Lidar and sky radiometer systems at Sapporo, Toyama, and Nagasaki, Japan, observed the vertical distributions and optical properties of (nonspherical) dust and spherical aerosol particles from March to May 2005 as part of the Atmospheric Brown Clouds–East Asia Regional Experiment 2005 (ABC-EAREX2005). Sky radiometer observations suggest that single scattering albedo at Nagasaki was smaller than that at Toyama and Sapporo. Relationships between the single scattering albedo and Ångstrom exponent suggest that aerosol particles observed at Toyama in March differed from those observed in April and May. In contrast, at Nagasaki, there was no obvious difference in aerosol particles among the 3 months. Aerosol optical thicknesses observed by the sky radiometer resemble the aerosol optical thicknesses observed by lidar and simulated by the Chemical Weather Forecasting System (CFORS) model. A new parameter that describes the aerosol vertical distribution, Hm, is the modified scale height of the extinction coefficient. Hm can be used as an index of the vertical aerosol extent even if the detailed structure of the vertical profile cannot be shown. Hm observed by lidar was consistent with Hm simulated by the CFORS. Relationships between the aerosol optical thickness and Hm obtained by lidar measurements and CFORS simulations suggest that dust aerosol particles are generally transported over Japan at higher altitudes than are spherical aerosol particles.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Growth of the East Asian economy has changed both social systems and the environment, with environmental changes including increases in the aerosol particles and greenhouse gases emitted into the atmosphere [Streets et al., 2000; Streets and Waldhoff, 2000]. Pyrheliometer measurements of broadband direct solar radiation [Luo et al., 2001] showed increases in the aerosol optical thickness for 1961–1990 over China. Recent satellite- and ground-based measurements have also shown that aerosol loading in this region is quite large compared to global averages. The aerosol optical thickness sometimes exceeds 1.0 at the wavelength of 500 nm [Qiu and Yang, 2000; Chu et al., 2003; Eck et al., 2005]. Observed aerosol particles over East Asia in springtime are often a mixture of polluted air particles transported from urban or industrial areas over western China and mineral dust particles transported from arid regions of northwestern China. These aerosol particles often affect the shortwave radiation budget at the surface [Nakajima et al., 2003].
 The Atmospheric Brown Clouds (ABC) project was initiated by the United Nations Environment Programme (UNEP) to study the impact of air pollution manifest as haze, smog, and acid deposition on climate, water resources, agriculture, and health. The ABC-East Asia Regional Experiment 2005 (ABC-EAREX2005) was conducted as part of ABC project activities with collaboration of Japan and Korea. The intensive field observations of aerosol particles and radiation took place mainly at Gosan, Cheju Island, Korea, in spring 2005 (T. Nakajima et al., An overview of the ABC EAREX 2005 regional experiment, submitted to Journal of Geophysical Research, 2007). During ABC-EAREX2005, measurements of aerosol particles and radiation were also taken at several SKYNET stations over Japan.
 This paper examines the vertical distributions and optical properties of aerosol particles during spring as measured by lidar and sky radiometer instruments at Sapporo, Toyama, and Nagasaki, Japan, during the ABC-EAREX2005 period. Vertical distributions of aerosol particles near this region have been studied in the past using data from lidar and/or aircraft-based instruments [Sugimoto et al., 2002, 2003; Huebert et al., 2003]. However, past studies have focused on short-term or event-like aerosol phenomena. This study introduces a new parameter to describe the vertical distribution of aerosol particles in analyses of lidar data that include a large amount of data. Observed aerosol particle properties are also discussed using output from an atmospheric chemical transport model.
2. Observations and Data Analysis
Figure 1 shows the locations of Sapporo (43.08°N, 141.34°E), Toyama (36.70°N, 137.65°E), and Nagasaki (32.78°N, 129.87°E) along the Sea of Japan. Sapporo and Nagasaki are separated by about 1500 km. Most of the air masses observed at these stations in springtime have been transported from the Asian continent [Sun et al., 2001; Koike et al., 2003; Liu et al., 2003]. Lidar and sky radiometer instruments recorded measurements at the three observation sites. Sky radiometer measurements were taken in daytime only, whereas lidar measurements were taken throughout the day. The instruments operated automatically, and data were transferred to the National Institute for Environmental Studies and the Center for Environmental Remote Sensing at Chiba University. Data obtained under clear-sky conditions in March, April, and May 2005 were analyzed. We used sky radiometer data to judge the clear sky condition. The data used in this study have a coincidence between measured and calculated sky radiances with an error less than 1%. In most of cirrus cases, spatial variability of sky radiance is much larger than clear sky condition although there might remain homogeneous cirrus condition.
 The lidar system used in this study automatically measured vertical profiles of backscattering intensity at 532 nm and 1064 nm. The depolarization ratio at 532 nm was also measured. The output laser power at each wavelength was 20 mJ with a repeat rate of 10 pps, and the receiver was a Cassegrain telescope with a 20-cm diameter mirror. The vertical spatial resolution and measurement cycle were 6 m and 15 min (5-min measurements and 10-min rests), respectively. This study used data that were 30-m averages in the vertical. The lidar measured vertical profiles of aerosol particles in the entire troposphere, but depolarization ratio measurements reached only to about 6 km. Consequently, the lidar data analysis was limited to 6 km in this study [Shimizu et al., 2004; Sugimoto et al., 2005].
 Depolarization ratio measurements were used to distinguish between spherical and dust (nonspherical) particles. The total (dust + spherical) extinction coefficient was retrieved from backscattering intensity by applying Fernald's method [Fernald, 1984] with a ratio of extinction to backscattering coefficients of aerosol particles, S1 = 50. The total was then partitioned into a dust extinction coefficient and spherical extinction coefficient [Sugimoto et al., 2003, 2005; Shimizu et al., 2004]. The ratio R of the dust extinction coefficient to the total coefficient was obtained from the following equation:
where δ is the observed depolarization ratio, and δ1 and δ2 are depolarization ratios of dust and spherical aerosol particles, respectively, such that δ1 = 0.35 and δ2 = 0.05 following Sugimoto et al. [2003, 2005]. There are various shapes in the aerosol particles; that is, pure dust particles have a range of depolarization ratio. Therefore the measured depolarization ratio exceeds the range 0.05–0.35 which are empirically determined and detailed discussions are shown by Shimizu et al. .
 Because of the large amount of lidar data, a scale height H0 was used to measure the vertical distribution of aerosol particles [Hayasaka et al., 1998]. Such a scale height is generally obtained by dividing the optical thickness τa (i.e., the vertically integrated extinction coefficient) by the extinction coefficient at the surface β0. This relationship is correct when the vertical profile of the aerosol extinction coefficient β(z) can be expressed as the following function of height z:
The lidar system in this study, however, could not measure β0 exactly because the optical axes of the transmitter and receiver were separated. The transmitter and receiver were aligned side by side, and the receiving field of view was limited. It was therefore difficult to measure correctly the extinction coefficient at the lowest layers, such as 100 m, even if the overlapping correction was done. In addition, aerosol vertical profiles sometimes differ from the extinction coefficient profile as shown by equation (2).
 A modified scale height Hm of the aerosol extinction coefficient can be used to overcome the above defect. Hm is defined as
Hm is exactly equivalent to H0 if the vertical profile of the extinction coefficient is expressed by equation (2). The implication is that Hm is the height below which 63.2% of the total columnar aerosol extinction coefficient is present, or the height at which the aerosol optical thickness from the top of atmosphere becomes exp(−1). Use of Hm instead of H0 robustly parameterized the vertical distribution even if the measure of the extinction coefficient at the surface contained errors and the vertical profile differed from equation (2).
 The black line in Figure 2 shows a lidar-observed vertical profile of the aerosol extinction coefficient at Toyama on 22 April. The aerosol extinction coefficient up to 90 m in this case is about 0.07 (1/km), but this value might include large errors. The scale height H0 derived from the extinction coefficient is 2.51 km. The vertical profile derived from this scale height H0 and equation (2) is shown by the red line in Figure 2. In contrast, the modified scale height Hm is 1.23 km. The vertical profile derived from this Hm and equation (2) with H0 = Hm is shown by the blue line.
Figure 3 compares the frequency distribution between Hm and H0 for lidar data in this study. H0 extends to higher than 10 km (the uppermost height is limited to 10 km in Figure 3), even though the lidar data analysis in this study was limited to 6 km. The H0 values higher than 6 km appear to be a virtual index. In contrast, Hm gives a more appropriate measure of the vertical profile of aerosol extinction. Differences between the observed vertical profile and the profile calculated with equation (2) and H0 as shown in Figure 2 were likely caused by inappropriate values for extinction coefficient β0 at the surface. The maximum value of the extinction coefficient sometimes occurs at higher altitudes, and more than one maximum may be observed, although the aerosol concentration is generally larger in the lower atmosphere than in the upper atmosphere. Although Hm may not always present the details of the aerosol vertical profile, it is nevertheless more appropriate than H0 as an index of the vertical profile of the aerosol extinction coefficient.
2.2. Sky Radiometer
 A sky radiometer (PREDE Co. Ltd., POM-01) measured both direct and scattered solar radiation at wavelengths of 315, 400, 500, 675, 870, 940, and 1020 nm. The scattering phase function and extinction coefficient of aerosol particles were determined by an integral equation of the size distribution and efficiency factor; thus aerosol particle size distributions were derived through the solution of an integral equation, i.e., by an inversion. The aerosol particle size distribution, optical thickness, Ångström exponent, and single scattering albedo in an air column were determined so as to be consistent with the observed direct and scattered solar radiation intensity at respective wavelengths, except for 315 nm and 940 nm, incorporating multiple scattering processes [Nakajima et al., 1983, 1986; Aoki and Fujiyoshi, 2003]. We assumed spherical particles in the analysis of sky radiometer data. It has been known that back scattering part of phase function is much affected by nonspherical particles [Pollack and Cuzzi, 1980; Nakajima et al., 1989]. In the actual measurement of scattered solar radiation by sky radiometer data, however, the scattering angle rarely covers backward direction such as 170° ∼ 180°. According to Dubovik et al. , retrieval of single scattering albedo in AERONET is not so much affected by an assumption of particle shape. The main purpose of this study was to investigate general aerosol properties over the entire observation period rather than to analyze each event; therefore the aerosol optical thickness, Ångström exponent, and single scattering albedo were mainly used. The Ångström exponent α relates the optical thickness to the particle size distribution. The wavelength dependence of the optical thickness is
if the number size distribution n(r) of aerosol particles can be approximated by a power law function,
where λ and r are the wavelength of light and the particle radius, respectively.
 The sky radiometer recorded measurements every 10 min. Very large values and large temporal variability of the retrieved optical thickness sometimes occurred, suggesting cloud contamination. Therefore the minimum value of the aerosol optical thickness in each hour was used if the variability was large. Nevertheless, some contamination caused by optically thin cirrus clouds may have been included in the data used in this study.
3. Chemical Transport Model Analysis
 The Chemical Weather Forecasting System (CFORS [Uno et al., 2003]) was used to synthesize the data obtained in this study. The CFORS is a regional-scale multitracer chemical transport model within the Regional Atmospheric Modeling System (RAMS [Pielke et al., 1992]) model. The CFORS simulates time and space variations of typical gas and aerosol components in the troposphere, including SO2/sulfate, black and organic carbon, mineral dust, and sea salt. Transport processes in the model include emission, advection, diffusion, dry/wet deposition, gravitational settling, and chemical reaction/conversion. Gravitational settling is applied to mineral dust and sea salt aerosol particles. In the present simulation, all anthropogenic emissions and biomass burning emissions inventories were calculated using a year 2000-based emission data set at 1° resolution [Streets et al., 2003]. A simple vertical dust deflation scheme proposed by Gillette and Passi  was used for natural dust emission. The dust source region was defined using the United States Geological Survey (USGS) land cover data set, and dust emission flux was calculated using a fourth-power law function of the simulated surface friction velocity in 12 bins for radii from 0.1 to 20 μm. Sea salt emission flux was calculated using the empirical relationship from Gong et al.  in 12 bins for radii from 0.005 to 20.48 μm. Sea salt transport was simulated with a two-bin model that considered a fine mode, for particles with radii between 0.005 to 2.56 μm, and a coarse mode, for particles with radii between 2.56 and 20.48 μm. The present simulation also included calculations of the aerosol optical thickness, extinction coefficient, and single scattering albedo at wavelengths of 550 and 860 nm.
 The CFORS simulation covered the period from 20 February to 1 June 2005 and included four-dimensional data assimilations based on NOAA National Center for Environmental Prediction (NCEP) global reanalysis data (6-h intervals with 2.5° resolution). The model domain was a 120 × 90 grid that covered East Asia with a resolution of 60 km. The model top was at 20 km. There were 40 nonuniform vertical layers in the terrain following the sigma-z coordinate. More detailed descriptions of the CFORS and its performance, and evaluation of aerosol optical parameters, can be found in reports by Uno et al.  and Satake et al. , respectively.
4. Results and Discussion
4.1. Aerosol Optical Thickness, Ångström Exponent, and Single Scattering Albedo
 Relationships of the aerosol optical thickness (AOT), Ångström exponent (AE), and single scattering albedo (SSA) were investigated using sky radiometer data to reveal general aspects of aerosol properties in the air column. Figure 4 shows scatter diagrams of AOT and AE. Blue, red, and green dots in each diagram indicate results from March, April, and May, respectively. At Toyama, AOT values larger than 1.0 were sometimes observed. Such large AOT values were not observed at Sapporo. Because the sky radiometer operations were automatic, cloud contaminations might affect AOT retrieval. At Sapporo, AE was smaller in May than in other months. In addition, AE values at Sapporo ranged from less than 0.5 to more than 2.0. AE values did not show this behavior at Toyama and Nagasaki, where there were few AE values exceeding 1.5. Small AOT with small AE was frequently observed at both Toyama and Nagasaki. Those aerosols might be mineral dust or sea-salt particles. Aerosol particles with large AOT and large AE, however, such as those found at Toyama in March and May and at Nagasaki in April and May, might be air pollution particles.
Figure 5 shows scatter diagrams of AE and SSA. SSA at Nagasaki was smaller than at the other two locations. Small SSA with large AE in March at Toyama may reflect polluted aerosol particles there, while small SSA was found with rather small AE at Sapporo and Nagasaki. Small SSA corresponded to small AE particularly at Sapporo, although the number of data points was limited. Small SSA at Sapporo resulted from dust aerosol particles, whereas that at Nagasaki resulted from a mixture of dust and polluted aerosol particles.
4.2. Comparison of Observed and Simulated Aerosol Optical Thickness and Modified Scale Height
 Lidar measurements in this study include the extinction coefficients of dust and spherical aerosol particles as a function of time and height. Figure 6 shows a time-height cross section of the aerosol extinction coefficient in April at Toyama for dust and spherical particles. Gray-shaded areas indicate regions where data were not analyzed because clouds and/or rain were present. White denotes areas where the aerosol extinction coefficient was less than 0.02 (1/km). Figure 6 shows that the vertical extent of dust aerosol particles had large variability. Spherical aerosol particles, in contrast, were generally loaded lower than about 3 km. Figure 6 shows typical results for lidar data, but data were limited to 1 month. Such a presentation has limitations for cases with large data amounts such as long-term measurements at many locations.
 The AOT is an important and convenient index of the integrated value of the extinction coefficient and is related to columnar aerosol loading. AOT is derived from lidar and sky radiometer measurements and from output from the CFORS simulation, and it is simple to intercompare AOT values. Figure 7 shows time series of AOT for total (Figure 7, top), dust (Figure 7, middle), and spherical (Figure 7, bottom) aerosol particles obtained by lidar and sky radiometer measurements, and also AOT calculated by the CFORS corresponding to Figure 6. Blue dots show AOT derived from lidar measurements, red open circles show AOT derived from sky radiometer measurements, and the black line shows results calculated from CFORS output. Sky radiometer measurements do not obtain both dust AOT and spherical AOT; thus sky radiometer measurements of AOT are for total AOT only. Although scavenging of aerosol particles by clouds was considered in the CFORS simulation, aerosol particles that remained in the clouds could be used to evaluate AOT. Thus AOT simulated by the CFORS was plotted as a continuous time series, whereas lidar and sky radiometer measurements were obtained for clear-sky conditions only.
Figure 7 shows that lidar measurements of AOT values were generally consistent with sky radiometer measurements, and both were consistent with CFORS calculations except around the middle of the month. The agreement of total AOT between measured and simulated values is better than the agreement for individual dust and spherical AOTs. The better agreement may be attributed to the discrimination method between dust and spherical aerosol particles; this method uses equation (1) and the depolarization ratio of the lidar signal after the retrieval of the total extinction coefficient [Sugimoto et al., 2003, 2005]. Dust aerosol particles simulated by the CFORS are only mineral dust particles generated in desert or arid regions. In reality, nonspherical aerosol particles measured by lidar as dust aerosol particles may be composed not only of pure mineral dust particles but also of a mixture of various components. Therefore the total AOT results seem better than individual dust and spherical AOTs results.
 The modified scale height Hm was introduced in section 2.1 to define the vertical distribution of aerosol particles with a single parameter. Figure 8 shows a time series of modified scale height Hm for dust (Figure 8, top) and spherical (Figure 8, bottom) particles obtained by lidar measurements and CFORS calculations corresponding to Figures 6 and 7. The Hm of dust aerosol particles is generally higher than that of spherical aerosol particles. Figure 8 also shows that Hm quantitatively presents the vertical extent of aerosol particles more distinctly than in Figure 6.
Figure 6 indicates that aerosol layers sometimes may include multiple layers. For example, Figure 9 shows vertical profiles of dust aerosol particles on 27 April and of spherical aerosol particles on 15 April obtained by lidar measurements (black solid line) and the CFORS (black dashed line). As illustrated in Figure 9a, the Hm of dust aerosol particles obtained from lidar measurements was 3.03 km; that estimated from CFORS output was 3.87 km. The blue and red lines, respectively, show vertical profiles of equation (2) with H0 = Hm and with H0 derived from AOT and the surface extinction coefficient β0. Figure 9b shows that Hm calculated from CFORS output was underestimated in this example with spherical aerosol particles. Discrepancies remain between lidar measurements and CFORS output, but the modified scale height Hm can be a useful index for the aerosol vertical profile.
 Using Hm, the long-term relationship between the AOT and the vertical extent of aerosol particles at many locations can be investigated. Figure 10 compares the AOT and Hm of the extinction coefficient retrieved from lidar measurements (Figure 10, left) and those derived from CFORS output (Figure 10, right) for 3 months at Sapporo, Toyama, and Nagasaki. Black dots show Hm for dust aerosol particles, and red circles show Hm for spherical aerosol particles. The upper limit of AOT is smaller than that of the sky radiometer because the analysis in Figure 10 is restricted to the lowest 6 km for both lidar measurements and the CFORS output. Contaminations due to high-altitude cirrus clouds can affect sky radiometer measurements only.
 As illustrated by Figure 10, the characteristics of AOT and Hm at the three locations in springtime were generally consistent whether measured by lidar or simulated by the CFORS. For example, the range of AOT was smallest at Toyama, and the minimum value of AOT was largest at Nagasaki. The Hm for dust aerosol particles was higher than that of spherical aerosol particles at all three locations. Lidar and aircraft measurements have previously shown that dust aerosol particles over Japan are loaded at higher altitudes in springtime and in other seasons [Murayama et al., 2001, 2003; Hayasaka et al., 1990, 1998]. The uppermost altitude analyzed using lidar data in this study was 6 km, but high aerosol concentrations are sometimes found in layers above 6 km. CFORS simulations suggest that the lidar-observed dust aerosol particles transported over Japan at higher altitudes are mineral dust particles that originated from desert or arid regions. In contrast, spherical particles, most of which are air pollutant aerosol particles of anthropogenic origin, are loaded over Japan at altitudes below the dust aerosol particles. The agreement between lidar measurements and CFORS simulations suggests that Hm can be used to present vertical aerosol profiles.
4.3. Horizontal Structures of Aerosol Optical Thickness and Modified Scale Height
Figure 10 suggests that the CFORS reproduced the relationship between AOT and Hm for 3 months at each location. CFORS output can also be used to calculate horizontal distributions of the relationship between AOT and Hm for dust and spherical aerosol particles. Figures 11a and 11b show 3-month (March–May 2005) averages of AOT (black lines) and Hm (color scale) for dust and spherical aerosol particles, respectively, over Asia. Hm is defined as the height from the ground surface (not from sea level) at individual locations. Although Hm can be assumed to be the height below which 63.2% of AOT is loaded, it does not contain detailed structures of the vertical stratification. There is little effect on Hm, for example, if the aerosol layer contains multiple layers or just a single layer. Nevertheless, Hm yields useful information on the aerosol layer.
Figure 11a shows that dust aerosol particles originate over western China and are transported to Japan at high altitudes. Spherical aerosol particles, in contrast, are emitted over southeast China and transported to Japan at lower altitudes, mainly below 3 km (Figure 11b). Figures 11a and 11b show that dust aerosol particles over Japan often exist at heights of at least several kilometers, but spherical aerosol particles rarely reach higher than 3 km. The horizontal structures of AOT and Hm of dust and spherical aerosol particles shown in Figures 11a and 11b likely do not occur simultaneously in most cases. The results in Figures 11a and 11b were obtained statistically as a 3-month average. While most previous studies of aerosol height have been limited to individual dust events, Figures 11a and 11b statistically show aerosol height information over a large area for 3 months. Figures 11a and 11b do not include the height of isentropic surfaces. However, a cold air mass might surround Japan when the air mass including dust aerosol particles moves in from the west, and the air mass may be lifted to a higher level under some conditions. Air masses at high altitudes over northern Japan contain high concentrations of dust aerosol particles under some meteorological conditions [Uno et al., 2004]. On the other hand, air masses including polluted aerosol particles from southeast of China moved over Japan in a lower atmospheric layer. An aerosol transport simulation for the ACE-Asia intensive observation period suggests that polluted aerosol particles can also be transported up to high altitudes [Satake et al., 2004]. However, the statistical analyses in the present analysis suggest that dust aerosol particles are generally loaded at higher altitudes than polluted aerosol particles in springtime over Japan.
 Lidar and sky radiometer observations of the vertical distributions and optical properties of dust and spherical aerosol particles were taken over Sapporo, Toyama, and Nagasaki, Japan, from March to April 2005 as part of ABC-EAREX2005. Results in this study are summarized as follows.
 1. Sky radiometer observations suggest that the SSA of aerosol particles at Nagasaki was smaller than at Toyama. The relationship between SSA and AE suggests that small SSA at Sapporo resulted from dust aerosol particles, and small SSA at Nagasaki resulted from a mixture of dust aerosol particles and polluted aerosol particles.
 2. Aerosol optical thicknesses observed by sky radiometer and lidar, and those simulated by CFORS, were consistent. The optical thicknesses of dust aerosol particles and of spherical aerosol particles were less consistent between lidar observations and the CFORS simulation, perhaps because of the method to distinguish between dust and spherical aerosol particles.
 3. This study introduces a new parameter to describe the aerosol vertical distribution, Hm. This parameter is the modified scale height of the extinction coefficient and can be used to manage large data volumes in lidar data analyses. Hm can be used as an index of vertical aerosol extent even if the detailed structures of the vertical profile are not known. Hm observed by lidar and that simulated by the CFORS were generally consistent. The relationship between AOT and Hm obtained by lidar measurements and CFORS simulations revealed that dust aerosol particles are generally transported over Japan at higher altitude than are spherical aerosol particles. Long-term or seasonal variations in the vertical extents of aerosol particles and their optical properties can be studied by applying methods similar to those in this study to elucidate the climatological behavior of aerosol particles around East Asia.
 Sky radiometer data were obtained in collaboration with T. Takamura, the Center for Environmental Remote Sensing, Chiba University, Japan, and with K. Arao, Nagasaki University. This study was supported by project 2-1 of the Research Institute for Humanity and Nature (RIHN).