Asymmetry parameters in the lower troposphere derived from aircraft measurements of aerosol scattering coefficients over tropical India

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113,, doi: /2008jd009795, 2008 Asymmetry parameters in the lower troposphere derived from aircraft measurements of aerosol scattering coefficients over tropical India S. Ramachandran 1 and T. A. Rajesh 1 Received 6 January 2008; revised 18 April 2008; accepted 9 May 2008; published 28 August [1] Aerosol scattering coefficients (total b sca and backscatter b backsca ) are measured on board an aircraft using an integrating nephelometer at 450, 550, and 700 nm in the 0 to 3000-m region over four locations in India in an air campaign held during March May b sca is a factor of two higher in the east (Bhubaneshwar, Chennai) than in the west (Trivandrum, Goa). b sca is about m 1 over Bhubaneshwar and Chennai. b backsca is about an order of magnitude lower than b sca. Seven-day air back trajectory analysis indicate that air masses originating from arid/semiraid regions, continents, and marine regions are found to influence the aerosol characteristics, in addition to local urban sources. No elevated aerosol layers are seen during the campaign. b, the aerosol backscatter fraction, is greater than 0.13 in the lower troposphere. The columnar mean Ångström exponent (a) is >1.75. Asymmetry parameter g profiles are derived for the first time over India in the lower troposphere, using the relation between b and g. 550-nm g corresponding to 30% RH is in the range over India. Higher b, higher a,andlowerg values over these locations suggest the dominance of submicron aerosols during the campaign. Scattering aerosols corrected to 30% RH in the 0 to 3000-m altitude region contribute about 20 35% to MODIS aerosol optical depths (AODs). The variation in the contribution of scattering aerosols to AODs highlights the spatial and vertical differences in aerosol properties. Citation: Ramachandran, S., and T. A. Rajesh (2008), Asymmetry parameters in the lower troposphere derived from aircraft measurements of aerosol scattering coefficients over tropical India, J. Geophys. Res., 113,, doi: /2008jd Introduction [2] Atmospheric aerosols influence the earth-atmosphere radiation budget through radiative forcing. In spite of the recognition of radiative effects of aerosols the magnitude is uncertain. The aerosol sources are widely distributed and vary on regional scales leading to regional differences in aerosol radiative forcing. The angular distribution of light scattered by aerosols, namely, the aerosol phase function is a crucial parameter which influences the aerosol contribution to radiative forcing. Asymmetry parameter (g) depends both on the size distribution and composition of aerosol particles. g is also a function of the relative humidity. Radiative transfer computations use parameterizations of the angular distribution of scattered light or the aerosol phase functions of different aerosol distributions. [3] In spite of its important role in radiative transfer studies in situ measurements and estimates of asymmetry parameter g are rare. An attempt was made in 1979 to retrieve the aerosol phase functions and g over Alaska using laser nephelometer on board an aircraft in the altitude region of 11 to 13 km [Grams, 1981]. Vertical profiles of g were derived during October 1991 and April 1 Space and Atmospheric Sciences Division, Physical Research Laboratory, Ahmedabad, India. Copyright 2008 by the American Geophysical Union /08/2008JD over Hyderabad (17.5 N, 78.6 E) for volcanic aerosols produced due to the Mt. Pinatubo eruption from balloon borne measurements of angular distribution of scattered radiation intensities [Ramachandran et al., 1994]. Several researchers have studied the relation between b (hemispheric backscatter fraction, the ratio of light scattered into the backward hemisphere to the total light scattered measured by nephelometer) and g, Wiscombe and Grams [1976], Charlson et al. [1984], and Andrews et al. [2006] to name a few. In this study, for the first time in India, from the aircraft measurements of aerosol scattering coefficients using an integrating nephelometer b and g for four locations (two each in east and west) in the 0 to 3000 m altitude region are determined. 2. Integrated Campaign for Aerosols, Gases, and Radiation Budget: Air Segment [4] A Beech craft Super King Raytheon Air 2000 series aircraft of National Remote Sensing Agency (NRSA, Hyderabad) was used for conducting the air segment of an Integrated Campaign for Aerosols, gases and Radiation Budget (ICARB) [Moorthy, 2006] in March May A total of 26 air sorties were conducted for measuring aerosol characteristics in different transects and configurations. The air sorties were conducted from Bhubaneshwar (20.2 N, 85.8 E, 46 m ASL) and Chennai (13.04 N, E, 16 m ASL) on the east, Trivandrum (8.5 N, 77 E, 8 m ASL) and Goa (15.4 N, 73.8 E, 52 m ASL) on the west. Four (one 1of10

2 from each station) multilevel flights (Table 1 and Figure 1) were conducted during which measurements of aerosol scattering coefficients were made at 8 levels located at around 500 m, 800, 1100, 1400, 1700, 2000, 2500 and 3000 m in addition to near surface altitudes. Results obtained from the aerosol scattering coefficient measurements made in the multilevel configuration are analyzed and discussed. Features in aerosol scattering coefficients (b sca, b backsca ) obtained from the multilevel flights during the campaign over tropical India are compared with results obtained from other oceanic regions and continental locations. Variabilities in Ångström exponents (a) and asymmetry parameter (g) obtained are discussed and inferences are drawn. 3. Measurements, Data Analysis, and Methodology 3.1. Instrument: Nephelometer [5] A three-wavelength integrating nephelometer (model 3563, TSI Inc., USA) which continuously measures total aerosol scattering (b sca ) and the hemispheric backscattering (b backsca ) coefficients at 450, 550 and 700 nm was used during the campaign. The instrument draws air sample through the large diameter inlet port into the measurement volume. An automated ball-valve built into the inlet is activated periodically and all of the aerosol sample is diverted through a high efficiency filter to obtain a measure of the clean-air signal pertaining to the operating environment. Built-in temperature and pressure sensors allow calculation of changes in Rayleigh scattering coefficients at three wavelengths. From the total signal, the clean-air signal and dark current of photomultiplier are subtracted to obtain the aerosol scatter signals. [6] During the campaign the ambient air was sampled through a stainless steel bent pipe fitted under the nose of the aircraft. Through the set up the inlet opened to the incoming air as the aircraft flew. The inside of the aircraft was unpressurized and the cabin temperature was maintained at 25 C. The aircraft cabin was typically colder than the ambient conditions in the lower altitudes. The built-in sample heater was turned on during the measurements to minimize condensation on the instrument walls as it operated in an air-conditioned environment. The sample temperature inside the measurement volume was about 35 C. The sample relative humidity (RH) measured inside the nephelometer measurement volume were used as inputs while scaling the aerosol scattering coefficients to 30% RH. The instrument was operated at a flow rate of 20 Lmin 1 and data were collected at 1-s resolution. Aerosol scattering coefficients are further averaged for 2-min, analyzed and presented in this study. The data are checked for quality before averaging to remove any flaws caused due to electronics during ascents and descents from/to different altitudes, and only those data after the aircraft stabilized at that particular altitude are considered. The data at each height correspond to 15 minutes. The duration of flights in multilevel configuration was about 2.5 hrs. Each air sortie traversed about 4 in either latitude/longitude or in both which translates to scattering coefficient measurements of about 400 km away from the base station (Table 1 and Figure 1) Effect of Relative Humidity on Measured Aerosol Scattering Coefficients and Estimation of Scaling Factors [7] Aerosol scattering strongly depends on the relative humidity (RH) at which it is measured, as the sizes of hygroscopic aerosols can increase with increasing relative humidities. The effect of relative humidity on the measured b sca and b backsca are determined. The hygroscopic scaling factor (SF) expressed as the ratio of (=b sca at RH/b sca at 30% RH) is determined for urban and maritime polluted aerosol models [Hess et al., 1998]. Urban aerosol model refers to strong pollution in urban areas and consists of water soluble (which originate from gas to particle conversion mechanism and contain various kinds of sulfates, nitrates, and other organic, water soluble substances), insoluble (soil particles with a certain amount of organic material) and soot (absorbing black carbon). Maritime polluted aerosol model corresponds to a marine environment under anthropogenic influence and is made up of water soluble particles, sea salt in the accumulation and coarse modes, and soot [Hess et al., 1998]. Water soluble and sea salt are hygroscopic, while insoluble and soot aerosols are hydrophobic. b sca at 450, 550 and 700 nm for urban and maritime polluted aerosol models using the log normal distribution parameters (r m, mode radii and, s width of size distribution) [Hess et al., 1998] are calculated for RH varying from 30 to 80%. [8] SF at 550 nm for urban aerosol model are found to be 1.08 (40%), 1.16 (50%), 1.29 (60%), 1.41 (70%) and 1.69 (80% RH). SF for maritime polluted aerosol model are 1.13 (40%), 1.27 (50%), 1.42 (60%), 1.58 (70%) and 1.89 (80% RH). In the 40 80% RH range SF for maritime polluted aerosol model are higher by 4 10% than urban aerosol type, as maritime polluted aerosol model contains water soluble and sea salt aerosols. The measurement locations in the east (Bhubaneshwar and Chennai) are urban, industrialized centers while the study areas on the west (Trivandrum and Goa) are located very close to the coast and are unindustrialized. In this study, taking into account the measurement locations, the aerosol type over the location and air back trajectory analysis (discussed later), SF corresponding to urban aerosols are used for RH correction of b sca and b backsca over Bhubaneshwar and Chennai, while SF corresponding to maritime polluted aerosol type are used for RH correction over Trivandrum and Goa (Table 2). [9] SF estimated in this study for urban and maritime polluted aerosol regimes are in good agreement with the mean value of 1.7 ± 0.3 at 80% RH obtained by Hegg et al. [1996] and Charlson et al. [1984]. Hegg et al. obtained the scaling factors for ambient conditions ranging from clean maritime background air to mildly polluted continental air. The mean SF values obtained during the Indian Ocean Experiment (INDOEX) on board C-130 aircraft in the 0 to 1000 m and 1000 m to 3000 m were 1.58 ± 0.21 and respectively over the Northern Indian Ocean (>5 N) [Sheridan et al., 2002]. The mean SF value was about 1.5 for the scattering coefficient measurements made over Kaashidhoo Climate Observatory (KCO, 5 N, 73 E) during INDOEX [Clarke et al., 2002]. Note that the SF values over the Northern Indian Ocean and KCO were obtained for the reference RH of 40% and the SF values mentioned above correspond to b sca at 85% RH/b sca at 40% RH. In comparison SF values used in the current study correspond to b sca at 80% 2of10

3 Figure 1. Tracks of multilevel air sorties conducted over Bhubaneshwar (28 March 2006), Chennai (5 April), Trivandrum (24 April), and Goa (1 May). See Table 1 for more details. RH/b sca at 30% RH and are higher than the SF values obtained over the Northern Indian Ocean and KCO. However, SF at 80% RH reduces to 1.56 and 1.67 respectively for urban and maritime polluted aerosol models if b sca is scaled to 40% RH instead of 30% RH, which agree well with the SF values obtained over the Northern Indian Ocean and KCO. [10] RH of the air sample measured in the measurement volume of the nephelometer, as a function of altitude, is given in Figure 2 at four locations. RH measured in the sample volume declines sharply to less than 20% at around 1000 m over Bhubaneshwar and is more or less constant till 3000 m. Similar features in RH are seen in Chennai. The sample volume RH measured over Trivandrum shows an exponential decrease with altitude. At 1000 m the sample volume RH over Trivandrum is about 40%. At Goa RH measured in the sample volume of the nephelometer is in the 45 60% range below 1000 m. The sample volume RH at all locations are above 30% from near surface to up to about 1000 m, while at the other altitudes RH was below 30% (Figure 2). The daily mean ambient RH in the 1000 to 700 hpa (near surface to 3 km) obtained from NCEP reanalysis ( cdc.noaa.gov) was found to be in the range of 31 72% (Bhubaneshwar, 28 March), 61 78% (Chennai, 5 April), 57 81% (Trivandrum, 24 April) and 55 87% (Goa, 1 May). The 1000 to 700 hpa mean RH was found to be about 30% higher than the sample volume RH measured in the nephelometer over the four measurement locations Uncertainty in Aerosol Scattering Coefficient Measurements [11] The uncertainties in nephelometer measurements arise due to noise in the filtered air scattering coefficient, calibration drift, calibration of the instrument for Rayleigh scattering of dry air and CO 2, truncation of near scattered forward light, and uncertainty in instrument and pressure in converting the data to STP [Anderson et al., 1996; Anderson and Ogren, 1998]. The nephelometer was calibrated before the air campaign; the instrument was adjusted to read zero by passing particle free air and span calibration was performed by passing CO 2. A separate size cut-off for the aerosol samples was not used on the inlet of the nephelometer. The particle transport efficiency of TSI 3563 is greater than 95% of unit density particles from 0.05 to 5 mm diameter. [12] The uncertainties associated with aerosol inlets, tubing and losses within the nephelometer are stated to be insignificant for submicrometer particles [Clarke et al., 2002]. The losses within the nephelometer for supermicron particles could be higher in the range of 5 10% [Anderson and Ogren, 1998]. In the present study the measurements are done over urban areas and locations very close to the coast, where submicron particles are expected to dominate the aerosol size distribution. Truncation of near forward scattered light which is an inherent and an unavoidable problem in nephelometer could introduce an uncertainty of 1 10% [Clarke et al., 2002; Anderson et al., 1996]. The nephelometer measurements have been corrected for truncation errors following Anderson and Ogren [1998] and Anderson et al. [1996]. The overall uncertainty in b sca in the current study taking into account the above sources is estimated to be about 15% Derivation of Backscatter Fraction (b), Ångström Exponent (a), and Asymmetry Parameter (g) [13] b sca and b backsca corrected to 30% RH (Table 2) are used to calculate the hemispheric backscatter fraction (b). b is defined as the ratio of light scattered into the backward hemisphere (b backsca ) to total light scattering (b sca ) [Charlson et al., 1984]. The sample volume RH was greater than 30% in the 0 to 1000 m altitude region (Table 2). [14] b sca measured at 450, 550 and 700 nm are analyzed further to get an insight into the sizes of particles (smaller or bigger) which dominate the aerosol distribution. [15] Ångström exponent, a which describes the dependence of aerosol scattering coefficient on wavelength can be written as a ¼ ln ½ b scað700 nmþ=b sca ð450 nmþš : ð1þ ln½700=450š [16] Smaller a values indicate the dominance of larger particles in scattering while higher values arise due to increase of smaller particles in an aerosol size distribution. [17] Wiscombe and Grams [1976] derived a smooth relationship between b and asymmetry parameter g (their Figure 3) following the Henyey Greenstein approximation for g. Andrews et al. [2006] obtained an equation between b and g based on the plot of Wiscombe and Grams [1976], and used the equation to determine g. The equation con- Table 1. Details of Four Multilevel Air Sorties Conducted During ICARB a Time of Flight, hrs Coverage Date Location (Latitude, Longitude) Begin End Latitude Longitude 28 March Bhubaneshwar (20.2 N, 85.8 E) N E 5 April Chennai (13.04 N, E) N E 24 April Trivandrum (8.5 N, 77 E) N E 1 May Goa (15.4 N, 73.8 E) N E a Results are discussed in this study. 3of10

4 Figure 2. Altitude profiles of the mean sample volume relative humidity (%) of air measured in the measurement volume of nephelometer over Bhubaneshwar, Chennai, Trivandrum, and Goa. necting b and g is written as g = * b * b * b [Andrews et al., 2006, equation (4)]. The other methods discussed in Andrews et al. [2006] require information on the size distribution of aerosols such as number density, mode radius, width and refractive index of particles. Owing to the lack of availability of the other relevant aerosol optical, physical and chemical properties during the air campaign, in this study, vertical profiles of g are derived from the relation between b and g Supplementary Meteorological Data: Air Back Trajectory Analysis [18] Air back trajectories are important to identify the source regions from where the pollutants could have originated and subsequently got transported to the measurement sites at different altitudes. The air back trajectory analysis is a three-dimensional (latitude, longitude and height) description of the movement of an air parcel as a function of time [Draxler and Hess, 1998]. It is also important to know the heights from where aerosols descended while analyzing vertical profile measurements. As the residence times of various aerosol components is about a week in the lower atmosphere, 7-day air back trajectory analysis are performed. The back trajectories describe back in time the origin of air parcels and the paths they traveled at different heights before reaching the measurement site. In Figures 3a 3d the 7-day air back trajectories are drawn for 10, 100, 500, 1000 and 2500 m heights corresponding to each measurement location. The altitudes from where the air masses originated and the heights at which the air masses traveled before reaching the measurement locations are also plotted for Bhubaneshwar, Chennai, Trivandrum and Goa. [19] At each location the trajectories at 10 and 100 m indicate no difference in source regions for the 10 and 100 m air masses, indicating a well-mixed aerosol up to 100 m. At 500 m and above the air masses originate from different arid/semiarid, continental and marine locations suggesting different source regions and aerosol types. [20] Over Bhubaneshwar air back trajectory analysis (Figure 3a) showed that aerosols at different altitudes could have originated from different source regions. The air masses originated from the marine and arid regions in the 10 to 2500 m heights, and passed through densely populated urban regions of west and central India. [21] Over Chennai air masses at 10 and 100 m originate from the Northern Indian Ocean. 500 m trajectory originates near Sri Lanka. The higher altitude air masses are found to originate from the semiarid (Pakistan) and marine (the Arabian Sea) locations, and travel through peninsular India (Figure 3b). [22] Up to 500 m air back trajectories for Trivandrum (Figure 3c) originated from and passed through the marine region (the Arabian Sea) m air trajectory originates from the arid/semiarid region (Iran) and travels close to the east coast of India m air trajectory originates from the arid location (Rajasthan) travels through mainland India (urban and industrialized regions) crosses over to the marine region (the Bay of Bengal). [23] Over Goa (Figure 3d) the 10, 100, 500 and 2500 m air trajectories originate and pass through the marine regions (the Northern Indian Ocean and the Arabian Sea). The 1000 m air trajectory for Goa originates from the arid (Saudi Arabia) region and passed through the marine region (the Arabian Sea) Complementary Satellite Data: MODIS (Terra) Retrieved Aerosol Optical Depths [24] The Moderate resolution Imaging Spectroradiometer (MODIS) is a remote sensor with two Earth Observing System (EOS) Terra and Aqua satellites which measures aerosol characteristics from space on a nearly global scale [Remer et al., 2005]. In this work, Level 3 MODIS Collection V005 daily global aerosol optical depth (AOD) from Terra at 1 Latitude 1 Longitude grid are used. This Table 2. Sample Volume Relative Humidity (%) Measured Inside the Nephelometer Measurement Volume During the Four Multilevel Air Sorties in the 0 to 1100-m Altitude Region a Bhubaneshwar Chennai Altitude (m) RH (%) b sca (10 5 m 1 ) Corrected b sca (10 5 m 1 ) SF RH (%) b sca (10 5 m 1 ) Corrected b sca (10 5 m 1 ) SF Ground level b Trivandrum Goa Ground level b a Total aerosol scattering coefficients (b sca ), aerosol scattering coefficients corrected for relative humidity effects (Corrected b sca ) and the hygroscopic scaling factors (SF) over Bhubaneshwar, Chennai, Trivandrum, and Goa. SF for an urban aerosol model are used for Bhubaneshwar and Chennai, while SF corresponding to maritime polluted aerosol type are used for Trivandrum and Goa. b Ground level data correspond to values measured at the surface of each location. 4of10

5 Figure 3. Seven-day air back trajectories for the days of multilevel air sorties conducted over (a) Bhubaneshwar (28 March), (b) Chennai (5 April), (c) Trivandrum (24 April), and (d) Goa (1 May). The back trajectories for each day are plotted at an hourly interval at 10-m, 100-m, 500-m, 1000-m, and 2500-m heights. The symbols are shown at 24-h intervals and represent a day. Below the back trajectory plots the heights from where the air masses descended/ascended and reached the observation point at an interval of 1 hr at each location are shown. analysis is performed to infer the contribution of scattering aerosols to AOD over each location. MODIS Terra and Aqua satellites operate at an altitude of 705 km with Terra spacecraft crossing the equator at about 1030 LST (ascending northward) while Aqua spacecraft crosses the equator at around 1330 LST (descending southward [Remer et al., 2005]). As the multilevel air sorties were conducted between 0900 and 1200 hrs local standard time AODs from Terra are utilized in the present study. 4. Results and Discussion 4.1. Total Scattering and Backscattering Aerosol Coefficients [25] Vertical profiles of aerosol scattering coefficients (b sca ) and aerosol back scattering coefficients (b backsca,m 1 ) at 550 nm from near surface to 3000 m over the four measurement locations, (1) Bhubaneshwar, (2) Chennai, (3) Trivandrum, and (4) Goa, are plotted in Figure 4. The nephelometer measured b sca and b backsca at all wavelengths have been scaled to 30% RH using the appropriate scaling factors whenever RH exceeded 30% (Table 2). At the outset, b sca values are found to be higher over locations in the east (Figure 1) on an average when compared to the locations in the west of India. Variations in b sca as function of altitude are seen. [26] Bhubaneshwar (population of 0.66 million based on 2001 census), Chennai (4.2 million), Trivandrum (3.2 million) and Goa (0.8 million) are densely populated urban/coastal locations, and have a large number of automobiles. Thus the gas to particle reaction products of the 5of10

6 Figure 4. Profiles of 550 nm b backsca, b sca and backscatter fraction b =(b backsca /b sca ) from near surface to up to about 3000 m over (a) Bhubaneshwar, (b) Chennai, (c) Trivandrum, and (d) Goa. b sca and b backsca plotted here have been scaled to 30% RH by using the appropriate scaling factors (Table 2) applicable to each study location. Horizontal bars in b backsca and b sca profiles represent ±1s from their respective mean values at each height. exhausts from the industry and automobiles contribute to the fine mode particles in these locations. At any given location and altitude both natural and anthropogenic sources contribute to the aerosol concentration. The percentage contribution of natural vs. anthropogenic sources to the aerosol characteristics will therefore depend on aerosol emissions, residence times, scattering, and absorption characteristics, in addition, to the aerosols which are long range transported. [27] Over Bhubaneshwar and Chennai the near surface b sca values are the highest in the m altitude region. b sca values over Trivandrum and Goa are almost about half of those obtained over Bhubaneshwar and Chennai (Figure 4 and Table 2). The back trajectory analysis and vertical profiles of b sca over Trivandrum and Goa indicate that when the air masses spend a significant time over the oceanic regions then b sca values are lower (Figures 4d and 3d). [28] The variations in the origin of back trajectory locations, the pathways of the air masses at different heights and the variations in b sca suggest that different aerosol types contribute to b sca over Bhubaneshwar, Chennai, Trivandrum and Goa in addition to the local sources over these locations. b backsca at all the four locations show similar features as that of b sca ; b backsca values are lower than b sca by about an order of magnitude (Figure 4). [29] b sca values obtained during the ICARB air campaign are found to be lower than b sca values obtained during INDOEX 1999 [Sheridan et al., 2002]. During INDOEX two major types of aerosol vertical profiles were encountered: (1) type 1 profile was characterized by fairly steady b sca above the marine boundary layer (about 1000 m) and (2) in type 2 profiles an intense layer of elevated b sca values was seen [Sheridan et al., 2002]. During ICARB air campaign no such elevated aerosol layers are seen (Figure 4). b sca values are found to be almost steady from near surface to 3000 m. INDOEX was conducted during the northeast monsoon season (January March) when the winds transport anthropogenic pollutants from the southeast Asia across the Indian Ocean [Sheridan et al., 2002]; while ICARB was conducted during the inter monsoon season of March May Aerosol Backscatter Fraction b [30] The backscatter fraction b is found to lie in the range over tropical India (Figure 4). The column averaged b values in the m altitude region are found to be 0.20 ± 0.03 (Bhubaneshwar), 0.18 ± 0.01 (Chennai), 0.21 ± 0.04 (Trivandrum) and 0.24 ± 0.04 (Goa). The 1 variations are larger in Trivandrum and Goa because of the variability in b in this altitude regime. The variability in b could be due to the different aerosol types which originated from different sources (Figure 3). The near surface mean b values are 0.14 (Bhubaneshwar), 0.15 (Chennai), 0.18 (Trivandrum) and 0.21 (Goa). [31] b values near the surface, at different altitudes in the lower troposphere and the column average over India are found to be larger than the median b values obtained during INDOEX over the Northern Indian Ocean, the Southern Great Plains, Bondville, Sable Island and Barrow (Table 3). The monthly mean b value during over the Southern Great Plains, Oklahoma was found to lie in the range of ; annual mean b value was about b was found to show a dip in midsummer which was attributed to the increase in the fraction of larger particles [Sheridan et al., 2001] over the Southern Great Plains. Mean b values were estimated to be 0.131, and for the biomass burning aerosols at Cuiabá, Porto Velho and Marabá in the Amazon basin [Kotchenruther and Hobbs, 1998]. The higher b values obtained in the current study indicate the dominance of smaller size aerosols during the premonsoon season over the select measurement locations in India Ångström Exponent (a) [32] a values derived at different altitudes from the wavelength dependent b sca over Bhubaneshwar, Chennai, Trivandrum and Goa are plotted in Figure 5. The column mean a for the m altitude region are found to be 1.84 (Bhubaneshwar), 1.98 (Chennai), 1.77 (Trivandrum) and 1.82 (Goa). a is >1.7 over Bhubaneshwar and Chennai in the lower troposphere. a decreases with increase in altitude over Trivandrum. In contrast to other locations a profile in Goa is quite different; a is 2.0 in the m altitude range. The higher a values suggest the dominance of submicron fine mode aerosols in the altitude region of 0 to 3000 m over the measurement areas. These high a values corroborate the higher b values. Higher a and 6of10

7 Table 3. Comparison of Column Averaged Aerosol Optical Properties Obtained During the Air Segment of ICARB a at Four Locations With the Median Values Obtained During 1999 Over the Northern Indian Ocean b, Rural Continental and Remote Coastal/ Marine Surface Stations c in USA Location Date/Month(s) b a g Bhubaneshwar 28 March Chennai 5 April Trivandrum 24 April Goa 1 May Northern Indian Ocean February March (C-130 aircraft) 1999 Southern Great Plains, Oklahoma August Bondville, Illinois July Sable Island, Nova Scotia August Barrow, Alaska April a b, a, and g obtained during ICARB correspond to standard temperature and pressure, and 30% RH. b The Northern Indian Ocean (latitude >5 N) results are obtained from measurements made on board C-130 aircraft in the m altitude region. c The values correspond to the highest-aerosol month, which was defined as the one having the highest median b sca at 550 nm for particles of diameter <1 mm during the 3-year measurement period of The aircraft (the Northern Indian Ocean) and the station data are for particles of diameter <1 mm. All the data are reported at standard temperature and pressure and instrument RH (typically 30 40%) [Sheridan et al., 2002]. higher b indicate the dominance of smaller size aerosols in the size distribution, and a and b should track each other (Figures 4 and 5). In comparison a values were greater than 2 over the Northern Indian Ocean, Southern Great Plains, Bondville and Sable Island; a is 1.71 for Alaska (Table 3). b values for these locations lie between 0.10 and These results suggest that as backscattering function increases (Table 3) the size distribution shifts to smaller particles resulting in higher a values Asymmetry Parameter (g) [33] There exists no direct way of measuring the asymmetry parameter g of aerosols as on date. In this study asymmetry parameter g is derived following Andrews et al. [2006] and their equation 4. The relationship between the backscatter fraction b and g showed that as b increases g decreases [Andrews et al., 2006]. It was stated by Andrews et al. [2006] that in general the Henyey Greenstein approximation is appropriate both for size distributions comprising primarily of submicron aerosol (e.g., continental, man made aerosol), based on the Mie calculations for measured submicron size distributions and for size distributions including supermicron aerosol particles such as dust, sea salt with diameter <10 mm. Andrews et al. [2006] also primarily reported values for g at RH <40%. In this study the derived mean b values are higher than 0.14 indicating the dominance of more smaller particles, and when b is large Mie theory calculated g approached the Henyey Greenstein g value. [34] Asymmetry parameter g derived from b corresponding to 30% RH is plotted in Figure 5. g values corresponding to 30% RH in the four measurement locations are less than 0.60 in the 0 to 3000 m altitude regime. The column average g values are found to be 0.44 ± 0.05 (Bhubaneshwar), 0.48 ± 0.03 (Chennai), 0.43 ± 0.07 (Trivandrum) and 0.38 ± 0.06 (Goa) respectively (Table 3). The highest g value of 0.56 was obtained at Bhubaneshwar (Figure 5a) near surface and the lowest g value of 0.29 was seen at Goa in the m altitude region. The corresponding b values were 0.14 (Bhubaneshwar) and 0.29 (Goa) respectively. For the median b values given in Sheridan et al. [2002] for the Northern Indian Ocean, the Southern Great Plains, Bondville, Sable Island and Barrow (Table 3) g values were calculated. b values in Sheridan et al. [2002] are reported at standard temperature and pressure, and an instrument relative humidity typically in the range of 30 40%. g was greater than 0.6 in the Northern Indian Ocean, Southern Great Plains, Bondville, Sable Island and Barrow (Table 3). Lower g values over the Indian locations suggest that smaller aerosol particles dominate the size distribution corroborating the higher a values obtained. As the particle size increases g will increase and a will decrease leading to an anti correlation between a and g. The anti correlation between a derived using b sca at 450, 550 and 700 nm, and g at 550 nm is seen in all the four locations (Figure 5) Contribution of Scattering Aerosol Optical Depth to Total Aerosol Optical Depth [35] Spatial maps of daily mean MODIS Terra 550 nm AODs corresponding to 28 March (Bhubaneshwar), 5 April (Chennai), 24 April (Trivandrum) and 1 May (Goa) are plotted in Figure nm b sca in the 0 to 3000 m altitude region are used to determine the scattering AOD and to compare with total columnar 550 nm MODIS AODs. Figure 5. Vertical profiles of Ångström exponent a and asymmetry parameter g determined from airborne measurements of aerosol scattering coefficients scaled to 30% RH over (a) Bhubaneshwar, (b) Chennai, (c) Trivandrum, and (d) Goa. Horizontal bars in a profiles represent ±1s from the mean a values. 7of10

8 Figure 6. MODIS Terra Level 3 Collection V latitude longitude daily mean 550 nm aerosol optical depth over the measurement region. Daily mean aerosol optical depths over 5 25 N, E latitude longitude grid on (a) 28 March 2006, (b) 5 April 2006, (c) 24 April 2006, and (d) 1 May 2006 are shown. Gaps in the figure represent no data. [36] The scattering AOD is calculated as Z z2 Scattering AOD ¼ b sca dz; ð2þ z 1 where z 1 and z 2 correspond to minimum (0 m) and maximum (3000 m) altitudes respectively. This calculation assumes that b sca varies linearly between two consecutive altitudes. Increase or decrease in b sca values in between the two adjacent altitude levels, if any, are not considered. [37] The scattering AODs are calculated from b sca corrected to 30% RH using appropriate scaling factors, while the MODIS AODs correspond to ambient (higher RH) conditions. It is well known that the sizes of the hygroscopic aerosols can increase with increasing relative humidity. Differences in RH can contribute to differences in the scattering AOD and MODIS AOD values. The scattering AODs calculated from b sca scaled to 30% RH are about 1% lower over Bhubaneshwar and Chennai, and 5% lower over Trivandrum and Goa when compared to the scattering AODs estimated from b sca values uncorrected for sample volume RH effects (Table 2). The decrease in scattering AODs when using RH corrected b sca is relatively higher over Trivandrum and Goa because (1) the scaling factors applied are higher (maritime polluted) than the urban aerosol model (Table 2) and (2) the sample volume RH is 8of10

9 Table nm AOD From MODIS Level 3 Collection V Latitude Longitude Grid Data at Ambient RH in Comparison With the Scattering AOD at 30% RH in Each Measurement Location a Location Date Scattering AOD AOD % Contribution of Scattering AOD to AOD Bhubaneshwar 28 March Chennai 5 April Trivandrum 24 April Goa 1 May a The scattering AOD is calculated by integrating the measured aerosol scattering coefficients at 550 nm scaled to 30% RH in the altitude region of 0 to 3000 m at each location. The percentage contribution of scattering AOD to the total columnar AOD is estimated. higher than 30% till 1100 m over Trivandrum and 800 m over Goa, while RH was below 30% above 500 m over Bhubaneshwar and Chennai (Figure 2). For the urban aerosol model 550 nm b sca is found to increase from m 1 at 50% RH to m 1 at 80% RH; the 550 nm AODs increase from 0.48 at 50% RH to 0.64 at 80% RH [Hess et al., 1998]. The rate of increase in b sca is higher at 1.5 when compared to the rate of AOD increase (1.3) when RH increases from 50% to 80% for the urban aerosols. These results illustrate that differences in RH could be one of the possible factors for the differences between the scattering AODs (at 30% RH) and MODIS AODs (at ambient RH). [38] In addition to the RH differences, the other possible sources that can contribute to the differences between the scattering AODs and the MODIS AODs could be (1) the missing aerosols (particles larger than the inlet cut-off size), (2) contribution from aerosol absorption, and (3) differences in wavelength. However, in the present study the measurements were performed over urban areas and locations very close to the coast where submicron aerosols will dominate the aerosol size distribution, the size range in which the TSI nephelometer is found to be quite efficient. The contribution of absorption AOD to the total AOD would be less, as it has been found that on a global mean scale, the absorption AOD contributes a maximum of 7% to the total AOD [Chung et al., 2005]. The contribution due to differences in wavelength would be nearly zero as scattering AODs and MODIS AODs correspond to the same wavelength of 550 nm. [39] 550 nm scattering and MODIS Terra AODs corresponding to each location are compared in Table 4. AODs over Bhubaneshwar and Chennai are higher than AODs found around Trivandrum and Goa (Figure 6 and Table 4), consistent with the AODs calculated from measured scattering profiles. AODs are found to vary from a low of 0.44 over Trivandrum to >1 over Bhubaneshwar (Figure 6). The scattering AODs over Bhubaneshwar and Chennai are around 0.2 while the scattering AODs are and over Trivandrum and Goa respectively. The percentage contribution of scattering AOD to AOD is less than 20% over Bhubaneshwar and Goa. The percentage contribution of scattering AOD to AOD is higher over Chennai and Trivandrum at 36% and 33% respectively. It is interesting to note that though the scattering AOD did not vary much between Bhubaneshwar and Chennai, its contribution to AOD is a factor of two higher over Chennai. Note that Chennai AOD is about 50% lower than Bhubaneshwar AOD. The above comparison shows that the scattering aerosols in the 0 to 3000 m can contribute about 20 35% to the AOD over these locations. These results reveal that the percentage contribution of scattering aerosols can vary depending on the values of scattering coefficients and AOD, thus, highlighting the importance of spatial and vertical variations in aerosol optical properties. 5. Conclusions [40] Multilevel air sorties were conducted from four locations in India: Bhubaneshwar (east), Chennai (east) Trivandrum (west) and Goa (west) during the premonsoon season of March May 2006 as part of an Integrated Campaign for Aerosols, gases and Radiation Budget (ICARB). Vertical profiles of asymmetry parameter g in the lower troposphere have been derived for the first time over India from the aerosol scattering coefficients measured using a three-wavelength nephelometer on board an aircraft. [41] b sca values on an average are higher for the locations in the east than in the west. b sca values are in the range of to m 1 in Bhubaneshwar and Chennai, while over Trivandrum and Goa b sca are about half in the altitude region of 0 to 3000 m. b sca values are found to exhibit altitudinal variations. The space altitude variations of b sca have been examined by the air back trajectory analysis. Results indicate that though the measurement locations are urban centers, air masses originating from arid/semiraid regions, continents and marine regions which carry different aerosol components influence the aerosol characteristics. During ICARB air campaign no elevated aerosol layers are seen unlike INDOEX. b backsca exhibit similar features as that of b sca, though b sca values are higher than b backsca by about an order of magnitude. [42] b, the backscatter fraction, was found to lie between 0.14 and 0.31 in the 0 to 3000 m altitude region over tropical India. The columnar mean b values in the m altitude region are found to be 0.20 ± 0.03 (Bhubaneshwar), 0.18 ± 0.01 (Chennai), 0.21 ± 0.04 (Trivandrum) and 0.24 ± 0.04 (Goa). The larger variations in b over Trivandrum and Goa are due to the variability in b in this altitude regime. The variability in b could have occurred due to the different aerosol types originating from different sources. The near surface b values are 0.14 (Bhubaneshwar), 0.15 (Chennai), 0.18 (Trivandrum) and 0.21 (Goa). b values over India are found to be larger than those obtained over the Northern Indian Ocean, Southern Great Plains, Bondville, Sable Island, Barrow and the Amazon basin. Higher b values indicate the dominance of smaller size aerosols in the aerosol size distribution. [43] Ångström exponent a is derived from b sca at 450, 550 and 700 nm. The column averaged a values in the m altitude region are found to be 1.84 (Bhubaneshwar), 1.98 (Chennai), 1.77 (Trivandrum) and 1.82 (Goa). a decreases as a function of altitude in Trivandrum, while a is 2.0 in the m altitude range over Goa. The 9of10

10 higher a values obtained over Bhubaneshwar, Chennai, Trivandrum and Goa suggest the dominance of submicron fine mode aerosols in the altitude region of 0 to 3000 m. These high a values corroborate the higher b values. [44] Asymmetry parameter (g) values corresponding to 30% RH are less than 0.60 over Bhubaneshwar, Chennai, Trivandrum and Goa in the 0 to 3000 m altitude regime. The columnar mean g values are found to be 0.44 (Bhubaneshwar), 0.48 (Chennai), 0.43 (Trivandrum) and 0.38 (Goa) respectively. The highest g value of 0.56 was obtained near surface over Bhubaneshwar while the lowest g value of 0.29 was seen at Goa in the m altitude region. g values obtained over tropical India are lower than those obtained in the continental, remote coastal and marine stations in USA. Higher a and lower g values obtained in the lower troposphere indicate the dominance of smaller size aerosols during the premonsoon season over these locations in India. [45] Scattering aerosols in the 0 to 3000 m altitude region are estimated to contribute about 20 35% to the total aerosol optical depth (AOD) in these four locations. The scattering AODs obtained by integrating the aerosol scattering coefficients in the lower troposphere are higher and nearly the same for Bhubaneshwar and Chennai. Bhubaneshwar AOD is 50% higher than Chennai AOD; this reduces the contribution of scattering AOD to total AOD by half over Bhubaneshwar. Results indicate that the contribution of scattering aerosols in the lower troposphere to AOD can vary, thereby bringing out the importance of spatial and vertical variations in aerosol optical characteristics. [46] Acknowledgments. K. K. Moorthy, Project Director, ICARB and ISRO-GBP Programme Office, ISRO Headquarters, Bengaluru are acknowledged for their support. We are grateful to Kalyanaraman, Raghu Venkataraman, Madhusudhan Reddy of NRSA, Hyderabad pilot and aircraft crew for their help in on board measurements. HYSPLIT (Version 4) model is obtained from ( MODIS AODs are downloaded using NASA Goddard Earth Sciences Data and Information Services Center s Interactive Online Visualization and Analysis Infrastructure. We thank the reviewers for their helpful comments and suggestions. References Anderson, T. L., and J. A. Ogren (1998), Determining aerosol radiative properties using the TSI 3563 integrating nephelometer, Aerosol Sci. Technol., 29, Anderson, T. L., et al. (1996), Performance characteristics of a highsensitivity, three-wavelength total scatter/backscatter nephelometer, J. Atmos. Oceanic Technol., 13, Andrews, E., et al. (2006), Comparison of methods for deriving aerosol asymmetry parameter, J. Geophys. Res., 111, D05S04, doi: / 2004JD Charlson, R. J., D. S. Covert, and T. V. Larson (1984), Observation of the effect of relative humidity on light scattering by aerosols in Hygroscopic Aerosols, edited by T. H. Rubuke and A. Deepak, pp , A. Deepak, Hampton, Va. Chung, C. E., V. Ramanathan, D. Kim, and I. A. Podgorny (2005), Global anthropogenic aerosol direct radiative forcing derived from satellite and ground-based observations, J. Geophys. Res., 110, D24207, doi: / 2005JD Clarke, A. D., et al. (2002), INDOEX Aerosol: A comparison and summary of chemical, microphysical, and optical properties observed from land, ship, and aircraft, J. Geophys. Res., 107(D19), 8033, doi: / 2001JD Draxler, R. R., and G. D. Hess (1998), An overview of the HYSPLIT-4 modeling system for trajectories, dispersion and deposition, Aust. Met. Mag., 47, Grams, G. W. (1981), In-situ measurements of scattering phase functions of stratospheric aerosol particles in Alaska during July 1979, Geophys. Res. Lett., 8, Hegg, D. A., D. S. Covert, M. J. Rood, and P. V. Hobbs (1996), Measurements of aerosol optical properties in marine air, J. Geophys. Res., 101, 12,893 12,903. Hess, M., P. Koepke, and I. Schult (1998), Optical properties of aerosols and clouds: The software package OPAC, Bull. Am. Meteorol. Soc., 79, Kotchenruther, R. A., and P. V. Hobbs (1998), Humidification factors of aerosols from biomass burning in Brazil, J. Geophys. Res., 103, 32,081 32,089. Moorthy, K. K. (2006), Integrated Campaign for Aerosols, Gases and Radiation Budget (ICARB) Operational Plan Document, Indian Space Research Organisation, Bangalore, pp Ramachandran, S., A. Jayaraman, Y. B. Acharya, and B. H. Subbaraya (1994), Mode radius and asymmetry factor of Mt. Pinatubo volcanic aerosols from balloon-borne optical measurements over Hyderabad during October 1991, Geophys. Res. Lett., 21, Remer, L. A., et al. (2005), The MODIS algorithm, products and validation, J. Atmos. Sci., 62, Sheridan, P. J., D. J. Delene, and J. A. Ogren (2001), Four years of continuous surface aerosol measurements from the Department of Energy s Atmospheric Radiation Measurement Program Southern Great Plains Cloud and Radiation Testbed site, J. Geophys. Res., 106(D18), 20,735 20,747. Sheridan, P. J., A. Jefferson, and J. A. Ogren (2002), Spatial variability of submicrometer aerosol radiative properties over the Indian Ocean during INDOEX, J. Geophys. Res., 107(D19), 8011, doi: / 2000JD Wiscombe, W. J., and G. Grams (1976), The backscattered fraction in twostream approximations, J. Atmos. Sci., 33, T. A. Rajesh and S. Ramachandran, Space and Atmospheric Sciences Division, Physical Research Laboratory, Navrangpura, Ahmedabad , India. (rajeshta@prl.res.in; ram@prl.res.in) 10 of 10