Characteristics of Aerosol Properties of Haze and Yellow Sand Examined from SKYNET Measurements over East China Sea

Transcription

1 Journal September of the 2014 Meteorological Society of Japan, Vol. S. 92A, KITAKOGA 57 69, 2014 et al. 57 DOI: /jmsj.2014-A04 Characteristics of Aerosol Properties of and Yellow Sand Examined from SKYNET Measurements over East China Sea Shiho KITAKOGA, Yoko INOUE Graduate School of Humanities and Sciences, Nara Women s University, Nara, Japan Makoto KUJI Division of Natural Sciences, Faculty, Nara Women s University, Nara, Japan and Tadahiro HAYASAKA Center for Atmospheric and Oceanic Studies, Tohoku University, Sendai, Japan (Manuscript received 28 November 2013, in final form 30 May 2014) Abstract In this study, characteristics of atmospheric phenomena such as haze and yellow sand (Kosa) events were investigated in terms of aerosols by using sky radiometers, Light Detection and Ranging (LIDAR), and optical particle counter (OPC) observations at Fukue-jima and Amami-Oshima Islands from 2003 to As a result of the data analyses, we determined that aerosol properties such as loading, light absorptivity, particle size, non-sphericity, and vertical distribution showed specific features both in the atmospheric column and near the surface, depending on the atmospheric phenomena compared with normal atmospheric conditions. A specific case was clearly confirmed: the influence of limited light absorptivity dominated even during a Kosa event. In this study, it was confirmed that even if each ground-based instrument observed the phenomena with different ranges for the atmospheric column, lower layer, and surface, the retrieved aerosol properties were consistent. We demonstrated that the combined use of state-of-the-art instrumentation to detect aerosols is useful for quantitatively characterizing the atmospheric phenomena. Keywords Sky radiometer; LIDAR; OPC; haze; yellow sand; Kosa; atmospheric phenomenon 1. Introduction Aerosol is one of the most influential agents in the atmospheric environment. Recently, it has been determined that aerosols emitted from factories, vehicles, and slash-and-burn farming affect the environment Corresponding author: Makoto Kuji, Division of Natural Sciences, Faculty, Nara Women s University, Kita-uoya Nishimachi, Nara , Japan makato@ics.nara-wu.ac.jp 2014, Meteorological Society of Japan in East Asia. A severe aerosol event with haze or dust can cause visual hindrance (VH) and health problems. Accordingly, atmospheric phenomena such as haze, mist, and Kosa events accompanied by aerosols have been reported at meteorological observatories and recorded for a long time. A comprehensive study of aerosol properties in East Asia was conducted by the Asian Atmospheric Particle Environmental Change Studies (APEX; Nakajima et al. 2003; Sano et al. 2003), in which the significance of aerosols on direct and indirect radia-

2 58 Journal of the Meteorological Society of Japan Vol. 92A tive forcing in the region was revealed. Moreover, the chemical, physical, and optical characteristics of aerosols were investigated by the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) in spring 2001 (Huebert et al. 2003; Sano et al. 2003). The relationship between aerosol optical properties and shortwave direct radiative forcing was also investigated by the East Asia Regional Experiment 2005 (EAREX 2005; Nakajima et al. 2007). Ohta et al. (2013) further investigated the optical and chemical properties of aerosols at Amami-Oshima and Fukuejima Islands in spring They confirmed that the air mass blown from the East Asian continent during spring contains dust and anthropogenic particles. Although previous studies investigated the atmospheric environment in East Asia, such phenomena in terms of aerosols have not necessarily been investigated quantitatively by using mechanical or optical instrumentation. Therefore, the purpose of the present study is to characterize atmospheric phenomena including aerosols by using a sky radiometer, Light Detection and Ranging (LIDAR), and optical particle counter (OPC) observations recorded at Fukue-jima and Amami-Oshima Islands in 2003 and In this paper, Section 2 describes the observation and data analyses. In Section 3, the results and their interpretations are shown in terms of atmospheric phenomena including aerosol properties. The study s findings are summarized in the final section. Furthermore, supporting figures are provided as a supplementary file, which were omitted to avoid redundancy in the main text. 2. Observation and data analyses Among Sky Radiometer Network (SKYNET) sites, Fukue-jima (32.8 N, E) and Amami-Oshima Islands (28.4 N, E) were selected because they are located in the lee of the continent, and there are few local pollution sources (Fig. 1). Fig.ure 1 clearly shows that these two sites are suitable for monitoring atmospheric phenomena around the eastern edge of the East China Sea. In this study, four types of data including meteorological observation, sky radiometer, LIDAR, and OPC measurements were analyzed. Meteorological observation by the Japan Meteorological Agency (JMA) is routinely conducted in 3 h intervals to identify visibility, cloud amount, and weather, and to continuously detect atmospheric phenomena such as haze and Kosa. Visual observation can be compared to mechanical observation with satellite monitoring to obtain semi-quantitative results. However, visibility observation can allow us to comprehend atmospheric turbidity and provides long-term data worldwide over comparatively broad areas near observation sites. But, mechanical sensors such as the sky radiometer, LIDAR, and OPC have specific measurement ranges and cover all layers (i.e., total column), lower layers, and the ground vicinity, respectively. Correlation analyses were conducted with the retrieved aerosol optical and microphysical properties at 9:00, 12:00, and 15:00 JST (UTC + 9) when meteorological data were available. The measurements with the sky radiometer and OPC were synchronized with the meteorological observation within 5 min, whereas the LIDAR measurement was fully synchronized with the meteorological observation. Cloud screening based on the visual observation of cloud amount is routinely conducted by JMA at 3 h interval. The fine weather conditions were selected with cloud amounts of not more than unity out of 10. The criteria were applied to all optical measurements as a result of the correlation procedures. Therefore, the main purpose of this study is to characterize the atmospheric phenomena with VHs under fine weather. The yield of fine weather was eventually approximately 10.0 %, and was estimated as follows: The number of cloud amount observations was three times per day at 9:00, 12:00, and 15:00 JST. Thus, the total number of observations from 2003 to 2004 at Fukue and Amami conducted over 731 days was Fine weather occurred 437 times during the period. Accordingly, the yield of the fine weather was approximately 10.0 %. Furthermore, the correlation procedures were applied to the optical measurements such as sky radiometer, LIDAR, and OPC with the fine weather. As a result, approximately 80 and several cases under normal conditions and atmospheric phenomena, respectively, ultimately remained. 2.1 Atmospheric phenomena As is defined by JMA (1998), haze is a phenomenon in which subvisible dry, small particles are suspended in the atmosphere. The sky appears milky because of the large number of particles. Moreover, visibility is less than 10 km, denoting VH, and the relative humidity is generally less than 75 %. On the contrary, Kosa is a phenomenon in which a large amount of mineral dust is blown into the sky from continental loess regions, transported westerly, and gradually falls widely to the ground. The sky assumes a yellowbrown color during severe conditions. In addition, the Sun s brightness decreases remarkably. Note that visibility with Kosa is not

3 September 2014 S. KITAKOGA et al. 59 Fig. 1. Map of the Sky Radiometer Network (SKYNET) sites from which aerosol data analyses were conducted in this study. Fukue-jima Island (32.7 N, E) and Amami-Oshima Island (28.4 N, E) are indicated by gray solid circles. necessarily less than 10 km generally. However, only Kosa events with visibility less than 10 km (VH) were selected for this study. 2.2 Observation data In this subsection, the four types of observation data such as meteorological observation, sky radiometer, LIDAR, and OPC data are outlined. a. Meteorological observation Meteorological observation data such as visibility, cloud amount, weather, and atmospheric phenomena were obtained from observatories in Fukue (32.7 N, E) and Naze (28.4 N, E). The observation was conducted at 3:00, 6:00, 9:00, 12:00, 15:00, 18:00, and 21:00 JST every day from 2003 to 2004, whereas correlation analyses included only the data obtained at 9:00, 12:00, and 15:00 JST. b. Sky radiometer A sky radiometer (PREDE POM-01) measures solar and sky irradiance, including aerosol, by using specific wavelengths from the visible to the near infrared spectral region and specific scattering angles near the Sun (Aoki and Fujiyoshi 2003). Aerosol optical depth τ a, Ångström exponent α, and single scattering albedo ω 0 are estimated using the inversion package SKYRAD.pack ver. 4.2 (Nakajima et al. 1996), in which the scheme is based on the Mie theory for homogeneous and spherical particles. The aerosol optical depth τ a is a parameter related to columnar aerosol loading in the atmosphere if other properties are identical. The Ångström exponent α is an index of aerosol size; α is smaller (larger) when larger (smaller) particles dominate. The single scattering albedo ω 0 is a ratio of scattering to extinction coefficient with a single scattering event; ω 0 is smaller (larger) when aerosol light absorptivity is larger

4 60 Journal of the Meteorological Society of Japan Vol. 92A (smaller). c. LIDAR LIDAR (Shibata Scientific Technology Model L2S-SM II) measures scattered signals at wavelengths of 532 nm and 1064 nm and depolarization at 532 nm with 30 m vertical resolution. The vertical profiles of the aerosol extinction coefficients are converted from backscatter coefficients. The extinction coefficients are classified as dust and sphere for non-spherical particles such as Kosa and spherical particles such as sulfates, respectively (Shimizu et al. 2004). In the present study, total represents the sum of dust and sphere aerosols for the LIDAR products. d. OPC The aerosol number size distribution was obtained from OPC measurements (Rion KC-01D) with five diameter ranges such as more than and equal to 0.3, 0.5, 1.0, 2.0, and 5.0 µm. The relationship between particle sizes and scattered signals detected with OPC was calibrated using spherical polystyrene latex particles with a laser beam at wavelength of 780 nm. 2.3 Data analyses From the sky radiometer measurements, aerosol optical depth τ a, Ångström exponent α, and single scattering albedo ω 0 were retrieved as described in Section 2.2.b. In addition, aerosol properties were obtained using LIDAR and OPC measurements. From the LIDAR products, aerosol non-sphericity and optical scale height were derived at the lower atmosphere. From the OPC data, however, indices of aerosol loading and number size distribution were estimated near the surface. a. Ratio of non-spherical particles at lower atmospheric layer η From LIDAR products, a ratio of non-spherical particles at the lower atmospheric layer η was estimated: τ η = τ + dust, (1) τ dust sphere where τ dust and τ sphere are the integrated extinction coefficients over the lower layers for dust and spheres, respectively, which is defined as the first consecutive layer from the ground with only positive extinction coefficients for the dust and sphere particles. τ τ dust sphere = σdust z + σdust z+ z z ( ( ) ( )), 2 σdust> 0 = σsphere> 0 (2) σsphere z + σsphere z+ z z ( ( ) ( )), (3) 2 where σ dust (z) and σ sphere (z) show the vertical profiles of the extinction coefficients at height z for dust and spheres, respectively, and Δz is the vertical resolution of 30 m. Takamura et al. (2007) previously introduced a non-sphericity index of particles, known as the Yellow Sand Index (YSI), which is defined at each atmospheric level. Because our analyses focus on the lower atmosphere near the surface, the parameter η is defined with integrated extinction coefficients in Eqs. (1) to (3). Note that the geometrical thickness of the first consecutive layer from the ground with only positive extinction coefficients may differ for dust and sphere particles. b. Aerosol optical scale height H m Assuming that the vertical profile of extinction coefficients is exponential, the aerosol optical scale height H m is calculated from the following formula (Hayasaka et al. 2007): Hm 0 σ τ 1 ( zdz ) = a( 1 e ) τ a, (4) where σ (z) is a vertical profile of the extinction coefficients of dust, spheres, or total particles, and τ a is the integrated extinction coefficient. In this study, the aerosol optical scale height H m was derived such that the extinction coefficient σ (z) was integrated within the first consecutive layer from the ground in which σ (z) was positive while the integrated value did not exceed τ a. Aerosol optical scale height is a measure of aerosol vertical distribution; when aerosols are distributed in higher altitudes, H m is larger. c. Ratio of coarse particles γ s From the OPC data, the number ratio of coarser to total particles γ s was estimated by the following formula: Nd 2µ m γ s = log10 N, (5) total where N d 2 µm is the number of particles larger than and equal to 2 µm in diameter, and N total is the total particle number such that the diameter is more than and equal to 0.3 µm; this includes the total number of

5 September 2014 S. KITAKOGA et al. 61 Fig. 2. Characteristics of atmospheric phenomena with aerosol properties. Circles represent normal atmospheric conditions such that the visibilities are equal to or more than 10 km, rhombuses represent points of VH, stars indicate Kosa, triangles represent haze, and the values in parentheses indicate the number of datasets. The error bar is the standard deviation. (a) Single scattering albedo ω 0 versus Ångström exponent α with the color bar showing aerosol optical depth τa at 500 nm, as obtained from sky radiometer measurements. (b) Ratio of non-spherical particles η versus the ratio of coarse particles γs, with the color bar showing the total volume of aerosols Vtotal in a common logarithmic scale, as obtained from LIDAR and OPC measurements. all observed particles. The parameter γ s was used to concisely determine the particle size distribution. d. Total volume concentration of aerosols V total The total volume concentration of aerosols V total in µm 3 L 1 was estimated by 5 Dj Vtotal = 4 3 π n j 3 2, (6) j= 1 where D j is the representative diameter of the particle and n j is the number of particles for each diameter bin. The variables D j and n j are explained in the Appendix. 3. Results In this section, the atmospheric phenomena were characterized with aerosol optical and microphysical properties for Fukue-jima and Amami-Oshima Islands in the East Asia region. Specific events were investigated in detail. 3.1 Characteristics of atmospheric phenomena with aerosol properties Figure 2 shows the characteristics of atmospheric phenomena with aerosol properties, which are summarized in Table 1. Aerosol properties such as single scattering albedo ω 0, Ångström exponent α, and aerosol optical depth τ a were obtained from sky radiometer measurements. As a result of the statistical analyses with the columnar aerosol properties, for haze events, the average ± standard deviation for the ω 0 was 0.96 ± 0.01; α, 1.26 ± 0.09; and τ a, 0.47 ± 0.16 (Fig. 2a). Conversely, for the Kosa event, the values were 0.99, 0.70, and 0.42, respectively. Note that single scattering albedo ω 0 for Kosa was 0.99, which was larger than 0.96 ± 0.01 recorded for haze. After the correlation analyses with sky radiometer measurements based on the atmospheric phenomena, the numbers of the analyzed datasets were 79 (Normal) and 5 (VH); nearly all data were observed at Fukue-jima Island. These data were selected because they simultaneously met the following requirements, as shown in Fig. 2a: (1) The fine weather data with a cloud amount of not more than unity out of 10 were selected to suppress cloud contamination, (2) the first consecutive layer from the ground for total particles (dust + spheres) in the LIDAR profiles had an upper limit of more than or equal to 3000 m and lower limit of less than or equal to 120 m, and (3) the measurements obtained with the sky radiometer were synchronized with the meteorological observation within 5 min. Aerosol properties such as the ratio of non-spherical particles η, ratio of coarse particles γ s, and total volume of aerosols V total were obtained from LIDAR

6 62 Journal of the Meteorological Society of Japan Vol. 92A Table 1. Characteristics of atmospheric phenomena with aerosol properties. Normal indicates the atmospheric condition in which visibility is more or equal to 10 km; VH indicates VH in which visibility is less than 10 km and consists of haze and Kosa events. Single scattering albedo ω 0, Ångström exponent α, and aerosol optical depth τ a were obtained from sky radiometer measurements. The ratio of non-spherical particles η, ratio of coarse particles γ s, and total volume concentration of aerosols V total were derived from LIDAR and OPC measurements. The statistics are shown with the average ± standard deviation. Number of data is the sum of the data used in Fukue-jima and Amami-Oshima Islands; the number in parenthesis indicates that of Fukue-jima Island only. Aerosol properties Normal VH Kosa ω 0 α τ a Number of data (Fukue) η γ s V total Number of data (Fukue) 0.94 ± ± ± (78) 0.25 ± ± ± (46) 0.97 ± ± ± (5) 0.38 ± ± ± (7) 0.96 ± ± ± (4) 0.35 ± ± ± (6) (1) 0.47 ± ± ± (1) and OPC measurements. As a result of the statistical analyses with aerosol properties near the ground, for haze events, the average ± standard deviation for the ratio of non-spherical particles η was 0.35 ± 0.12, the ratio of coarse particles γ s was 2.71 ± 0.32, and total volume of aerosols V total was 4.63 ± 0.26 (Fig. 2b). On the contrary, for the Kosa event, the values were were 0.47 ± 0.02, 2.20 ± 0.45, and 4.56 ± 0.23, respectively. At the end of the correlation analyses with LIDAR and OPC measurements based on the atmospheric phenomena, the numbers of analyzed datasets were 74 (Normal) and 8 (VH) and were observed mainly at Fukue-jima Island. These data were selected because they simultaneously met the following requirements, as shown in in Fig. 2b: (1) The fine weather data with a cloud amount of not more than unity out of 10 were selected to suppress cloud contamination, (2) the first consecutive layer from the ground for total particles (dust + spheres) in the LIDAR profiles had an upper limit of more than or equal to 3000 m and lower limit of less than or equal to 120 m, and (3) the measurements with OPC and LIDAR were synchronized with the meteorological observation within 5 min and fully, respectively. Figure 2 clearly indicates that the error bars with normal atmospheric conditions (Normal) are generally larger than those with atmospheric phenomena (VH). It is suggested that aerosols over the East China Sea generally have a wide variety of features, whereas the specific properties are dominated with VH. A comparison between Normal and VH in terms of τ a and V total also indicated that aerosol loading was heavier with atmospheric phenomena than with normal conditions. In comparison with atmospheric phenomena, it was discovered that aerosol loading was heavier with haze than with Kosa. In terms of ω 0, α, and γ s, it was indicated that Kosa consisted of hardly light-absorbing and larger particles. On the contrary, haze contained weak light-absorbing and smaller particles. Non-sphericity η was larger with Kosa than with haze at the lower layer, and the ratio of coarse particles γ s was larger at the surface, which suggests that dust particles were falling to the ground. Therefore, it is suggested that haze events are significantly influenced by anthropogenic aerosols (Ohta et al. 2013; d Almeida et al. 1991). It was also determined that the relationship among these aerosol parameters is consistent for both atmospheric column (Fig. 2a) and lower layer (Fig. 2b). That is, all optical sensors such as sky radiometer, LIDAR, and OPC were sensitive to the same aerosols near the surface accompanied by the atmospheric phenomena. Figure 3 shows a boxbar chart for aerosol optical scale height H m estimated using Eq. (4) for the first consecutive layers. It is clear that aerosols dominated near the lower level of 1 km in altitude. A comparison between Normal and VH indicates that the vertical range of distribution with Normal was wider than that with VH, which is consistent with the other aerosol properties shown in Fig. 2. It was also determined that H m was higher and lower with dust and sphere aerosols, respectively, within the VH conditions and that aerosol heights were generally lower with haze than with Kosa.

7 September 2014 S. KITAKOGA et al. 63 Fig. 3. Boxbar chart of aerosol optical scale height Hm for the atmospheric condition based on LIDAR measurements. Normal indicates the atmospheric condition in which visibility is more or equal to 10 km; VH indicates VH when visibility is less than 10 km. The VH cases included six and two events for haze and Kosa, respectively. The value in parentheses is the number of datasets. The solid line indicates dust and the dotted line indicates spheres. The bars show maximum, 75th percentile, median (50th percentile), 25th percentile, minimum, and average, according to the legend. 3.2 Specific events According to the detailed match-up procedures, 12 atmospheric phenomena were observed at Fukuejima and Amami-Oshima Islands in 2003 and The weather conditions of the 12 cases were generally under high pressure systems, as revealed by the weather charts in the supplemental file. All characteristics of aerosols with the atmospheric phenomena are summarized in Table 2. The 12 atmospheric phenomena listed in Table 2 correspond with the number of VH data in Table 1. Note that the VH data from the sky radiometer and from LIDAR and OPC in Table 1 were five and eight, respectively; simply, their sum was 13. However, one event was the same Kosa event as case L in Table 2, and the total number of the synchronized cases, from cases A to L, was 12. Further details are shown in the weather charts, backward trajectories, and maps created with the Chemical Weather Forecasting System (CFORS; Uno et al. 2003) in the supplemental file. The JMA s records of atmospheric phenomena suggest that cases A to C, E to F, and H to J were considered as consecutive haze events. The consecutive mist events occurred between cases H and I, which indicates that water vapor increased during the mist events. Table 2 indicates that the 12 correlation datasets had a seasonal deviation, including seven cases in spring from March to May, three in summer in June and July, and two in winter in December and February. There were no correlation data in autumn in this study. However, there was no specific deviation in local time among 9:00, 12:00, and 15:00 JST. The haze events were frequently observed in the spring with 6 cases from A to F of 10 from A to J. Kosa events were observed at the end of winter (case K) and in the middle of spring (case L). The atmospheric phenomena and sky radiometer measurements coincided in cases G to J and L, whereas the atmospheric phenomena and LIDAR as well as OPC coincided in cases A to F, K, and L. All measurements were synchronized in case L. In terms of the aerosol properties retrieved from sky radiometer measurements, such as single scattering albedo ω 0, Ångström exponent α, and aerosol optical depth τ a, the individual values did not appear to be as irregular as the haze events for cases G to J, as indicated in Fig. 2 and Table 1. However, the value for case L appeared to differ somewhat from the Kosa

8 64 Journal of the Meteorological Society of Japan Vol. 92A Table 2. Characteristics of individual atmospheric phenomena with aerosol properties. In addition to the properties listed in Table 1, optical scale heights, H dust, H sphere, and H total were tabulated for dust, spheres, and total aerosols, respectively. indicates that no synchronization was observed. The Kosa event (case K) was observed at AmamiOshima Island only; the other events occurred at Fukuejima Island. Case Atmospheric phenomena Date [JST] ω 0 α τ a η γ s V total H dust (km) H sphere (km) H total (km) A B C D E F G H I J K L Kosa Kosa Mar. 26, :00 Mar. 26, :00 Mar. 26, :00 May 01, :00 May 23, :00 May 23, :00 Dec. 24, :00 Jun. 30, :00 Jul. 01, :00 Jul. 01, :00 Feb. 27, :00 Apr. 22, : event in the previous study, such that the ω 0 of 0.99 is larger than the value up to 0.90 (e.g., Ohta et al. 2013) and the α of 0.70 is equivalent to the value of approximately 0.81 (e.g., Takamura et al. 2007). Case L with the Kosa event is further investigated subsequently. In terms of the aerosol properties estimated by LIDAR and OPC measurements, such as ratio of non-spherical particles at the lower atmospheric layer η, the ratio of coarse particles γs, and the total volume concentration of aerosols V total, we determined that η was higher than 0.3 in the spring generally from February to May. Note that the η of 0.13, as in case F, dropped abruptly from 0.39, as in case E, by a factor of three for 6 h. Cases E and F with haze events are further investigated subsequently. Furthermore, in terms of the aerosol optical scale heights such as H dust, H sphere, and H total for non-spherical, spherical, and total particles, respectively, it is apparent that cases E and F with haze events and case L with the Kosa event were even higher than the others. Moreover, the average was approximately 1 km, as shown in Fig. 3, which indicates that the air mass may have originated from remote regions. To concisely investigate the air mass paths, trajectory analyses with the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT; Draxler and Rolph 2013) were conducted at Fukue-jima and Amami-Oshima Islands for several days for all 12 atmospheric phenomena. It was confirmed that air masses at levels of 300 m and 1000 m in altitude over the two sites travelled over industrialized cites in China in 6 cases (A, B, C, D, G, and H) of the 10 haze events in cases A to J. Additional information is shown in the weather charts, backward trajectories, and CFORS maps in the supplemental information. The specific events with haze (cases E and F) and Kosa (case L) are investigated in detail in the following subsection. a. events events, cases E and F, were observed at Fukue-jima Island on 09:00 and 15:00 JST 23 May

9 September 2014 S. KITAKOGA et al , respectively (Table 2). The ratios of coarse particles γ s were 3.09 (case E) and 3.04 (case F). These values are even smaller than those of the other haze events shown in Table 2, which indicates that smaller particles dominated at that time. The non-sphericity η abruptly decreased from 0.39 to 0.13 by a factor of three for 6 h. The aerosol optical scale height H dust also decreased, whereas H sphere increased. These variations in aerosol non-sphericity and vertical distribution are attributable to the decrease in extinction coefficient for non-spherical particles (dust) at a level of approximately 2000 m, as shown in Figs. 4a and 4b. Furthermore, seven-day backward trajectory analyses at levels of 1000, 2000, and 3000 m based on Fukue-jima Island indicates that the air mass originated from the Siberian forest fire area, which is shown in Fig. 4c and the supplementary file. Therefore, it is suggested that the aerosol occurring with the extensive wild fire event near Siberia in spring 2003 was observed over Fukue-jima Island as a haze event. b. Kosa events As shown in Table 2, for case K with Kosa, the ratio of coarse particles γ s of 1.88 was significantly larger than the others at Amami-Oshima Island. Furthermore, the backward trajectory analyses indicated that dust particles existed near the ground and that air mass over Amami-Oshima Island was transported from the Gobi and Takla Makan desert regions. Therefore, the dust particles were transported from the desert regions in China. A Kosa event was also observed on 22 April 2004 as case L in Table 2. At that time, γ s with case L was smaller than that with case K. In addition, the single scattering albedo ω 0 of 0.99 was relatively large for case L, which is usually observed with hardly light-absorbing particles such as sulfates. Figure 5a shows the vertical profile of the aerosol extinction coefficients for dust and sphere particles. It is apparent that these coefficients are comparable even with the Kosa event. According to the distribution map of sulfates with CFORS, a comparatively heavy concentration of sulfate particles existed over Fukuejima Island (Fig. 5b) even with the Kosa event (Fig. 5c). Even though Figs. 5b and 5c are the forecasted results, it is well explained that the Kosa event of case L in spring over the East China Sea appeared to contain mineral dust from inland China as well as sulfates from East or southeastern Asia. Biomass burning events accompanied with slash-and-burn agriculture in that season usually emit sulfate aerosols (e.g., Zhang et al. 2003). These results are also consistent with the fact that anthropogenic particles traveled with the Kosa event in spring in the East China Sea (Ohta et al. 2013). 4. Concluding remarks The characteristics of the atmospheric phenomena were investigated in terms of aerosols by using sky radiometer, LIDAR, and OPC observations at Fukuejima and Amami-Oshima Islands during 2003 and The characteristics of the haze and Kosa events with the aerosol properties are summarized as follows: The aerosol loading was heavier with the atmospheric phenomena than with normal conditions and with haze than with Kosa among atmospheric phenomena. The Kosa consisted of hardly light-absorbing and larger particles. The haze contained weak light-absorbing and smaller particles, and the non-sphericity with Kosa was larger than that with haze at the lower layer. The aerosol optical scale heights with haze were generally lower than those with Kosa. From the second point above, it is suggested that the anthropogenic particles with hardly light-absorption were mixed with Kosa because the value of the single scattering albedo of more than 0.95 appeared to be even higher with the Kosa event. These results confirmed the existence of a wide variety of aerosol species with the atmospheric phenomena under VH conditions over the East China Sea even with the relatively short-term data analyses. Furthermore, we determined that the retrieved aerosol properties were consistent, even if the sky radiometer, LIDAR, and OPC measurements showed different ranges for the atmospheric column, lower layer, and surface, respectively. This result suggests that the other optical sensors such as a sun photometer and an all-sky camera were also most sensitive to the heavy aerosol loading in the lower atmosphere in the visible spectral region when the horizontal visibilities were less than 10 km with the atmospheric phenomena under VH conditions. This case is illustrated clearly in the LIDAR extinction profiles. Therefore, it was demonstrated that a combination of state-of-the-art instrumentation in terms of aerosols is useful for quantitatively characterizing atmospheric phenomena that were identified by visual observation.

10 66 Journal of the Meteorological Society of Japan Vol. 92A Fig. 4. Vertical profiles of the extinction coefficients (km 1) and aerosol optical scale height Hm based on the LIDAR measurements at (a) 9:00 JST (case E in Table 2) and (b) 15:00 JST 23 May 2003 (case F in Table 2), respectively. The range of abscissa is limited to 1.0 km 1 to show the variation in the smaller extinction coefficients. The plots of the extinction profiles are limited to the first consecutive layer from the ground with only positive extinction coefficients. Blue, green, and red lines show the aerosol extinction coefficients for total (dust + spheres), dust, and spheres, respectively. (c) Results of backward trajectory analyses at Fukue-jima Island with the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT), beginning at altitudes of 3000 m (circle), 2000 m (square), and 1000 m (triangle) for the seven-day analysis. Acknowledgments The authors are grateful for their many useful conversations with Prof. Kazuma Aoki at Toyama University, Prof. Tamio Takamura and Dr. Pradeep Khatri at Chiba University, Dr. Akihiro Uchiyama at the Meteorological Research Institute/JMA, and Dr. Nobuo Sugimoto and Dr. Atsushi Shimizu at the National Institute for Environmental Studies. The authors are also grateful to Prof. Sachiko Hayashida, Ms. Mayumi Hibino, Ms. Yoshimi Azuma, and Ms. Ayano Ota at Nara Women s University for their

11 September S. KITAKOGA et al. Fig. 5. (a) Vertical profiles of the extinction coefficients (km 1) and aerosol optical scale height Hm based on the LIDAR measurements obtained at 9:00 JST 22 April 2004 (Case L in Table 2). The range of abscissa is limited to 1.0 km 1 to show the variation in the smaller extinction coefficients. The plots of the extinction profiles are limited to the first consecutive layer from the ground with only positive extinction coefficients. The CFORS distribution map shows (b) sulfate and (c) dust at 9:00 JST 22 April The CFORS maps were provided by NIES ( www-cfors.nies.go.jp/.) assistance with the data analyses. The comments from the two anonymous reviewers for improving the clarity of this paper are highly appreciated. The aerosol optical properties and meteorological data were kindly provided by the SKYNET project and JMA, respectively. It should be mentioned that the data of the sky radiometer, LIDAR and OPC were kindly provided by the University of Toyama, the National Institute for Environmental Studies (NIES), and Chiba University, respectively. This study was conducted through the joint research program of CEReS, Chiba University, Japan. Appendix The original size of the bin of the number of aerosols obtained from OPC measurements is shown in Table 3. The number concentration for the smaller bin such that the diameter d is more than or equal to 0.3 µm and less than 0.5 µm is estimated as n 1 = N1 N2 (0.3 d < 0.5) (7) In the same manner, the number concentration for other smaller bins is defined as shown in Table 4, and Vtotal is estimated with dj and nj in Eq. (7). The representative diameter Dj is estimated according to the arithmetic average as

12 68 Journal of the Meteorological Society of Japan Vol. 92A Table 3. Relationship between the range of aerosol diameter and number concentration with OPC measurements. Bin number i Range of diameter d (μm) d 0.3 d 0.5 d 1.0 d 2.0 d 5.0 Number concentration N i (Particles L 1 ) N 1 N 2 N 3 N 4 N 5 Table 4. Same as Table 3, but for a smaller size bin. The representative diameter is a mean value of each bin and is used in Eq. (6) to estimate the total volume concentration. Bin number j Range of diameter d (μm) Representative diameter D j (μm) Number concentration n j (Particles L 1 ) d < d < d < d < d n 1 n 2 n 3 n 4 n 5 D =. +. = 04., (8) 2 for the first bin (j = 1), for example. On the contrary, D 5 = 5.0 for the final bin. References Aoki, K., and Y. Fujiyoshi, 2003: Sky radiometer measurements of aerosol optical properties over Sapporo, Japan. J. Meteor. Soc. Japan, 81, d Almeida, G. A., E. P. Shettle, and P. Koepke, 1991: Atmospheric Aerosols: Global Climatology and Radiative Characteristics. A. Deepak Pub., 561 pp. Draxler, R., and G. Rolph, 2013: HYSPLIT (HYbrid Single-Particle Lagrangian Integrated Trajectory) Model access via NOAA ARL READY Website. NOAA Air Resources Laboratory, College Park, MD. [Available at Hayasaka, T., S. Satake, A. Shimizu, N. Sugimoto, I. Matsui, K. Aoki, and Y. Muraji, 2007: Vertical distribution and optical properties of aerosols observed over Japan during the Atmospheric Brown Clouds- East Asia Regional Experiment J. Geophys. Res., 112, D22S35, doi: /2006JD Huebert, B. J., T. Bates, P. B. Russell, G. Shi, Y. J. Kim, K. Kawamura, G. Carmichael, and T. Nakajima, 2003: An overview of ACE-Asia: Strategies for quantifying the relationships between Asian aerosols and their climatic impacts. J. Geophys. Res., 108, 8633, doi: /2003JD JMA, 1998: Meteorological observation companion. Japan Meteorological Agency. Nakajima, T., M. Sekiguchi, T. Takemura, I. Uno, A. Higurashi, D. Kim, B. Sohn, S.-N. Oh, T. Y. Nakajima, S. Ohta, I. Okada, T. Takamura, and K. Kawamoto, 2003: Significance of direct and indirect radiative forcings of aerosols in the East China Sea region. J. Geophys. Res., 108, 8658, doi: /2002JD Nakajima, T., G. Tonna, R. Rao, P. Boi, Y. Kaufman, and B. Holben, 1996: Use of sky brightness measurements from ground for remote sensing of particulate polydispersions. App. Opt., 35, Nakajima, T., S.-C. Yoon, V. Ramanathan, G.-Y. Shi, T. Takemura, A. Higurashi, T. Takamura, K. Aoki, B.-J. Sohn, S.-W. Kim, H. Tsuruta, N. Sugimoto, A. Shimizu, H. Tanimoto, Y. Sawa, N.-H. Lin, C.-T. Lee, D. Goto, and N. Schutgens, 2007: Overview of the Atmospheric Brown Cloud East Asian Regional Experiment 2005 and a study of the aerosol direct radiative forcing in east Asia. J. Geophys. Res., 112, D24S91, doi: /2007jd Ohta, S., N. Murao, and S. Yamagata, 2013: Optical and chemical properties of atmospheric aerosols at Amami Oshima and Fukue islands in Japan in spring, J. Meteor. Soc. Japan, 91, Sano, I., S. Mukai, Y. Okada, B. N. Holben, S. Ohta, and T. Takamura, 2003: Optical properties of aerosols during APEX and ACE-Asia experiments. J. Geophys. Res.,

13 September 2014 S. KITAKOGA et al , 8649, doi: /2002jd Shimizu, A., N. Sugimoto, I. Matsui, K. Arao, I. Uno, T. Murayama, N. Kagawa, K. Aoki, A. Uchiyama, and A. Yamazaki, 2004: Continuous observations of Asian dust and other aerosols by polarization lidars in China and Japan during ACE-Asia. J. Geophys. Res., 109, D19S17, doi: /2002jd Takamura, T., N. Sugimoto, A. Shimizu, A. Uchiyama, A. Yamazaki, K. Aoki, T. Nakajima, B. J. Sohn, and H. Takenaka, 2007: Aerosol radiative characteristics at Gosan, Korea, during the Atmospheric Brown Cloud East Asian Regional Experiment J. Geophys. Res., 112, D22S36, doi: /2007jd Uno, I., G. R. Carmichael, D. G. Streets, Y. Tang, J. J. Yienger, S. Satake, Z. Wang, J.-H. Woo, S. Guttikunda, M. Uematsu, K. Matsumoto, H. Tanimoto, K. Yoshioka, and T. Iida, 2003: Regional chemical weather forecasting system CFORS: Model descriptions and analysis of surface observations at Japanese island stations during the ACE-Asia experiment. J. Geophys. Res., 108, 8668, doi: /2002JD Zhang, M., I. Uno, G. R. Carmichael, H. Akimoto, Z. Wang, Y. Tang, J.-H. Woo, D. G. Streets, G. W. Sachse, M. A. Avery, R. J. Weber, and R. W. Talbot, 2003: Large-scale structure of trace gas and aerosol distributions over the western Pacific Ocean during the transport and chemical evolution over the Pacific (TRACE-P) experiment. J. Geophys. Res., 108, doi: /2002JD