Air Quality Modeling in East Asia: Present Issues and Future Directions

Transcription

1 Asia-Pac. J. Atmos. Sci., 50(1), , 2014 DOI: /s REVIEW Air Quality Modeling in East Asia: Present Issues and Future Directions Rokjin J. Park and Sang-Woo Kim School of Earth and Environmental Sciences, Seoul National University, Seoul, Korea (Manuscript received 2 December 2013; accepted 15 January 2014) The Korean Meteorological Society and Springer 2014 Abstract: The rapid economic growth has increased trace gas emissions in East Asia, resulting in various environmental issues, including acid deposition, regional haze, air quality degradation, and climate change, which are critical to the human existence. In particular, air quality degradation became an object of rising concern in East Asian countries. In order to understand sources, transport, and chemical transformation of air pollutants, scientists have widely used atmospheric chemical transport models (CTMs) in East Asia. Here we review our knowledge related to the present air quality issues and their modeling, focusing on O 3 and particulate matter in East Asia. We finally suggest a few recommendations for the next generation of air quality models to improve their capability and use in this region. Key words: Air quality, chemical transport model, ozone, particulate matter, East Asia 1. Introduction Increases in the use of fossil fuels in accordance with the rapid economic growth have resulted in a dramatic rise of trace gas emissions in East Asia (Ohara et al., 2007). As a result, countries in East Asia have confronted growing concerns for environmental consequences including acid deposition, regional haze, air quality degradation, and climate change (Carmichael et al., 2002; Ramanathan et al., 2008). These issues are very critical to the human existence and appear differently depending on the spatial and temporal distributions of atmospheric concentrations of air pollutants, which are either directly emitted from sources or chemically produced from precursors in the atmosphere. Sulfur dioxide (SO 2 ) and reactive nitrogen oxides (NO x NO + NO 2 ) are primary pollutants emitted from the burning of fossil fuels such as coals and gasoline, respectively. Oxidation of SO 2 and NO 2 by OH in the atmosphere produces secondary pollutants of sulfuric acid (H 2 SO 4 ) and nitric acid (HNO 3 ). Scavenging these species by hydrometeors is the major cause of acid deposition (Kim et al., 2011; Morino et al., 2011; Kuribayashi et al., 2012). An increase of atmospheric mass concentrations of these acids by the reaction with ammonia (NH 3 ), the formation of ammonium sulfate and nitrate aerosols, is one of the main reasons of the worsening particulate matter (PM) Corresponding Author: Rokjin J. Park, School of Earth and Environmental Sciences, Seoul National University, 1 Gwanak-ro, Gwanak-gu, Seoul , Korea. rjpark@snu.ac.kr air quality in East Asia (McNaughton et al., 2004). The hygroscopic growth of those inorganic ammonium sulfate and nitrate aerosols with relative humidity leads to degradation of visibility (regional haze) (Li et al., 2013a). Along with PM, ozone (O 3 ) is another key air pollutant that affects air quality for human health and vegetation growth (Nawahda et al., 2012). O 3 is photochemically produced by the oxidation of CO, methane, and non-methane volatile organic compounds (NMVOCs), by OH in the presence of NO x in the troposphere. In order to understand humans contribution to the environmental issues mentioned above, scientists have used numerical tools (called atmospheric chemical transport models, CTMs), which are developed based on their scientific understanding of the processes that determine those issues. The numerical tools are also appropriate for simulating the atmosphere as realistically as possible (Jacob, 1999). In addition, CTMs have played an important role in filling the gap for observations that reflect the status of the atmosphere but are often insufficient to represent the entire whole. Thus, the successful synthesis between the observations and models enables us to understand changes in atmospheric composition, the role of human activity, and the environmental consequences (Bowman, 2013). In this context, CTMs need to evolve continuously to cope with the progress of our knowledge and be tested against observations for better evaluation of their performances, which are sometimes far behind our expectations. Here we review our understanding of the present issues of air quality modeling in East Asia. We also suggest a few recommendations for the next generation of air quality models in East Asia. 2. Observations Observations are crucial to measure the capability of present modeling and to provide guidance for future development. We first discuss the past and present observations and some related issues in East Asia. Over the past few decades, several international field campaigns have been conducted in East Asia for studying atmospheric chemistry, aerosol characteristics, and radiative forcing: for example, the Perturbation by East Asia Continental Air Mass to Pacific Oceanic Troposphere (PEACAMPOT; Hatakeyama et al., 1995), the NASA Pacific Exploratory Mission in the Western Pacific Ocean (PEM-WEST; Hoell et al., 1996), the NASA Global Tropospheric Experiment Trans-

2 106 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES port and Chemical Evolution Over the Pacific (TRACE-P; Jacob et al., 2003), the Asian-Pacific Regional Aerosol Characterization Experiment (ACE-Asia; Huebert et al., 2003), the Intercontinental Transport and Chemical Transformation (ITCT), and the Pacific Exploration of Asian Continental Emission (PEACE; Parrish et al., 2004), the Atmospheric Brown Clouds- East Asian Regional Experiment (ABC-EAREX; Nakajima et al., 2007), the East Asian Studies of Tropospheric Aerosols: an International Regional Experiment (EAST-AIRE; Li et al., 2007), and so on. Measurements obtained during the PEM-WEST A (September-October 1991) and B (February-March 1994) missions provided new insight into the atmospheric chemistry of O 3, its precursors, and the atmospheric sulfur cycle in East Asia (Hoell et al., 1996). TRACE-P and ACE-Asia campaigns, which were exemplary in integrating ground-based, aircraft, and satellite measurements with CTMs, in spring 2001, made it clear that the large-scale natural and anthropogenic air pollutant exported from Asian continent was much more complicated and substantial than expected (Huebert et al., 2003; Jacob et al., 2003). ACE-Asia observations showed that Asian aerosols had a wide range of physical and optical properties, depending on the mixture of mineral dust with pollutants such as black carbon (BC), sulfate, nitrate, and organic aerosols (OA) (Alfaro et al., 2003; Carrico et al., 2003; Huebert et al., 2003; Maxwell-Meier et al., 2004). New particle formation during TRACE-P was found to take place in urban plumes, which consist principally of ammonium sulfate and ammonium nitrate, but not in biomass burning or volcanic plumes (Weber et al., 2003). Validation flights conducted for the Measurements Of Pollution In The Troposphere (MOPITT) satellite instrument revealed little bias (6 ± 2%) in the MOPITT measurements of CO columns. Eisele et al. (2003) showed agreement within 1-2% for O 3, CO, CO 2, CH 4, and photolysis frequencies between the two aircraft measurements and agreement within 10-20% for NO, PAN, and HNO 3. However, the discrepancies between the OH, HO 2, HCHO, and NO 2 measurements aboard the two aircraft represent a significant gap in our ability to test models of HO x and NO x chemistry. Kiley et al. (2003) intercompared results from three regional and four global CTMs applied to the simulation of TRACE-P CO observations. Although all CTMs used the same anthropogenic and biomass burning sources of CO, large differences were found between the simulations, particularly in the free troposphere. This can be attributable to the general difficulty in reproducing convective mass flux particularly near emissions source regions, and the locations and strengths of pollution plumes. In addition, occasionally the meteorological lift associated with excessive model-calculated mass fluxes leads to an overestimation of middle and upper tropospheric mixing ratios. The TRACE-P results indicate remarkable consistency with the Streets et al. (2003) inventory for hydrocarbons, NO x, and SO 2. However, Carmichael et al. (2003) found that CO, BC, and other combustion gases (ethane, acetylene) are underestimated in the boundary layer outflow originating from central regions of China. During the decade since TRACE-P and ACE-ASIA, the achievement in our understanding of natural and anthropogenic air pollutants from East Asia has been tremendous. The PEACE-A (January 2004) and PEACE-B / ITCT 2K2 (April- May 2004) provided coverage of the seasonal progression of O 3 photochemistry and O 3 precursor transport processes, from winter to late spring (Kondo et al., 2004). Liang et al. (2004) simulated global CO transport for the March 2001 through May 2002 time period with the GEOS-Chem CTM and reported that Asian transport events increased CO at Cheeka Peak Observatory by ppbv. Hudman et al. (2004) also used the global GEOS-Chem CTM and argued that PAN decomposition represents a major and possibly dominant component of the ozone enhancement in transpacific plumes carrying Asian emissions. By far the largest effort following TRACE-P and ACE-Asia has been the ABC-EAREX and EAST-AIRE, conducted in spring 2005 at Gosan, Korea and at the Xianghe and Tai Lake sites, China, respectively (Li et al., 2007; Nakajima et al., 2007). The purpose of two campaigns was to understand the chemical, physical, and optical properties of natural and anthropogenic aerosols and their precursor gases, and to estimate direct and indirect effects of these aerosols on radiation, cloud, precipitation, atmospheric circulation and the environment. It is worth noting that a number of other smaller to mediumsized field experiments (e.g., ABC-EAREX2007, Cheju ABC Plume-Asian Monsoon Experiment 2008 (CAPMEX)) and several ongoing long-term continuous measurement activities within the framework of ABC-Asia ( Acid Deposition Monitoring Network in East Asia (EANET; and WMO Global Atmospheric Watch (GAW), have also provided useful observation data for validating CTM simulations and estimating the inter-annual and decadal-scale variations (e.g., Luo et al., 2001; Wang and Shi, 2010). Although surface and intensive multiplatform observations have provided clear indications on the effects of pollution aerosols on atmospheric heating and surface energy budget, the representation of these processes as well as the indirect effects of atmospheric aerosols on clouds is still underdeveloped. In addition to surface network observations, the next step is to develop interdisciplinary intensive experiments to make substantial progress towards answering many of the questions posed in this work. 3. Chemical transport models (CTMs) a. General description CTMs are devised to obtain quantitative information of the concentrations of chemical species in the atmosphere, whose distributions depend on the controlling processes: emission, transport, chemistry, and deposition. This dependence is expressed mathematically by the continuity equation (equation 1)

3 31 January 2014 Rokjin J. Park and Sang-Woo Kim 107 based on the mass conservation of species, which is solved by CTMs (Jacob, 1999). q = emis+ tran+ chem+ dep (1) t where q is the concentration of species of interest, and its change with respect to time is controlled by four processes. In general, this continuity equation is highly nonlinear and can only be solved using the numerical approach. In addition, the dependence of the concentrations of species on the controlling processes should be taken into account simultaneously, because they often occur concurrently. However, the numerical approach is not feasible to solve this dependence simultaneously, in contrast with the time splitting method (Jacobson, 2005). Detailed descriptions of the individual controlling processes and their numerical representation are beyond the scope of this paper. Here, we briefly discuss the general description of the present CTM modeling in this section. Critical issues, however, will be addressed in the following sections below. The controlling processes in equation (1) are dependent on meteorological variables. One obvious example is the transport of species that is driven by horizontal and vertical winds in the atmosphere. Not only the transport but also other processes are functions of meteorological variables, whose information is needed to solve equation (1) in CTMs. Therefore, most CTMs require meteorological data at certain intervals (every 1 or 3- hr) as input to drive them. The meteorological input is generally pre-archived from simulations of independent meteorological models for specific cases or climatological conditions prior to CTMs simulations (Byun and Ching, 1999). One of the widely used CTMs in East Asia is the Community Multi-scale Air Quality (CMAQ) model, which was developed by the US EPA (Byun and Ching, 1999). A number of other CTMs have also been developed and applied for air quality studies in East Asia (Carmichael et al., 2008; and references therein). Those models may differ in terms of methods for representing the processes shown in equation (1), but their fundamental structures based on the 3-D Eulerian framework bear close similarity to the CMAQ. Those 3-D CTMs are coupled with regional or global meteorological models in an offline fashion. For example, the CMAQ can be driven by meteorological outputs from regional meteorological models such as the MM5 model, the Weather Research Forecasting Model (WRF), and others (Otte and Pleim, 2010). We refer to these CTMs as offline CTMs for the convenience. On the other hand, changes of chemical composition in the atmosphere perturb radiation balance and thus affect meteorology. The interaction between meteorology and air pollution is a growing issue for air quality control as well as for weather forecast and even for climate change. To address this issue, the Weather Research Forecasting Model with CHEMistry (WRF- Chem) was developed to incorporate two-way interactions between meteorology and air chemistry (Grell et al., 2005). Recently, the coupling between the WRF and CMAQ models (WRF-CMAQ) has been developed to account for the twoway interactions (Wong et al., 2012). Unlike conventional CTMs, the WRF-Chem calculates meteorological data that are used to simulate the controlling processes in equation (1) at each time step. This online approach in the WRF-Chem has an advantage over offline CTMs; a chemistry simulation is conducted in a consistent way with driving meteorology by incorporating the meteorological feedback. For example, the simulated concentrations of air pollutants from offline CTMs are often inconsistent with meteorological fields in terms of spatial and temporal distributions to the extent that high PM concentrations result in reduced surface insolation that might differ from the meteorological model. However, the online approach possesses a weakness; simulated errors can propagate and grow between meteorology and chemistry computations, since too many uncertainties are involved in complex interactions between the meteorology and chemical simulations. So it is often too difficult to nail down clear causes of simulated errors. Despite this weakness, the online model tends to be widely used for its unique advantage to offer insight into the interaction between the meteorology and the chemistry. b. Emission Burning fuels emit trace gases and aerosols into the atmosphere. Emissions in air quality models are typically estimated for ten major chemical species: SO 2, NO x, CO, CH 4, NMVOCs, submicron BC, submicron organic carbon aerosol (OC), and NH 3 with PM 2.5 (PM whose diameter is less than 2.5 µm) and PM 10 (PM whose diameter is less than 10 µm) mass concentrations. In addition, NMVOCs is further specified into subcategories based on chemical reactivity and functional groups. Quantification of the amount of trace gas emissions has generally been achieved by compiling the amount of fuel use and the burning condition of fuel (Klimont et al., 2002). Streets et al. (2003) first compiled a regional emission inventory of Asia for 2000 in a comprehensive and consistent way, although it possesses considerable uncertainties with some of the emission values because of the lack of national statistics in many countries and of knowledge about the performance of certain kinds of emitters. Along with the numerous modeling studies, many of the top-down approaches reveal high uncertainty, especially with OC and BC emissions that were found to be underestimated especially in China (Carmichael et al., 2003; Park et al., 2005; Jeong et al., 2011). Not only the magnitude of emissions but also their temporal variation is another key factor for the model (Jeong et al., 2011) to reproduce the observation. Recently there have been studies to improve regional emission inventory especially from China (Ohara et al., 2007; Zhang et al., 2009), which considerably have contributed to the improvement of air quality simulations (Kondo et al., 2011; Wada et al., 2012). However, there are still important missing sources of anthropogenic sectors such as ship, aircraft, and other miscellaneous uses of biofuels that need to be further

4 108 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES accounted for the updated emission inventory. In addition, natural emissions from wildfires and vegetation still remain highly uncertain but regarded as very important for air quality degradation in East Asia (Fu et al., 2007; Jeong et al., 2008; Fu et al., 2012; Zhang et al., 2012). Primary PM emissions from dry soil and the deserts in East Asia are responsible for the increases in PM concentrations in spring and are known to have large uncertainty (Lee et al., 2013). The estimates of soil dust emission have been typically parameterized using wind speeds and surface conditions in the model (Kim and Lee, 2013), which will be discussed in more detail below. c. Transport and chemistry Transport of emitted pollutants by winds is a key process to determine the spatial distribution of pollutant concentrations and is generally well taken into account by models (Carmichael et al., 2002), although predictions of accurate wind fields are considerably challenging. The discussion of meteorological simulations is beyond the scope of our paper, but we think that it is worthwhile to emphasize the importance of winds and their transport of pollutants for air quality predictions. Relative to the horizontal transport, a model has considerable uncertainty in the vertical transport of air pollutants from the surface within the PBL and out of the PBL. The former plays a significant role in the dilution of air pollutants with the diurnal variations of the PBL heights. The latter is a crucial factor to determining the long-range transport of air pollutants from East Asia (hemispheric and continental scale transport) and is driven by mid-latitude cyclones, with warm conveyor belts lifting ahead of the associated cold fronts and boundary layer outflow behind the fronts (Park et al., 2005). CTMs include a few hundreds of chemical reactions for chemical species, of which instantaneous concentrations are calculated at each time step. Among the four processes in equation (1) to be calculated in models, more than half of the simulation times are taken by the chemical production and loss calculations of species, which are very expensive. Therefore, air quality simulations often adopt an optimized condensed chemical mechanism for air pollutants of interest with the reduced number of species and chemical reactions to increase the computational efficiency but maintain reasonable accuracy (Gery et al., 1989). As the capability of computers and our scientific understanding have advanced, chemical mechanisms have also evolved with increased species and chemical reactions toward diminishing the existing uncertainties with models, which have primarily reflected ambient conditions in developed countries in Europe and North America. Because of the rapid economical growth, the large population, and diverse terrain, East Asia experiences by far a more complex chemical environment with high anthropogenic and natural emissions of gases and aerosols. Mixing of air masses with different sources could result in diverse chemical characteristics (Jacob et al., 2003). High aerosol concentrations provide surface areas for heterogeneous chemical reactions, which play an important role in the gas-phase O 3 chemistry in polluted urban regions (Song et al., 2009). Heterogeneous uptake of gases by dust aerosols occurs every spring when massive dust aerosols from the arid regions are mixed with anthropogenic emissions (Pradhan et al., 2010). Previous studies investigated the reaction probability of gas uptake by dust aerosols and the mixing assumption of aerosols, and showed considerable changes in inorganic salt formation and aerosol aging in East Asia (Song et al., 2007, 2012). Mixing and aging of aerosols have important climate implications (Park et al., 2005). However, models still poorly account for all these interactions between the gases and aerosols with a constant uptake coefficient and a uniform mixing assumption. The effects of aromatic and biogenic NMVOCs on O 3 and PM concentrations are yet to be fully understood in East Asia. Extensive observation campaigns, consisting of in-situ, aircraft, ships, and satellite measurements, thus would be needed to achieve a better understanding of the chemical environment in East Asia. d. Deposition processes Physical loss processes including dry and wet depositions affect atmospheric burdens and residence times of air pollutants, of which those properties are critical for the human exposure to air pollution. The deposition of air pollutants causes serious environmental issues including acid rain and soil acidification, as addressed in many modeling studies in East Asia (e.g., Larssen and Carmichael, 2000; Carmichael et al., 2002). The Model Inter-Comparison Study for Asia (MICS-Asia II) evaluated regional air quality models with particular focus on the acid deposition, and found large discrepancies between the models and observations because of the uncertainty with deposition calculations in the models (Wang et al., 2008). Those loss processes are typically parameterized in models employing empirical relationships between the observed dry and wet deposition fluxes of species and key meteorological variables. For example, dry deposition is calculated in typical CTMs as a first order process using dry deposition velocity, which is parameterized as functions of surface types and atmospheric stability conditions (Wesely, 1989). However, its parameterization in models is highly uncertain because of complexities arisen from surface conditions at sub-grid scales (Wu et al., 2011). Wet deposition is also parameterized as a simple first-order process with a constant or a simple form depending on rain intensity (Mircea et al., 2000; Andronache, 2003). Most of these deposition schemes used in models were developed based on the observations in the United States or Europe. Therefore, their parameterizations (e.g., seasonal categories, land use information, wet scavenging efficiency, etc.) may not be suitable for the Asian region. Further development and testing of parameterizations for East Asia are critically

5 31 January 2014 Rokjin J. Park and Sang-Woo Kim 109 Fig. 1. Daily (24-h) mean ambient mixing ratios of ozone observed at Mt. Happo (dots) for the period Also shown are the trend component (solid line) of the best-fit curve for daily means and its uncertainty range (gray region) since Adapted from Tanimoto (2009). needed (Carmichael and Ueda, 2008). 4. Issues in air quality modeling in East Asia a. O 3 air quality O 3 background concentrations in East Asia have gradually increased as shown in the observations at a mountain site in Japan (Tanimoto, 2009). Figure 1 shows the daily mean mixing ratios of O 3 measured at the Mt. Happo Observatory (HPO; o N, o E, 1850 a.m.s.l) for , which is located on a ridge on Mt. Happo and faces the East Sea (Sea of Japan). A continuous increase over the study period was noted with a total increase of about 8 ppbv. The mean growth rate for the 8 years was ~1.0 ppbv yr 1. This increase of mean O 3 concentrations is driven by the increase at the high percentile of O 3 concentrations, indicating the increases of the number of days exceeding O 3 air quality criteria with time. The observed O 3 changes are attributed to the increases in anthropogenic emissions of O 3 precursors in East Asia, particularly in China, because Japan and South Korea have not shown any significant increases in their national anthropogenic emissions over the past decade (Tanimoto et al., 2009). Yamaji et al. (2012) used a regional air quality model to reproduce the observed trend of O 3 concentrations at remote sites in Japan, simply by increasing NO x emissions in eastern China, based on the changes in the bottom-up emission inventory (Fig. 2). The response of O 3 concentrations at the remote sites to anthropogenic emissions is pretty much linear especially to the emission amount from China, indicating the significant role of Chinese emission in determining regional O 3 concentrations in East Asia (Yamaji et al., 2012). According to their modeled O 3 increase, the Chinese anthropogenic NO x emission perturbations alone might account for approximately 60% of the observed O 3 trends at the remote Japanese sites. A similar study was conducted by Jeong and Park (2013), who quantified individual contributions to the observed O 3 concentrations, of anthropogenic emission increases, and the changes of meteorological conditions in East Asia over the Fig. 2. Scatterplot of springtime surface O 3 anomalies and the NO 2 VCDs ratios in April Missing NO 2 data in 2004 are estimated by interpolating NO 2 of March and May in Monthly medians (solid squares) from observed hourly surface O 3 concentrations at Japanese EANET monitoring sites (see Fig. 1) and monthly averages from NO 2 VCDs over east central China measured by SCIAMACHY, their least squares fit (solid line), and interquartile range (error bars). Monthly medians (open circles) from simulated hourly O 3 concentrations for Japanese EANET monitoring sites and monthly averages from NO 2 VCDs over east central China based on the BFM experiment within the perturbation range of 30 to +30% (from the 2004 level), their least squares fit (dotted line), and interquartile range (shaded). Both fitted curves were obtained in the range of of the NO 2 fraction. Adapted from Yamaji et al. (2012). past decades. They suggested using their model results that the meteorological variability accounts for 30% of total O 3 increases in East Asia over the past two decades. The most important meteorological variable affecting O 3 concentrations in East Asia is the cloud cover from their analysis (Fig. 3). However, models typically have large uncertainties in simulating hydrological processes including the cloud cover in East Asia. It is necessary to investigate the interplay between the meteorological conditions and air quality degradation, in the context of chemistry and climate interaction for both the present and the future. b. PM modeling issues Being exposed to a high PM concentration in surface air is detrimental to humans well-being (Nawahda et al., 2012). Aerosol concentrations in East Asia have gradually increased over the past decades (Qin et al., 2010; Guo et al., 2011; Li et al., 2013b), likely due to the increase in anthropogenic emissions in China (Streets et al., 2009). Figure 4 shows a long-term trend of the observed optical depths of absorbing aerosols (ODA) and optical depths of scattering aerosols (ODS) averaged over China from 1957 to The column aerosol loading in China shows a clearly increasing trend over the past four decades, although the rate of increase gradually gets flattened over time. However, Qu et al. (2010) analyzed observed air pollution index records in 86 Chinese cities for and showed a decreasing trend of PM 10 concentrations for the northern cities and no significant change over time in the other regions. This PM 10 trend in China appears to

6 110 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES Fig. 3. Spatial distributions of the simulated 5-yr averaged springtime O3 concentrations (ppbv) for a) , b) and c) differences between two five year period in surface air from the model with the fixed anthropogenic emissions as of 2006 in East Asia. Changes of meteorological variables for d) surface temperature (K), e) large-scale precipitation (mm mon 1 ), f) relative humidity (%), g) cloud fraction (%), h) mixing depth (100 m), and i) surface wind speed (m s 1 ) between 5-yr average periods ( )-( ). Correlation coefficients R of O3 versus meteorological variables for two periods and regression slope S (ppbv D 1, where D is dimension of each meteorological variable) are shown inset. Values are only shown when significant with 99% confidence. Adapted from Jeong and Park (2013). be inconsistent with the observations at clean background sites in East Asia such that the analysis of aerosol concentrations measured at EANET sites in Japan showed a modest but increasing trend of PM concentrations in clean background air, which could be attributed to the anthropogenic emission increases in China (Jeong and Park, 2013). So it is still an open question as to whether or not regional PM concentrations in East Asia have increased especially during the recent past and to what they will be in the near future. Annual and 24-h mean concentrations of PM 10 are typically regulated in countries in East Asia. PM 10 in the atmosphere consists of sulfate, nitrate, ammonium, OA, BC, and soil dust aerosols. Anthropogenic emissions are the main source of these aerosols except for soil dust, which is wind blown mineral from arid and semi-arid regions. Sulfate is the most dominant aerosol species in East Asia, followed by OA and nitrate aerosol (Zhang et al., 2007). One of the long-standing issues in PM modeling in East Asia is that typical air quality models underestimate observed sulfate aerosol concentrations but significantly overestimate nitrate aerosol concentrations (Song et al., 2008). Insufficient oxidation of SO 2 both in the gas-phase and the aqueous-phase could account for the modeled low bias of sulfate aerosol. In particular, the aqueous-phase oxidation of SO 2 is highly model-dependent; most models include aqueous-phase oxidation of SO 2 by H 2 O 2 and O 3, but do not account for SO 2 oxidation in the aqueous phase by other oxidants including O 2, catalyzed by Fe (III) and Mn (II), NO 2, NO 3, and HNO 4, which are thought to be of little importance on a global scale, but significant regionally (Alexander et al., 2009; and references therein). The formation of sulfate on haze particles with high water content might be another contributing factor, which is not included in the model. The formation of fine nitrate aerosol in the atmosphere is

7 31 January 2014 Rokjin J. Park and Sang-Woo Kim 111 Fig. 4. Annual mean optical depths of absorption aerosols (ODA) and optical depths of scattering aerosols (ODS) of 36 sites from 1957 to 2007 and the AODs estimated by Luo et al. (2001) and Qin et al. (2010). Adapted from Wang and Shi (2010). Fig. 5. The variations of the annual mean concentrations of (a) nss- SO 4 2 and (b) NO 3 in TSP, PM 10, and PM 2.5 measured at a Gosan site in Jeju, South Korea. Adapted from N. K. Kim et al. (2011) determined by precursor HNO 3 concentrations as well as by the availability of NH 3 beyond that required for sulfate neutralization to ammonium sulfate (NH 4 ) 2 SO 4 (Seinfeld and Pandis, 1998). The latter is often a critical factor in East Asia (Song et al., 2008), implying the importance of NH 3 emission for reducing uncertainty in fine ammonium nitrate formation in models. An accurate NH 3 emission inventory in East Asia is critically needed for improving PM modeling especially in winter when fine nitrate aerosol is a major contributor to worsening PM air quality. In addition, OA concentrations produced from gas-phase precursors (secondary organic aerosols) are significantly underestimated in models (Heald et al., 2005; Jeong et al., 2011). With these issues discussed above, therefore, a good agreement between the model and the observations for PM mass concentrations does not properly reflect the level of our understanding of sources and regulating processes of PM. It is necessary to validate the chemical composition of aerosols in models extensively by comparing it with long-term observations of aerosol chemical compositions in East Asia to enhance our scientific understanding and to improve models. Although regional observations of chemical compositions were not available, there have been a few site measurements to validate simulated aerosol compositions in East Asia. N. K. Kim et al. (2011) showed a gradual decrease of sulfate aerosol concentrations from 1992 to 2003 and an increase of sulfate aerosol concentrations after 2003 by using chemical observations at a clean background site (Gosan) in Jeju, South Korea for (Fig. 5a), which is largely consistent with the emission trend in China. On the other hand, nitrate aerosol concentrations have increased over the period and have gradually become the dominant aerosol in East Asia (Fig. 5b). One interesting change is that solar absorbing BC aerosol concentrations have also increased (not shown) and resulting radiative warming nearly compensated for the radiative cooling by solar scattering aerosols in 2008 (N. K. Kim et al., 2011). However, models generally suggest decreases in radiative forcing (negative forcing) in recent years relative to the past because of a gradual increase in inorganic scattering aerosol concentrations in China (Jeong and Park, 2013; Li et al., 2013b). This apparent inconsistency should be further investigated using region-wide observations and an improved model. The great challenge of PM modeling is to simulate naturally driven aerosols such as soil dust, biomass burning aerosols, sea salt, and etc. In particular, soil dust aerosols from Mongolia and the Taklimakan and Gobi deserts are one of the serious air quality concerns in East Asia, causing substantial economic losses and environmental damage in spring (Ku and Park, 2011; Wang et al., 2013). In favorable synoptic conditions, Asian dust aerosols can even be transported across the Pacific, affecting the air quality in the western United States (Husar et al., 2001; Fairlie et al., 2007). Emission of dust aerosols is typically parameterized in models using wind speed, soil water content, and vegetation cover but is known to be highly uncertain (Ku and Park, 2013; and references therein). A dust model inter-comparison study that examined the current regional dust models, was applied to East Asia, arriving at the conclusion that the dust aerosol transport patterns from the source regions were usually very similar, but the simulated

8 112 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES dust concentrations in the surface air differed by over two orders of magnitude in the dust source regions mainly because of the large uncertainties in the dust emission simulation (Uno et al., 2006). Therefore, reducing these uncertainties and accurately quantifying dust emissions are critical for air quality models in East Asia. 5. Effects of short-lived chemicals on regional climate In the troposphere, O 3 absorbs effectively the radiation on both UV/VIS and IR wavelengths, and the global mean net radiative forcing of tropospheric O 3 has been estimated in the range of 0.25 to 0.65 W m 2 (IPCC, 2007). Because O 3 is chemically produced from its precursor species, such as CH 4, CO, NMVOCs, and NO X, the O 3 radiative forcing can be attributed to these precursor emissions (Shindell et al., 2009). In East Asia, as discussed above, O 3 background concentrations have increased (Tanimoto et al., 2009), and are expected to increase significantly in the near future owing to the increase in NO x emissions (Ohara et al., 2007; Yamaji et al., 2012). This observation implies the increase in climate warming by the O 3 forcing in East Asia in the future. Although the long-term change in aerosols needs to be better understood, East Asia is the largest source region for anthropogenic aerosols (Streets et al., 2009). Soil dust aerosols from the deserts and smoke aerosols from wildfires and agricultural burnings are additional important seasonal contributors to aerosol loadings in East Asia. The MODIS-derived aerosol optical depths (AODs) showed a large seasonal variation of aerosols, a maximum in spring and a minimum in autumn and winter (Kim et al., 2007). The AERONET Sun/ sky radiometer measurements also showed high values of the monthly mean AODs in June over the industrialized coastal regions of China and the Yellow Sea, the Korean Peninsula, and Japan (Fig. 6). The temporal and spatial distributions of aerosols in East Asia are significantly modulated by synoptic meteorology and regional climate systems such as the Asian monsoon (Shimadera et al., 2013). Several recent studies showed a close relationship between the pollution aerosols and the Asian summer monsoon system (Yoon et al., 2010; Zhao et al., 2010; Yan et al., 2011). An observational study by Yoon et al. (2010), for example, showed that the increase of about 40-50% in the AOD in July 2005 over China, the Yellow Sea, and the Korean Peninsula, compared to the eight-year July average, can be attributable to spatial patterns of Asian summer monsoon circulation and associated precipitation (Fig. 7). The regional effects of aerosols on climate are complex depending on the chemical compositions of aerosols (Ramana et al., 2010). Typical scattering aerosols such as ammonium sulfate and nitrate result in significant surface cooling, whereas absorbing aerosols including BC and brown carbon aerosols cause atmospheric heating (Ramanathan and Carmichael, 2008). Based on the seven-year analysis of AERONET observations at Gosan, Korea, Kim et al. (2010) showed a relatively low single scattering albedo (SSA < 0.93) for February-May, which is indicative of moderate to high absorbing aerosols and supportive of the large surface dimming and equally large atmospheric heating. The average AERONET retrieved SSA is about 0.9 in the visible wavelength range at all stations in China during EAST-AIRE (Li et al., 2007). Most studies, based on shipboard (Markowicz et al., 2003), ground-based radiometers (Kim et al., 2005; Yoon et al., 2005), aerosol transport-radiation models (Takemura et al., 2003), in situ measured aerosol data and radiative transfer models (Conant et al., 2003), and satellite radiance measurements (Nakajima et al., 2003), generally revealed strong negative aerosol forcing at the surface and positive aerosol forcing greater than 10 W m 2 in the atmosphere. For example, at Gosan, the annual average clear-sky direct forcing at the surface was ± 9.21 W m 2 and at the top of the atmosphere (TOA) was ± 4.44 W m 2, thereby leading to an atmospheric absorption of ± 5.82 Wm 2, which is translated to an atmospheric heating of 1.5 to 3.0 K day 1 (Kim et al., 2010). This solar absorption by aerosols is one of the critical issues with large uncertainties in climate change studies. Among lightabsorbing aerosols, BC is known to be the most effectively absorbing solar radiation (Ramanathan and Carmichael, 2008; and references therein). Ramana et al. (2010) directly observed a large enhancement of atmospheric heating by BC from unmanned aerial vehicle (UAV) measurements in Jeju Island, Korea during summer Brown carbon aerosols, which are certain fractions of OA, were also found to be ubiquitous and to effectively absorb solar radiation (Alexander et al., 2008). As shown in Fig. 8, Park et al. (2010) estimated that the averaged radiative forcing of brown carbon aerosol over East Asia is 0.43 W m 2 and 0.05 W m 2 at the surface and at the TOA, accounting for about 15% of total radiative forcing ( 2.2 Wm 2 and 0.33 W m 2 ) of absorbing aerosols (BC + brown carbon aerosols) by using the GEOS-Chem model. The question is how the perturbation of earth s radiation balance by aerosols has affected and will affect climate in East Asia. This question is particularly challenging because the aerosol forcing is highly sensitive to the change of climate. In other words, climate and aerosols or short-live chemicals, in general, are interwoven nonlinearly. An understanding on the complex two-way interaction between the two is required to quantify the effect of the short-lived chemicals on climate; it can be achieved by using a sophisticated earth system model that includes a fully coupled chemistry-climate simulation integrated with observational constraints. A few modeling studies, although they did not account for the two-way interaction between the short-lived chemicals and climate, investigated the effect of aerosols on regional climate in East Asia. For example, Menon et al. (2002) argued that the presence of absorbing aerosols such as BC has modified precipitation trends in China over the past several decades, with increased rainfall in the south and drought in the north. Youn et al. (2011) used a general circulation model with simulated aerosols concentrations from the GEOS-Chem to examine the effects of Siberian fire aerosols on regional meteorology in East

9 31 January 2014 Rokjin J. Park and Sang-Woo Kim Fig. 6. Spatial distributions of monthly mean AOD at 550 nm determined from MODIS measurements and NCEP/NCAR reanalysis wind vectors (m s 1) at the 850 hpa pressure level over East Asia from April 2000 to June Adapted from Kim et al. (2007). 113

10 114 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES Fig. 7. Spatial distribution of aerosols and rainfall for July 2005: (a) MODIS-derived AOD at 550 nm with NCEP/NCAR reanalysis wind vectors (m s 1 ) and geopotential height (m) at 850 hpa and (b) their anomalies. (c) GPCP rainfall (mm day 1 ) with equivalent potential temperature (K) and (d) their anomalies. Adapted from Yoon et al. (2010). Asia during May 2003, and they suggested that the observed changes in meteorological variables, including the surface temperature, surface pressure, and precipitation over the North- West Pacific, were likely due to Siberian fire aerosols, indicating a direct and significant impact of aerosols on regional meteorology. These results suggest that an accurate quantification of short-lived air pollutants effects on regional climate, and vice versa, is needed to better understand regional-scale climate changes in particular under the future global warming conditions. 6. Future of air chemistry modeling With the advance of the monitoring capability by in-situ, aircraft, and satellite measurements, integrating the observations and the models became even more necessary to allow deeper understanding of sources, transport, and transformation of air pollutants (Bowman, 2013). Inverse modeling has been proven to be an effective way to provide observational constrains on a priori information and to reduce errors in simulations by incorporating observations in models (Rodgers, 2000). A significant advance in inversion methods has been achieved, and they are widely applied in meteorological models for weather forecasting (Kalnay, 2003). Because of the continuity of atmospheric phenomena, a meteorological model is very sensitive to the initial condition that the inverse modeling tries to optimize based on the observations. A CTM, however, is sensitive to both initial and boundary conditions, and the latter is often more important than the first especially for short-lived air pollutants. The most important boundary condition is the emission of trace gases and aerosols from various sources, which is highly uncertain especially in East Asia. Reducing this uncertainty in the emission inventory can effectively be achieved by using the

11 31 January 2014 Rokjin J. Park and Sang-Woo Kim 115 Fig. 8. Simulated radiative forcing of BC (left) and brown carbon aerosols (right) at the surface (top) and at the TOA (bottom) over East Asia in spring Simulated aerosol mass concentrations were used in MIE code for calculating AODs assuming the lognormal distribution. Radiative transfer computations were conducted using the GEOS-3 assimilated meteorology field with the simulated AODs. Adapted from Park et al. (2010). inverse modeling in CTMs with atmospheric observations of air pollutant concentrations. The use of such inverse methods with CTMs, however, has relatively been limited because of the lack of in-situ chemical observations; therefore analytical inversion by combining in-situ observations and CTMs has been used to constrain sources of CO (Palmer et al., 2003) and some aerosol species (Hakami et al., 2005; Ku and Park, 2011) in East Asia. On the other hand, satellites observations are desirable for a large state vector, enabling us to use the advanced inverse methods such as the adjoint method (Kopacz et al., 2009), the four-dimensional ensemble Kalman filter (Sekiyama et al., 2010) and the four-dimensional variational method (Henze et al., 2009; Park and Russell, 2013) in CTM models. Their observations of column concentrations of trace gases and aerosols have been successful in detecting changes in anthropogenic and natural emissions (Richter et al., 2005; van der A et al., 2008) and in validating CTMs (Han et al., 2009; Park et al., 2011). Previous studies have also applied the various inverse techniques to integrating satellite observations and CTMs to update top-down emission inventory of important species for air pollution (Heald et al., 2004; Martin et al., 2006; Henze et al., 2009; Kopacz et al., 2009). Along with the top-down estimates of emission, the bottomup emission inventory should further be developed in terms of improving its spatial and temporal allocation by which the model resolution is limited. Typical bottom-up emissions are compiled based on the national statistics of fossil fuels use from which emissions of trace gases and aerosols are spatially and temporally allocated, based on some proxy data such as population density and traffic volume. However, the use of these proxy data often causes significant errors in the bottomup emission inventory, since the anthropogenic emission is not necessarily proportional to the population density (Zhang et al., 2009). The improvement of the bottom-up emission inventory is particularly necessary for gases such as NH 3 and NMVOCs (Tsigaridis et al., 2005; Zhu et al., 2013). The importance of NH 3 emission for PM modeling is discussed above. For NMVOCs the improvements are twofold: biogenic sources and the chemical speciation of total NMVOCs. Although NMVOCs emissions from anthropogenic sources are dominant in polluted urban regions (Zhang et al., 2009), the biogenic sources are more important for East Asia (Fu et al., 2007). Moreover, mountains and vegetative surfaces, where biogenic VOCs emissions occur considerably, typically surround mega-

12 116 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES cities in East Asia. O 3 concentrations are generally low due to the titration of high NO x from the heavy mega-city traffic, but it increases dramatically in the case in which high NO x in polluted air is mixed with biogenic VOCs from surroundings (Kim et al., 2013). An improved estimate of biogenic VOCs emission is thus key to the success of O 3 modeling and possibly OA simulations as well. The bottom-up emission inventory generally provides the total mass of NMVOCs, which are comprised of numerous chemical species with different reactivity. The chemical speciation of total NMVOCs is often carried out depending on the chemical scheme used in models. Therefore, despite the identical total NMVOCs emission used in different models, a wide range of O 3 concentrations can be calculated, indicating the large sensitivity of models to the chemical speciation of NMVOCs. Although its significance is well known in the community, the information is limited by the measureable NMVOC species from different anthropogenic sources. Environmental agencies in developed countries (e.g., U.S. EPA) have provided limited but available chemical speciation information of NMVOCs in their domestic inventories and have continued to improve their database using improve measurement techniques. Further emphasis is much needed to improve the chemical speciation information of NMVOCs in the bottom-up emission inventory in East Asia. Climate changes, induced by long-lived greenhouse gases as well as short-lived climate pollutants, affect meteorological conditions, which determine distributions of air pollutants. The two-way interaction between the meteorology and air quality would be a great challenge to advance our capability to predict environmental changes in the future. Along with this interaction, a more accurate projection of future emission changes is crucial to prepare an adaptation strategy for future environmental changes. East Asia is a complex region politically and economically, including one of the most developed countries, Japan, one of the most dynamic countries, China, and one of the poorest countries, North Korea. Therefore, it is difficult to project future societal and economical changes and their resultant emissions, which were mostly assessed using an emission model developed by Europe and the United States. It is necessary to develop an emission model suited for East Asia to address all the sensitive regional issues and to project regional changes in the future. 7. Conclusions Concerns for air quality degradation have grown in East Asia because of increases in trace gas emissions particularly from China. CTMs have widely been used to understand sources, transport, and chemical transformation of air pollutants. Here, we reviewed our knowledge related to the present air quality issues and their modeling focusing on O 3 and PM in East Asia. CTMs have been developed to obtain quantitative information of the concentrations of chemical species in the atmosphere, whose distributions are determined by the four controlling processes: emission, transport, chemistry, and deposition. In East Asia, a large uncertainty still remains with the emission inventory, which limits the capability of CTMs. In addition, the dry deposition processes used in CTMs were developed in Europe and North America and may not properly represent conditions in East Asia. A need to improve or develop deposition modules for East Asia is required to improve CTMs for the future. O 3 background concentrations in East Asia have gradually increased over the past. Many studies attributed this O 3 increase to the increase in anthropogenic emissions in China, which could not explain the observed trend entirely. Not only changes in precursor emissions but also meteorological conditions driven by greenhouse gases may have contributed to the longterm O 3 change, which entails a further study of chemistryclimate interaction in this region. On the other hand, the PM trend is highly debated whether it has increased over the past or not. Some observations showed a clear increase but others showed the opposite. Even though the model reproduces observed PM mass concentrations, chemical species consisting of PM such as sulfate, nitrate, BC, OA, and soil dust aerosols are not captured by the model, which indicates our lack of understanding of the source, transformation and transport, deposition of those individual species in East Asia. We need region-wide long-term observations of those chemical species to evaluate and improve the model in East Asia. O 3 and PM play a critical role in the earth s radiation budget and affect the climate as short-lived climate forcers. Thus, the highest concentrations of these species in East Asia have a huge climate implication, which needs to be studied extensively focusing on their effects on regional climate and the atmospheric responses, which affect air quality as well. This complex interplay between air chemistry and climate would be a challenging issue in CTMs for this region. With the advance of the monitoring capability by in-situ, aircraft, and satellite measurements, integrating the observations and the models has became necessary to provide a deeper understanding of sources, transport, and transformation of air pollutants. An inverse modeling is proven to be an effective way to provide observational constrains on a priori information and to reduce errors in simulations by incorporating observations in models. Reducing the uncertainty in the emission inventory can effectively be achieved using the inverse modeling in CTMs with atmospheric observations of air pollutant concentrations. We think that an application of inverse modeling and its technical development should be further emphasized in CTMs for East Asia. Along with the top-down estimates of emission, the bottomup emission inventory should further be developed in terms of improving its spatial and temporal allocation in East Asia. Especially, it is necessary to develop an emission model suited for East Asia to address all the sensitive regional issues and to project regional changes in the future, which enables us to