Optical and microphysical particle properties, including the particle single-scattering albedo, were derived from multiwavelength aerosol Raman lidar observations at Gwangju (35.10°N, 126.53°E), and Anmyeon Island (36.54°N, 126.33°E), South Korea. The results present aerosol properties in various heights of the atmospheric aerosol layers on 12 different measurement days. The measurement cases differ in terms of aerosol loading as well as aerosol types (long-range transported urban/industrial haze from China, regional/local haze that mainly originated from the Korean peninsula, and smoke from forest fires in east Siberia). The origin of the particle plumes was determined from chemical transport modeling with the FLEXPART model. We find comparably clear differences between the optical and microphysical properties of the aerosol types. Local haze aerosols show effective radii of 0.32 ± 0.02 μm at relative humidity of 60–80%. The effective radii of urban/industrial haze and smoke aerosols are approximately 0.26 μm and 0.27 μm at relative humidity of 35–60%. Light absorption, expressed in terms of single-scattering albedo, is 0.87 ± 0.02 (at 532 nm) for urban/industrial haze from China. This value is considerably lower than the single-scattering albedo of smoke aerosols from Siberia and northern China (0.92 at 532 nm) and of regional/local haze aerosols (0.97 ± 0.01 at 532 nm). We find a hygroscopic growth factor (from relative humidity of 30% to relative humidity of 85%) of 1.49 ± 0.36, if we consider all measurements.
 In that regard the aerosol single-scattering albedo (SSA) is a key parameter that determines the influence of aerosols on global and regional climate change [Novakov et al., 2003]. Aerosol particles cause atmospheric cooling under almost all conditions when the SSA exceeds 0.95 (at visible wavelengths). If the SSA is less than 0.85, aerosols heat the atmosphere significantly and tend to cause warming at the surface, even though the surface underneath the aerosol layers in that region may receive less sunlight [Ramanathan et al., 2001; Novakov et al., 2003]. In addition, there are indications that absorbing aerosols alter regional atmospheric stability and vertical motions (strength of convection) [Wendisch et al., 2008], and affect the large-scale circulation and the hydrologic cycle (like evaporation) with significant regional climate effects [Menon et al., 2002]. All these reasons call for an increased effort to characterize the light-absorption characteristics of aerosols under ambient conditions and on a vertically resolved scale.
 In the past decade a large number of field experiments, as for example the Smoke, Clouds, Aerosols, Radiation-Brazil (SCAR-B) experiment [Kaufman et al., 1998], the Second Aerosol Characterization Experiment (ACE 2) [Raes et al., 2000], the Tropospheric Aerosol Radiative Forcing Observational Experiment (TARFOX) [Russell et al., 1999], the Indian Ocean Experiment (INDOEX) [Ramanathan et al., 2001], the Southern African Regional Science Initiative (SAFARI) [Swap et al., 2002], and the Zambian International Biomass Burning Emissions Experiment (ZIBBEE) [Eck et al., 2001] were carried out with the goal of better understanding the direct and indirect effects of the light-absorption characteristics of aerosol on the climate and to reduce the uncertainty of climate forcing studies.
 With regard to the situation of light-absorbing aerosols over East Asia we still have a very poor understanding of such particles, in particular with regard to particles in lofted layers of the atmosphere. Asian aerosol sources are unlike those in Europe and North America. Asian aerosols add more light-absorbing soot because of the high amount of fossil fuel that is used in East Asia [Streets et al., 2001; Huebert et al., 2003]. The amount of BC aerosol emissions, mainly produced by the combustion of coal and biofuels in residential areas, is particularly large in China [Streets et al., 2001; Cao et al., 2006; Ramanathan and Carmichael, 2008].
 Aerosol composition is particularly complex in East Asia, because of the mixing of anthropogenic and natural aerosol types in that region. East Asia is one of the major source regions for anthropogenic aerosols in the world [Li et al., 2007; Streets et al., 2007]. Next to fossil fuel combustion there is strong emission of mineral dust from desert regions in Central Asia [Iwasaka et al., 1988]. Forest fire smoke, which is transported from Siberia to east China, Korea, and Japan each year, is also believed to be a major source of particulate pollution in the region [Lee et al., 2005].
 The combined effect of expanding arid dust-producing regions, increasing regional populations, and increasing fossil fuel usage in China causes an annual increase of the atmospheric aerosol burden, which is in contrast to a decreasing trend in other source regions of the world [Lee et al., 2006]. Furthermore, the regionally produced light-absorbing aerosols affect the atmospheric environment on a global scale due to long-range transport that results from the specific meteorological conditions in that region [Ramanathan and Feng, 2009].
 The largest coordinated field activity that collected information on light-absorption characteristics in the upper planetary boundary layer and the free troposphere over East Asia to date has been ACE Asia [Huebert et al., 2003]. The activities however delivered only information on the basis of airborne and ship-borne in situ measurements over the east China Sea. On the one hand these data do not represent ambient atmospheric conditions, as in general particles are dried to comparably low relative humidity during the particle collection process. On the other hand we cannot exclude loss effects by the particle inlet system of the research aircraft, which means that the derived optical and microphysical properties may not fully describe the true aerosol situation. Furthermore the core measurement period of ACE Asia was comparably short, thus limiting the statistical representativeness of the aerosol data.
 The main objective of this paper is to investigate the light-absorbing characteristics of atmospheric aerosols over East Asia on the basis of their origin. We present vertically resolved optical and microphysical particle properties and specifically the single-scattering albedo of atmospheric aerosols over the Korean peninsula. We furthermore present some results on the dependence of some aerosol parameters on relative humidity. The observations were made episodically since 2004 at Gwangju, Korea (35.10°N, 126.53°E) and in Anmyeon Island (36.54°N, 126.33°E) which is located 200 km northwest of Gwangju and off the midwest coast of South Korea. Table 1 gives an overview on 12 measurement cases that we selected for our analysis.
Table 1. Mean Values of Ångström Exponent (Å), Lidar Ratio (S in sr), and Relative Humidity (RH in Percent) Measured for the Observed Aerosol Layers
Section 2 describes the multiwavelength Raman lidar and the data analysis methodology that is used for the optical and microphysical particle characterization. Section 3 presents the measurements. We discuss our results in section 4. We close our contribution with a summary in section 5.
2.1. Optical and Microphysical Particle Parameters
 A detailed description of the multiwavelength aerosol Raman lidar system, the methodology of analysis of the optical data and the uncertainty analysis are given by Noh et al. [2007, 2008, 2009]. Vertical profiles of the particle volume extinction coefficients are derived at 355 and 532 nm with the use of nitrogen vibrational Raman signals detected at 387 and 607 nm, respectively [Ansmann et al., 1990]. A sliding average of 360 m below 2400 m altitude and 600 m above 2400 m altitude is applied to the range-corrected nitrogen Raman signals. The relative uncertainty is 10–20% for the aerosol layers discussed in this contribution. Particle backscatter coefficients at 355, 532, and 1064 nm are calculated with the Raman method [Ansmann et al., 1992]. Relative uncertainties are ≤5% in the center of the aerosol layers.
 Vertical profiles of relative humidity are provided by radiosonde observations conducted four times (0300, 0900, 1500, and 2100 local time) a day by the Korean Meteorological Administration (KMA) at Gwangju airport. The airport is located about 5 km southwest of the lidar site.
 The inversion algorithm that is used for the retrieval of the microphysical parameters from the optical data is described by Müller et al. [1999a, 1999b] and Veselovskii et al. [2002, 2004]. Particle backscatter coefficients at 355, 532, and 1064 nm and particle extinction coefficients at 355 and 532 nm are taken from selected height segments of the optical profiles and serve as input to the data inversion algorithm. The inversion algorithm provides approximations of volume size distributions from which we compute particle effective radius, volume and surface area concentration of the particle size distribution, and the complex refractive index. Particle size distribution and complex refractive index are used to calculate the particle single-scattering albedo with a Mie-scattering code [Bohren and Huffman, 1983]. A detailed description of the retrieval procedure for the single-scattering albedo is given by Noh et al. . The single-scattering albedo is given for the wavelength at 532 nm.
 Several studies on retrieval errors have been carried out in the past [e.g., Müller et al., 1999b; Veselovskii et al., 2002, 2004; Böckmann et al., 2005]. The studies show that for measurement errors as given in our present study effective radius may be retrieved with an accuracy of better than 30%. Volume and surface area concentrations are derived with an accuracy of better than 50%. With regard to the complex refractive index, its real part can be inferred to an accuracy of approximately ±0.05 if measurement errors are less than 15%. The imaginary part can be estimated to ±50% if it is >0.01i. If the imaginary part is less than 0.01i, uncertainties can grow to 100%.
 We point out that in all cases presented here, the complex refractive index is given as a mean, wavelength-independent property which is valid for the wavelength range covered by the Nd:YAG laser. The inversion was carried out ten times for each optical data set. The noise level of the optical data was 10–20%, and we used a Gaussian distribution of the error bars. The standard deviations of the ten inversion results for each data set are reported as uncertainties in Table 3.
2.2. Backward Dispersion Model
 FLEXPART [Stohl et al., 2005] simulations were used to identify the origin of the observed aerosol layers. FLEXPART is a Lagrangian particle dispersion model. The input wind fields are derived from the meteorological analyses of the Global Forecast System (GFS). The simulations are initiated with particles released within the height range and observation time of the actual aerosol layers above our lidar site. The simulations presented in this paper were performed 10 days backward in time. The residence times of the particles were used to reproduce the pathways of the aerosol plumes over Asia.
 We also used the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (R. R. Draxler and G. D. Rolph, HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model, 2003, http://www.arl.noaa.gov/ready/hysplit4.html) for computations of 10 day backward trajectories of air parcels arriving over the Gwangju and Ammyeon site in altitudes in which we observed aerosol layers. The HYSPLIT results were used to complement the FLEXPART results.
Noh et al.  studied average optical and microphysical properties of different atmospheric aerosol types on the basis of haze events over Gwangju, Korea. However, these studies were insufficient to understand long-range transport of Asian aerosols in the upper planetary boundary layer and the free troposphere, because the haze events were observed only within the planetary boundary layer (PBL). In this contribution we focus our data analysis on the light-absorption properties for different air mass transport patterns and observation heights of the investigated aerosols.
Figure 1 shows the mean profiles of the particle backscatter and extinction coefficients measured on four days, i.e., on 10 February, 13 and 14 June, and 30 October 2004. High values of particle extinction and backscatter coefficients were found up to 3.5 km height above ground on 10 February 2004. On the other three days the extinction and backscatter coefficients strongly decrease to background values around 1.5–2 km height above ground. The geometrical depth of the planetary boundary layer was around 1.5–2 km on those days according to aerosol vertical profiles and virtual potential temperature profiles from radiosonde.
Figure 2 presents the FLEXPART backward simulations and HYSPLIT backward trajectories for aerosol layers in heights that correspond to increased backscatter coefficients. The source regions of the observed aerosol layers can be clearly distinguished. The FLEXPART and HYSPLIT computations show that the source regions of the aerosol emissions on these four measurement days were located in urban/industrial areas of China, in Siberia and in northern China, and in the Korean peninsula. We find that these different source regions were responsible for the aerosol load in different heights on each of the measurement days.
 Different source regions and origin of each observed aerosol layer will induce different aerosol compositions. Smoke particles consist primarily of black carbon (BC) and organic carbon (OC). Urban and industrial haze which results from fossil fuel combustion in populated urban/industrial areas, likely consists of sulfates (SO42−), nitrates (NO3−), OC, BC, and ammonium (NH4+). The different light-absorbing characteristics, according to source regions, stem from varying mass fractions of BC in these aerosols [Omar et al., 2005]. Particle emission characteristics (as for example temperature of combustion processes, composition of fuel, coating of absorbing particles with nonabsorbing components, external mixtures of different particle types) also have influence on the light-absorbing characteristics [Dubovik et al., 2002].
 The observed aerosols are categorized as long-range transported haze, local haze and smoke based on the FLEXPART and HYSPLIT backward simulations. We assume that the long-range transported haze is mainly composed of anthropogenic pollution particles generated from urbanized and industrialized regions in east China. Regional/local haze may contain a mixture of anthropogenic particles (emitted from urban and industrial regions) with emissions from agricultural activity (burning of crop waste) around our measurement site.
 The aerosol layer on 10 February 2004 was above the PBL (>1.0 km above ground level) and its geometrical depth varied between 1.0 and 3.5 km during the measurements. The aerosol originated from industrial regions of China. We assume that local sources along the transport path had little effect as the transport altitude was above 1.0 km, according to the HYSPLIT results shown in Figure 2b. We classify this aerosol layer as long-range transported haze.
 The air mass movement pattern within the PBL below 2 km (above ground) on 13 June is shown in Figures 2c and 2d. MODIS fire maps were obtained from the Fire Information for Resource Management System (FIRMS; http://maps.geog.umd.edu/firms/) [Davies et al., 2009] for the period of 8–12 June 2004; the results from FIRMS are not shown in this paper.
 The maps show many fire spots in west Mongolia and northeast Siberia (around Lake Baikal) during this period in June. Although the air mass was transported from Siberia and northern parts of the Korean peninsula on 13 June, the air mass did not pass over source regions of smoke. For this reason, we classify the aerosol layer in the PBL as regional/local haze that mainly originated from within the Korean Peninsula. It is unlikely that this haze was affected by smoke transported from Siberia, when we consider the air mass movement pattern as suggested by FLEXPART, HYSPLIT backward trajectories and MODIS fire maps. The lidar signals above the PBL were too weak and could not be analyzed.
 The air mass observed on 14 June shows different transport patterns with altitude. The air mass observed within the PBL shows the same pattern as on the previous day. For this reason we classify the aerosol layer in the PBL as regional/local haze, too. However, the aerosol layer observed above the PBL originated from Siberia and Mongolia. The aerosol above the PBL rather likely consisted of pure smoke particles, when we consider the MODIS fire maps and the source region as indicated by the FLEXPART and the HYSPLIT results.
 The aerosols observed on 30 October also show an air mass pattern that varies with altitude. According to the FLEXPART results, the aerosol layer observed within the PBL below 1.5 km (above ground) most likely consisted of local/regional haze that originated from inside the Korean peninsula. This classification is corroborated by the HYSPLIT results. The air mass did not pass over source regions in China. The aerosol layers above the PBL are different from those within PBL regarding their source region. According to the HYSPLIT simulations these aerosol particles were long-range transported from China, and we therefore classify them as long-range transported haze.
3.2. Anmyeon Island Site
 The measurement site at Anmyeon island is located at an altitude of 42.7 m above sea level. The island is near the west coast of Korea. The site is located about 500 km across the east coast of China and 100 km west of the Seoul metropolitan area. Main aerosol types of atmospheric aerosols observed on the measurement days reported here were long-range transported particles from China, locally generated urban haze particles from the Seoul metropolitan area, and haze particles emitted from agricultural activity in the Korean peninsula.
Figure 3 shows the mean profiles of the particle backscatter and extinction coefficients measured between 27 May and 7 June 2005. High values of particle extinction coefficients up to 0.3 km−1 at 532 nm are found for the measurement period from 27 to 30 May. Top heights of the aerosol layers during that measurement period varied from 2 to 5 km.
Figure 4 presents the FLEXPART backward simulations and HYSPLIT backward trajectories. In the case of 27 May the air mass was directly transported from China. We assume that the impact by locally generated aerosols on the aerosol loading was negligible, since there are no dominant aerosol sources around the observation site, and Seoul was not directly in the transport path of the air that arrived over Anmyeon Island.
 The measurements from 28 and 29 May also show that the aerosols were mainly transported from China. In that case there may have been some impact by urban haze generated closer to the observation site since the air mass passed over the Seoul metropolitan area prior to arrival over Anmyeon island.
 A different air mass transportation pattern was observed on 30 May. In contrast to the previous days, the air mass arrived from easterly directions, and we assume that the air was loaded with regional/local haze from the Korean peninsula. The same advection pattern of air from more easterly directions continued throughout the rest of the observation period (7 June).
4.1. Optical Properties
 The profiles of the measured optical properties were split into sublayers of variable height. Table 1 lists the Ångström exponents, the particle lidar ratios at 355 and 532 nm, and relative humidity of the selected particle layers and classifies them according to the measurement sites and main source regions.
 The Ångström exponent in the wavelength range from 355 to 532 nm varies between 0.83 and 1.86. Smallest Ångström exponents were measured for regional/local haze that was observed in the PBL over Gwangju on 13 and 14 June and 30 October 2004. Values vary from 0.83 to 1.03 with an average of 0.96 ± 0.09. We measured relative humidities of 70% ± 7%.
 The measurements of long-range transported haze at Anmyeon Island show the highest Ångström exponents. Values range between 1.24 and 1.86. The average value is 1.40 ± 0.27, indicating comparably small particles. Relative humidity of 41% ± 8% is comparably low.
Table 1 summarizes the lidar ratios of each layer. We see a variation of the lidar ratio from 49 sr to 102 sr and 44 sr to 96 sr at 355 and 532 nm, respectively. The lidar ratios vary depending on the measurement sites and under the various air mass transport patterns identified in this study.
Figure 5 summarizes the results for the lidar ratios. Table 2 presents a comparison of the lidar ratio measured in this study to results from previous studies. We find a comparably clear separation of the lidar ratios according to our aerosol type classification. The aerosol that is classified as regional/local haze at Gwangju shows lower lidar ratios with average values of 52 ± 3 sr and 59 ± 5 sr at 355 and 532 nm, respectively. In contrast to the Gwangju results, the lidar ratios observed at Anmyeon island show comparably high values of 70 ± 7 sr and 82 ± 10 sr at 355 and 532 nm, respectively. These considerably larger lidar ratios may be the result of different aerosol sources at the lidar two sites. Anmyeon island may have been more affected by particles transported from the Seoul metropolitan area. In contrast to Anmyeon island, Gwangju was not directly downwind of Seoul during the measurement times considered here.
Table 2. Mean Values and 1 Standard Deviations of Lidar Ratios at 355 and 532 nm According to Aerosol Types and Source Region
 The lidar ratios at Anmyeon island are also considerably higher than lidar ratios measured for south Indian aerosols and southeast Asian aerosols [Müller et al., 2007]. In these cases we find values of 37 ± 10 sr at 532 nm and 51 ± 20 sr at 355 nm. These areas are mainly affected by biomass burning aerosols or mixtures of biomass burning aerosols and urban, light-absorbing haze (comparably high soot concentration). In contrast, higher lidar ratios of 65 ± 16 sr at 532 nm were found for north Indian pollution which is characterized by a high concentration of light-absorbing urban/industrial pollution [Franke et al., 2003; Müller et al., 2007].
 The average lidar ratios of smoke aerosols are 69 ± 1 sr and 63 ± 2 sr at 355 and 532 nm, respectively. Similarly high lidar ratios were found from multiwavelength Raman lidar observations of Siberian forest fire smoke detected over Tokyo, Japan [Murayama et al., 2004] and of biomass burning aerosols advected from Canada to Germany [Wandinger et al., 2002]. Our numbers are different from lidar ratios of low light-absorbing forest fire smoke observed over Central Europe in 2003 [Müller et al., 2005, 2007].
 The aerosols that were advected from China to the Gwangju site differ in lidar ratios depending on the observation day. The average lidar ratios of long-range transported haze observed on 10 February 2004 are 58 ± 2 sr and 50 ± 6 sr at 355 and 532 nm, respectively. The lidar ratios at 532 nm are similar to or slightly higher than those of haze aerosols observed in north (47 ± 6 sr) and south (38 ± 7 sr) China [Müller et al., 2007]. In the latter case (south China) we assume that a comparably high amount of biomass burning particles (like burning of agricultural waste) and high relative humidity may be responsible for the very low lidar ratios; see also the results for south Indian pollution. This mixture of the urban/industrial haze with biomass burning aerosols in China seems to generate considerably lower lidar ratios than the regional/local haze over Korea.
 High lidar ratios are typically observed for light-absorbing particles that are transported from urban/industrial areas in China to Gwangju [Noh et al., 2008]. The long-range transported haze observed at Anmyeon Island also shows comparably high lidar ratios of 61 ± 5 sr and 73 ± 6 sr at 355 and 532 nm, respectively. As in the case of the regional/local haze observed over Anmyeon island the high lidar ratios indicate comparably strong light-absorbing aerosols, see also Franke et al.  who report on lidar ratios of Indian pollution and Müller et al.  who provide an overview on lidar ratios for different regions in the world. Furthermore, Ferrare et al.  observed high lidar ratios of 68 ± 12 sr at 355 nm in the southern Great Plains of north-central Oklahoma (USA) and reported that such high lidar ratios were associated with air masses from urban/industrial areas.
 The major factors controlling the lidar ratio may be different aerosol composition according to source region and relative humidity. The combined effect of these two factors can explain why the lidar ratio often shows such different values even if the source regions are similar.
 Atmospheric aerosols show different composition and mixture of organic and inorganic compounds with different source regions, therefore the particulate scattering characteristics may be quite different. Relative humidity plays an important role in particle properties and thus for the lidar ratio. Particles absorb or release water in response to changes in relative humidity. Their optical properties such as scattering, backscattering and absorption can be significantly affected by these changes [Kovalev and Eichinger, 2004]. Takamura and Sasano  examined the lidar ratio variation in dependence on relative humidity. Their analysis shows that lidar ratios decrease from 80 sr to 50 sr (at 355 nm) if relative humidity increases from 40% to 95%.
 We find an unusually high lidar ratio on 30 October 2004. At the moment we cannot explain in full detail the reason for the high values, i.e., 92 ± 10 sr at 355 nm and 87 ± 7 sr at 532 nm.
 Reason may be the long transport time in combination with the source region which may have been South Asia, e.g., North India. The backward trajectories show that the air mass observed on 30 June was slowly advected at low altitudes from South Asia, see Figures 2k and 2i. During transport the air from South Asia mixed with urban/industrial haze over China. The slow transport may have given the air mass plenty of time to uptake a high amount of light-absorbing material.
 The optical data of the aerosol layers discussed in section 4.1 were averaged and subsequently used for the data inversion. Figure 6 shows the results for particle effective radius, imaginary part of the complex refractive index and single-scattering albedo (at 532 nm). Table 3 summarizes the results. Table 4 summarizes for comparison the results for particle effective radius, real and imaginary parts of the refractive index, and the single-scattering albedo of aerosols observed with Raman lidars and Sun photometers in East Asia.
Table 3. Microphysical Parameters Derived From the Inversion of the Lidar Data
4.2.1. Vertical Variation on 14 June and 30 October 2004
 Before we move to a detailed discussion on the results of the microphysical properties according to aerosol types, we briefly discuss some features of the vertical variation of microphysical parameters. We selected two days measurement cases; 14 June and 30 October 2004. The aerosol layers on these two days were detected to high altitude (>3.5 km), in contrast to the other measurement cases where we detected particles only to comparably low altitude. The origin of the air mass on these two days varied with altitude, as explained in section 3.1.
 The lidar ratio at 532 nm increases with height from 60 sr to 65 sr and from 60 sr to 90 sr on 14 June and 30 October, respectively. The lidar ratio depends on the absorption properties of the particles, and their mean size [Franke et al., 2003; Müller et al., 2003].
 Effective radii were approximately 0.3–0.35 μm within the PBL on 14 June and 30 October, and thus larger than the particles above the PBL. We find values of approximately 0.27 μm on 14 June and 30 October. The single-scattering albedo decreases with altitude from 0.98 to 0.92 and from 0.97 to 0.85 on 14 June and 30 October, respectively. These variations with height likely result from the different sources of the aerosols.
 Atmospheric aerosols have different chemical composition depending on their source regions. Kim et al.  report that mass concentrations of particles transported from east and north China are higher than mass concentrations in air that has its origin from within the Korean peninsula. The authors furthermore point out that particles with a high contribution of ionic and carbon components, especially black carbon, were identified when the air mass was long-range transported from China. According to FLEXPART and HYSPLIT results, the air masses that arrived above the PBL over our measurement site had predominantly crossed forest fire smoke regions over Siberia and Mongolia on 14 June, and had passed over industrial regions in China on 30 October; see Figure 2. In contrast, the air mass that we observed within the PBL likely originated from inside the Korean peninsula.
 Relative humidity is another reason for the change of particle size and single-scattering albedo with altitude. We found high relative humidity (>65%) only at low altitude (within the PBL) on both days. In contrast, the air mass above the PBL showed only a low relative humidity (<53%). From our data we find that the light-absorbing capacity (in terms of single-scattering albedo) decreases with increasing relative humidity. However, our data set is quite limited. We have no information on the mixing state of the particles (internal versus external mixing of aerosol types). We do not know if and to what degree absorbing particles (soot core) were coated with nonabsorbing, hygroscopic material. We have no chemical analysis of pollution particles. For these reasons we skip any further going interpretation of our results. In a future study we will specifically deal with the hygroscopic properties of pollution over our lidar site.
4.2.2. Particle Size
 The effective radius varies from 0.24 to 0.35 μm for the measurements at Gwangju. These values vary according to the origin of the observed aerosol particles. Particle effective radii vary between 0.3 and 0.35 μm for local/regional haze measured at Gwangju within the PBL (13 and 14 June and 30 October 2004). We find slightly lower effective radii of 0.24–0.28 μm for the long-range transported haze advected from China. Smoke particles transported from Siberia are in about the same size range (0.27 μm).
 As for the results representing the measurements at Anmyeon Island we find mean effective radii of 0.2 ± 0.02 μm and 0.24 ± 0.02 μm for long-range transported haze and regional/local haze, respectively. We think that the effect by sea salt aerosol can be neglected because we analyzed data that represent the situation above the marine boundary layer. We assume three possible reasons for the lower effective radii particular for the regional haze situation at Anmyeon, compared to the Gwangju results: (1) relative humidity was lower during the measurements at Anmyeon (24–51%) compared to the conditions over Gwangju, (2) Anmyeon is much less affected by local aerosol sources like biomass burning of agricultural waste, and (3) Anymeon may always be influenced by urban haze from the Seoul metropolitan area.
Figure 7 shows that particle effective radius increases with decreasing Ångström exponent. The correlation describes the well-known property that the extinction coefficient, which determines the Ångström exponent, is mainly influenced by particle size. A simple linear regression fit was applied. We find a correlation coefficient of approximately 0.62 if we do not distinguish among the aerosol types studied in this contribution.
 The effective radius of the long-range transported haze observed over Gwangju is considerably larger than the particle size of anthropogenic (urban/industrial) pollution from North America and Europe for which we find values around or even less than 0.2 μm [Müller et al., 2004, 2005]. Our numbers are similar to or slightly higher than effective radii of mixtures of urban/industrial/biomass burning haze observed in the Pearl River Delta, (0.24 ± 0.07 μm) and urban/industrial emissions over Beijing (0.23 ± 0.06 μm) in China [Müller et al., 2006]. In view of the retrieval uncertainties we believe that the differences for long-range transported haze are not significant. In contrast, there is a notable difference to regional/local haze over Korea, which may be strongly affected by the burning of agricultural waste.
 The smoke particles that we observed above the planetary boundary layer on 14 June 2004 are smaller than effective radii of aged, long-range transported forest fire smoke observed over Central Europe [Müller et al., 2005]. Reason for the lower effective radii may be transport time which seems to have some influence on particle growth [Müller et al., 2007]. Müller et al.  report that effective radius of forest fire smoke particles increases with duration of transport. For instance, effective radius of Siberian forest fire smoke observed over Tokyo, Japan (35.66°N, 139.80°E) was 0.22 ± 0.04 μm [Murayama et al., 2004]. Transport time of the smoke aerosols was 4 days. In our case we estimate a transport time of around 6–7 days. Particle size that we find in our study seems to be consistent with the assumption of particle growth during transport time [Müller et al., 2007].
 In Korea, open-field burning of agricultural waste after harvest is commonly practiced by many farmers [Ryu et al., 2004]. There are two typical biomass burning phenomena in Korea: one is the burning of agricultural waste after the harvesting of barley in late spring, and the other one is the burning that follows the harvesting of rice in the fall.
 If we consider the observation date of the regional/local haze in Gwangju and the low height in which the haze was observed, it is rather likely that emissions from open-field burning around Gwangju were present. We may assume that the regional/local haze that was observed over Gwangju consisted of a mixture of urban/industrial emissions and combustion products of biomass like fuel wood, rice or barley straw. The burning of biomass leads to larger particles compared with particles from diesel combustion [Venkataraman and Rao, 2001], which can explain why particles from regional/local haze differ in effective radius to long-range transported haze from urban/industrial emission from China.
 We assume that relative humidity is another reason for the relatively large particle sizes observed in our study. Figure 8 shows that the Ångström exponent on average decreases and particle effective radius increases with relative humidity. A poor correlation coefficient is found if we consider all data points regardless of the origin of the particle plumes, i.e., if we mix different aerosol types in our correlation analysis.
 We find a mean hygroscopic growth factor (f(RH)), defined as reff (RH = 85%)/reff(RH = 30%), of 1.49 ± 0.36 if we consider all measurements. We obtain growth factors of 1.50 ± 0.45 for long-range transported haze (all data point from Gwangju and Anmyeon for this aerosol type) and 1.35 ± 0.41 for regional/local haze (all data points from Gwangju and Anmyeon for this aerosol type).
 However, we have only a very low number of data points for this aerosol-type-dependent extraction of growth factors. These growth factors of 1.5 and 1.35 therefore can only be considered as a first rough estimate of the true conditions.
 In fact we have at hand a considerable higher number of measurement cases from 2004/2005 and 2009. However, the relocations of the lidar between Gwangju and Anmyeon caused significant changes in instrument performance. Continuous upgrades and redesigns of the instrument also had significant impact on instrument performance. On the one hand we could not keep a continuous time series of lidar observations after 2005. On the other hand the analysis of the existing data is extremely time consuming. Thus a more statistical analysis of our data has to be left for a future contribution.
 For comparison, urban/industrial aerosols advected from China show a growth factor of 2.75 ± 0.38 [Kim et al., 2006]. European anthropogenic urban/industrial aerosol pollution observed during ACE-2 [Carrico et al., 2000] shows a growth factor of 1.46 ± 0.10. Growth factors of 1.81 ± 0.37 to 2.30 ± 0.24 are found for urban/industrial aerosols observed at the U.S. East Coast during TARFOX [Kotchenruther et al., 1999]. Kim et al.  report on f(RH) values of 1.91 ± 0.16 for particles generated over Korea In contrast, smoke particles emitted from forest fires near Lake Baikal and in North Korea show a rather low f(RH) of 1.60 ± 0.20 [Kim et al., 2006].
4.2.3. Real Part of Complex Refractive Index
 The mean values of the real part of the refractive index are rather similar for the different aerosol types. We find values between 1.37 ± 0.02 to 1.44 ± 0.05. The numbers for these aerosol types are on average lower than what has been reported for similar aerosol types observed in other places. Wandinger et al.  report on higher real parts for biomass burning particles and continental European pollution observed over Central Europe after long-range transport from Canada (Lindenberg Aerosol Characterization Experiment (LACE 98), Germany 1998). Values range from 1.56 to 1.66 for biomass burning aerosol and from 1.48 to 1.58 for continental European pollution. Müller et al.  report on real parts of 1.47 ± 0.07 for forest fire smoke particles and 1.52 ± 0.01 for anthropogenic (urban/industrial) emissions advected from North America to Central Europe. Real parts of 1.48–1.56 were reported for continental European pollution observed during the Second Aerosol Characterization Experiment 2 (ACE 2) [Müller et al., 2002].
Dubovik et al.  report on aerosol properties derived from 8 years of observations with AERONET Sun photometer [Holben et al., 1998]. Real parts for biomass burning smoke range from an average of 1.47 for Amazonian forest region smoke to 1.52 for South American cerrado smoke. These numbers are higher than real parts for urban/industrial aerosols. The authors find a value of 1.39 for aerosols at the east coast of the United States, 1.4 for Creteil in France, 1.47 for Mexico City, and 1.44 for the Maldives (Indian Ocean).
 The variation of the real part of the complex refractive index is affected by relative humidity because of the uptake of water. von Hoyningen-Huene et al.  report on increasing real parts with decreasing relative humidity. Their results rest upon measurements carried out under different relative humidity conditions. Dubovik et al.  state that low real parts of the refractive index are likely associated with high relative humidity resulting in aerosol hygroscopic growth.
 The low values of the real part obtained in our research work might also be caused by the uptake of water vapor, though the quality of our study suffers a bit from the fact that we did not observe aerosol conditions with relative humidity above 80%. Noh et al.  report comparably low values of the real part of the refractive index of 1.44 ± 0.01 and 1.41 ± 0.01 for urban/industrial haze (long-ranged transported from China) and smoke aerosols, respectively, at high relative humidity (70–90%). The data describe lidar observations in the height range from 0.78 to 1.62 km (within the planetary boundary layer) at the Gwangju site. The values from that previous study agree well to the findings discussed in this contribution.
 We find comparably low values of the real part of the refractive index of 1.41 ± 0.02 (at Gwangju) and 1.37 ± 0.02 (at Anmyeon) for regional/local haze at a relative humidity of 69 ± 7% and 61 ± 15%, respectively. In contrast we find slightly higher values of 1.44 ± 0.05 and 1.43 ± 0.04 for long-range transported haze at Gwangju and Anmyeon island, respectively, at lower relative humidities (<45%).
4.2.4. Light-Absorption Properties
 Our results for single-scattering albedo are summarized in Table 3. There exist differences of the light-absorbing properties according to the source regions of the observed particles. We find comparably low single-scattering albedos of 0.85–0.89 (at Gwangju) and 0.9–0.92 (at Anmyeon) for long-range transported haze that originated from urban and industrial regions in China.
 In contrast we find higher values of 0.96–0.98 for regional/local haze at Gwangju and 0.93–0.96 for regional/local haze observed at Anmyeon island. The single-scattering albedo of regional/local haze at Anmyeon island is slightly lower than at Gwangju. We assume that aerosols, that originated from the Seoul metropolitan area, had some impact on the formation of the regional/local haze at Anmyeon island. According to a recently released emission inventory (D. Streets, emission data, 2006, http://www.cgrer.uiowa.edu/EMISSION_DATA_new/index_16.html), the black carbon emissions in Seoul and its surrounding region are as high as 3048 tons per year per 0.5° grid cell. For comparison, the black carbon emissions are 866 tons per year per 0.5° cell for Gwangju and its surrounding area. A moderately low single-scattering albedo of 0.92 is found for smoke aerosols.
Figure 9 shows the correlation between single-scattering albedo and the imaginary part of the complex refractive index. Both parameters represent the average conditions within the measurement wavelength range of the Nd:YAG laser. We find a very strong correlation of the single-scattering albedo with the imaginary part. The correlation coefficient is 0.91.
Table 4 lists single-scattering albedos of aerosols in other regions of East Asia. Figure 10 displays the single-scattering albedo discussed in the following text. We may separate the following numbers for single-scattering albedo according to observations near the source regions and observations downwind of the source regions. For instance, single-scattering albedo was determined for aerosols observed at various locations in China, including urban and suburban areas, on the basis of short-term in situ measurements and ground-based remote sensing.
 A single-scattering albedo (at 532 nm) of 0.77 ± 0.12 and 0.78 ± 0.11 was retrieved on the basis of combined one-wavelength Raman lidar and Sun photometer observations in the Pearl River Delta (South China) in October 2004, and in Beijing (North China) in January 2005, respectively [Müller et al., 2006; Tesche et al., 2008]. An average midvisible (550 nm) single-scattering albedo of 0.89 was derived from Sun photometer measurements at two urban sites (Beijing and Julin) in China [Eck et al., 2005]. Andreae et al.  report on an average single-scattering albedo of 0.86 at ambient atmospheric conditions in the urban area of Guangzhou (eastern part of China). Qiu et al.  find a single-scattering albedo of 0.80–0.85 for six north Chinese cities from 9 years of routine pyrheliometer and pyranometer observations. A nationwide mean value of 0.89 ± 0.04 for China is reported by Lee et al. . That number was derived from a combination of ground-based spectral transmittance and spaceborne top-of-the-atmosphere reflectance observations. The enormous amount of soot emitted over China may explain the low single-scattering albedo. According to an emission inventory (D. Streets, emission data, 2006, http://www.cgrer.uiowa.edu/EMISSION_DATA_new/index_16.html) the soot emission in China is about 1.811 × 106 tons per year, which amounts to 60% of the total soot emissions in Asia.
Table 4 shows that pollutants transported from China to downwind areas over Korea and Japan tend to have higher single-scattering albedos than pollutants close to their source regions. Noh et al.  reports a single-scattering albedo from 0.87 to 0.93 (average 0.90 ± 0.03) for anthropogenic aerosols that originated from eastern and northern China. The numbers describe multiwavelength Raman lidar measurements of aerosols in the upper PBL and the free troposphere. Single-scattering albedos were measured with in situ instrumentation aboard ships cruising along the coasts of Korea and Japan during ACE-Asia (spring 2001). A mean value of 0.94 ± 0.03 at a relative humidity of 71 ± 13% [Carrico et al., 2003] was found. Though not explicitly mentioned by the authors, sea salt may have impacted the measurements to some extent.
 The single-scattering albedo of particles over the Chinese mainland is on average similar or slightly lower (approximately 0.05) than the single-scattering albedo that we find for urban/industrial haze advected from China to our field site, considering the uncertainties in our data analysis. Possible reason for the somewhat higher single-scattering albedos (compared to the results for the Chinese mainland) may be the reduction of the light-absorption capacity as a result of the aging process (secondary aerosol formation) during the transport of aerosols from China to Korea and Japan. Cheng et al.  report that aerosol single-scattering albedo may increase very fast as the result of rapid secondary particle formation and condensation. Observations in highly polluted regions in northeastern China serve as a database.
 We note two significant outliers in the numbers in Table 4. Single-scattering albedos that describe the conditions found with lidar/Sun photometer in the Pearl River Delta and over Beijing are considerably lower than any other values reported here. We point out that the mean values of single-scattering albedo of these two cases show good agreement to results from ground-based in situ observations that were carried out at the site of the lidar/Sun photometer observations. The high error bars are caused by the limited optical information that was available for the retrieval of single-scattering albedo. Detailed information about the retrieval method that was used for these data is explained by Tesche et al. .
 The lower values of single-scattering albedo in the Pearl River Delta and over Beijing may be caused by the fact that relatively fresh particles were observed in these two areas. Next to aging effects during transport time, particle size may be another reason for the significantly lower single-scattering albedo, which was derived from these combined lidar/Sun photometer observations. Particles over the Pearl River Delta on average are larger than what we report in our present study. For fixed imaginary part of the complex refractive index a larger particle size results in a lower single-scattering albedo.
 The highest value of single-scattering albedo is found for regional/local haze over Korea. We find an average value of 0.97 ± 0.01. We assume that this high single-scattering albedo may be caused by comparably high relative humidity conditions in combination with the type of burned waste. The temperatures during the burning of the waste may have been comparably low so that mostly nonabsorbing smoke (white smoke from smoldering fires) was produced, which would also be in accordance with the relatively low lidar ratios.
Carrico et al.  find a pronounced correlation between changing relative humidity and single-scattering albedo for anthropogenic particles that are transported from the Asian continent. The main composition of these anthropogenic particles is a mixture of nss-SO42−, NO3−, NH4+ and carbon. The measurements were performed with in situ instruments aboard ship near Korea and Japan during ACE-Asia (spring 2001). The authors report an increase of single-scattering albedo from 0.91 at relative humidity 40% to 0.96 at a relative humidity of 85%. In contrast to in situ or column-integrated measurements carried out at the surface, aircraft measurements performed during ACE-Asia in coastal areas of Korea and Japan show a low single-scattering albedo of 0.88 ± 0.03 at low relative humidity [Anderson et al., 2003]. Values for relative humidity are not given by Anderson et al. .
 Other reasons for the differences in the single-scattering albedo of regional/local haze to single-scattering albedo of long-range transported haze may be a different mixture of different fuel types (crop waste versus fossil fuel) and differences in the combustion. The burning of fuel at high temperature generates on average a higher concentration of light-absorbing soot. Combustion at lower temperatures (like the burning of agricultural waste) results in so-called smoldering fires, which tends to give low light-absorbing particles and thus a higher single-scattering albedo. Pollution control technology may also have some impact on the light-absorption capacity of regionally produced haze [Eck et al., 2005].
 We find a mean single-scattering albedo of 0.92 for smoke particles, which is higher than the single-scattering albedo of long-range transported haze. Our single-scattering albedo is at the lower end of numbers that are usually reported for forest fire smoke. In view of the transport height of the air, as shown by the HYSPLIT trajectories (see Figure 2d), we believe that the smoke particles were not contaminated with urban/industrial haze.
Müller et al.  report on single-scattering albedos of Siberian biomass burning plumes that were observed with Raman lidar over Leipzig, Germany, in May 2003. The authors find a mean single-scattering albedo of 0.93 ± 0.03 at 532 nm. These values are higher than what has usually been observed near the sources of fires [Hobbs et al., 1997; Eck et al., 2001, 2009]. Condensation of nonabsorbing gases on the smoke particles, as well as particle aging effects during long-range transport may have been the reason for the comparably high single-scattering albedo [Müller et al., 2005]. This value is lower than for smoke particles measured under conditions of high relative humidity in Korea [Noh et al., 2009] and Japan [Murayama et al., 2004]. Noh et al.  report on a single-scattering albedo of 0.96 ± 0.02 at a high relative humidity of 80–90% for smoke particles transported from Siberia to Korea in October 2005. Murayama et al.  find a similar value of 0.95 ± 0.06 at relative humidities of 60–80% for Siberian smoke observed in May 2003.
 Optical and microphysical properties were retrieved from several days of multiwavelength Raman lidar measurements in Korea. We studied the light-absorbing characteristics of aerosol layers according to their source regions and the impact of relative humidity on particle size and single-scattering albedo. The observed aerosol layers were classified according to their source regions as long-range transported haze (anthropogenic pollution particles generated from urbanized and industrialized regions in the eastern parts of China), smoke (Siberia and northern China), and regional/local haze particles (a mixture of anthropogenic particles emitted from urban and industrial regions with emission from agricultural activity, i.e., burning of crop waste around our measurement sites). The identification of the source regions was made with HYSPLIT backward trajectory analysis and FLEXPART simulations. From our study, we infer that aerosols that are transported from China to Korea in general have a higher light-absorbing capacity than aerosols produced in Korea. In that latter case the burning of agricultural waste, which is a main source of pollution during spring and autumn, produces aerosols with comparably low light-absorption capacity.
 The single-scattering albedos can be distinguished according to their source regions. The lowest single-scattering albedo value was retrieved for long-range transported haze. We find a mean value of 0.87 ± 0.02 (representative for the wavelength at 532 nm). Moderately light-absorbing capacity is found for smoke particles. The single-scattering albedo is approximately 0.92. Regional/local haze particles show the highest single-scattering albedo of 0.97 ± 0.01 at 532 nm. We assume that differences in fuel types, combustion methods, and pollution control technology may be some of the reasons for the observed differences between long-range transported haze and regional/local haze. Another possible factor in these differences may be relative humidity.
 Highest effective radii of 0.32 ± 0.02 μm are found for regional/local haze particles. Smoke and long-range transported haze show intermediate values of 0.27 and 0.26 ± 0.02 μm, respectively. These particle types are slightly larger than the particle sizes observed at the source regions. We assume that particles in the smoke and long-range transported haze plumes grow slightly during transportation as a result of the aging effect (external mixture and/or coated core) and hygroscopic growth at moderately to high relative humidity. In contrast, locally produced haze particles seem to be mainly affected by hygroscopic growth under high relative humidity conditions (60–80%) and the specific fuel type (crop waste). We find that the light-absorbing capacity of regional/local haze decreases as a result of hygroscopic growth. The other two types of aerosols considered in our study were not significantly affected by relative humidity.
 This work was funded by the Korea Meteorological Administration Research and Development Program under grant CATER 2009-3112. The authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model and/or READY website (http://www.arl.noaa.gov/ready.html) used in this publication. The wind fields for the FLEXPART simulations are from the Research Data Archive (RDA), which is maintained by the Computational and Information Systems Laboratory (CISL) at the National Center for Atmospheric Research (NCAR). NCAR is sponsored by the National Science Foundation (NSF). The original data are available from the RDA (http://dss.ucar.edu) in data set ds083.2.