting to note that the effect of diabatic generation of EAPE (Figure 7d), which is partly linked to a weak increase of aerosol concentration from NOEMIS to CTRL over Alaska (Figure 1d), leads to a significant increase of EAPE in this region. However, negative anomalies of baroclinic conversion (Figure 7c) and EAPE advection (Figure 7b) largely over-compensate this increase of diabatic generation.

[22] Further partitioning of the baroclinic conversion (Figure 7c) indicates that the meridional component of the term (i.e., λvθ′(∂ Θm/∂ y), where inline image) plays the major role, as is expected given the predominance of meridional temperature gradients in the winter time-mean flow. Figure 8 shows the differences of the time-mean eddyheat flux (vθ′, Figure 8a) and temperature gradient (λ(∂ Θm/∂ y), Figure 8b) at 700 hPa level between the CTRL and NOEMIS experiment. The distribution of the changes of the meridional temperature gradient shown in Figure 8b is consistent with that of the changes in Eady growth rate given in Figure 3d. Both the eddy heat flux and time-mean temperature gradient decrease in a zone extending from the east of Japan to Alaska, and thus contribute to the decrease of baroclinic conversion found in Figure 7c. On the other hand, eddy heat flux is clearly a more important factor among the two given its larger magnitude of decrease in the Bering Sea. The increase of baroclinic conversion (Figure 7c) and subsequently the EAPE (Figure 7a) from the central North Pacific to the west coast of North America also appear to be mainly driven by local increases of eddy heat flux (Figure 8a).

[23] When the total eddy energy (EKE + EAPE) is considered, the key difference due to the inclusion of East Asian anthropogenic aerosols is a marked decrease of the total eddy energy in a southwest-northeast oriented zone extending from the east of Japan toward Alaska (Figure 9a). This change is consistent with the EKE and EAPE changes discussed above. The convergence of the total eddy energy flux is the only term on the right-hand side of equation ((3)) that has not been shown before. The change of this term (Figure 9b) does not project effectively onto that of the total eddy energy (Figure 9a), which further confirms that baroclinic conversion (and the conversion from EAPE to EKE) is the main process responsible for the simulated differences in the North Pacific transient eddy energy between the CTRL and NOEMIS experiment.

[24] The eddy energy budget analysis above focuses on submonthly-scale transients, which are defined as departure from monthly mean. We also conducted similar diagnosis for the total atmospheric transients that are defined as departure from the long-term time-mean. Consideration of total atmospheric transients allows easier separation of the influence of changes in eddy quantity from the influences of changes in background states on the final eddy statistics. For example, Figure 10 shows the differences in some eddy energy budget terms obtained by using the same NOEMIS time-mean state in the calculation of energy conversions while taking transient eddies from CTRL and NOEMIS separately as before. Comparing Figure 10a to Figure 9a, Figure 10b to Figure 7c, and Figure 10c to Figure 8a, we can clearly see that change in baroclinic conversion, which is dictated by change in meridional eddy heat flux, is the main process responsible for the simulated changes in the intensity of the total atmospheric transients from east of Japan to the Bering Sea and Alaska.

4 Conclusion

[25] Emissions of anthropogenic aerosols in East Asia lead to significant changes in the aerosol type and concentration over the North Pacific in boreal winter. The direct and indirect effects of aerosols modify radiation, cloud, and precipitation processes, and eventually cause changes in the properties of atmospheric circulation across the basin due to the interaction between thermodynamical and large-scale dynamical processes. Through idealized perpetual winter simulations made with the National Center for Atmospheric Research CESM, our analysis contrasts basic properties of the North Pacific circulation, especially atmospheric transients, between the scenarios of normal and zero emission of East Asian anthropogenic aerosols.

[26] Statistically robust changes in the transient eddy activity are identified when the emission of anthropogenic aerosol is introduced into the model. The most significant signal is a decrease of transient eddy activity over a southwest-northeast-oriented zone extending from the east of Japan to Alaska. The drop of the transient eddy activity in the equilibrium state is 5–8% with respect to the zero-emission scenario when measured in terms of the total transient eddy energy or the RMS of the filtered 500 hPa height field, and up to 12% when measured in terms of the EAPE. There is a corresponding minor enhancement of transient eddy activity from the central North Pacific to the west coast of North America, which is most evident in the EAPE field. Further partitioning in the frequency domain reveals that the transient eddy response is dominated by changes in low-frequency (10–30 days) eddies. The changes in the activity of synoptic-scale, high-frequency (2–6 days) eddies, in terms of the spatial distributions, are generally consistent with those of the low-frequency eddies except for a distinct drop of synoptic eddy activity near the northwest coast of North America that causes a decrease of precipitation over western North America. Analysis of the local energy budget of transient eddies indicates that the main changes in eddy activity are associated with anomalous baroclinic conversion and subsequently the conversion from EAPE to EKE when eddy kinetic energy budget is considered. Additional calculation shows that changes in eddy meridional heat flux dominate anomalies of baroclinic conversion. The results presented here serve as a preliminary assessment of the projection of East Asian anthropogenic aerosols onto the properties of the equilibrium-state atmospheric circulation across the North Pacific. The sensitivity of the simulated response to the choice of aerosol schemes and the effect of bringing a climatological seasonal cycle are currently being investigated. Future investigations will focus on delineating the dynamical pathway through which aerosol-induced heating anomalies modify eddy quantities such as the meridional heat flux.

Acknowledgments

[27] The MODIS data used in this study were provided by the National Aeronautics and Space Administration, and the CESM was provided by the National Center for Atmospheric Research (NCAR). The USTC author (Zhou) was supported by “the Fundamental Research Funds for the Central Universities” (WK2080000011) and “National Key Basic Research Development Program (973) Project” (2010CB428603). The Georgia Tech author (Deng) was supported by the DOE Office of Science Regional and Global Climate Modeling program under Grant DE-SC0005596 and NASA Energy and Water Cycle Study program under Grant NNX09AJ36G.