Analysis of the continuous and collocated measurements of columnar spectral aerosol optical depths (AODs) and mass size distributions in the marine atmospheric boundary layer (MABL) over the Bay of Bengal (BoB), carried out from 27 December 2008 to 29 January 2009 during the Winter Integrated Campaign for Aerosols, Gases and Radiation Budget (W-ICARB), revealed distinct regional features in the spatial variations of the aerosol properties in the MABL and column. In general, AODs were high over the northern and northwestern parts of the BoB, with pockets of very high values, within which the AODs were as high as ∼0.8 while the smallest values (∼0.1) were observed over the northeastern BoB, off the Myanmar and Bangladesh coasts. Interestingly, though, this region had the highest Angstrom wavelength exponent α (∼1.5), notwithstanding the generally high values that prevailed over the eastern as well as northern coastal regions of India. Back trajectory analyses revealed the significant role of the advected aerosols in the observed spatial pattern. Within the MABL, high accumulation mode mass concentrations (MA) prevailed over the entire BoB with the accumulation fraction ranging from 0.6 to 0.95, whereas very high fine-mode (r < 0.1 μm) aerosol mass fractions (∼0.8) were observed over the northeastern and western coastal BoB adjoining the Indian mainland (where α was high to very high). The vertical distributions, inferred from the columnar and MABL properties as well as from the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations data, revealed better homogeneity in the northeastern and eastern BoB, whereas significant heterogeneity was seen over other regions.
 Atmospheric aerosols, natural or anthropogenic, have important roles in global climate, ecosystem processes, and human health [Forster et al., 2007]. Sea-salt aerosols from the global oceans form the largest single source of aerosol mass in the atmosphere. They play an important role in radiative transfer, both directly and indirectly, and hence have important climate implications [e.g., Charlson et al., 1992; Pandis et al., 1994]. The marine aerosol system has two main components: (1) those advected from the continents and (2) those produced in situ (composed of sea salt and non-sea salt) [Fitzgerald, 1991]. Advection of continental aerosols (both anthropogenic as well as natural) over to the oceanic environments has been extensively investigated and its impacts have been assessed [e.g., Hoppel et al., 1990; Prospero et al., 2003; Moorthy et al., 2001, 2005; Johansen and Hoffmann, 2003; Bates et al., 2004; Zhu et al., 2007]. The net result of such advection and in situ production is the large spatiotemporal heterogeneity in aerosol properties over global oceans, especially those adjoining inhabited continents [Smirnov et al., 1995, 2000, 2002; Sakerin and Kabanov, 2002]. Quantifications of these heterogeneities are important in the better understanding of the dynamics of the aerosols and for accurate estimation of regional radiative forcing.
 Viewed in light of the above, the Bay of Bengal (BoB), a small oceanic region, confined between 80°E and 100°E longitude and 5°N and 22°N latitude and surrounded by densely populated continental landmasses that have complex geographical features and varying anthropogenic activities, assumes regional significance in Asia. Despite its significant role in the Asian summer monsoon, extensive investigations leading to detailed characterization of the aerosol heterogeneities over the BoB and its consequences on regional radiative forcing have been attempted only very recently during the Integrated Campaign for Aerosols, Gases and Radiation Budget (ICARB) [Moorthy et al., 2009]. Even though a few studies exist on aerosol properties over the BoB [Vinoj et al., 2004; Sumanth et al., 2004; Satheesh et al., 2006], they were limited to aerosol optical depths (AODs) over western BoB, adjoining the Indian landmass. The first attempt to examine the BoB in near totality was the ICARB of 2006 [Moorthy et al., 2009; Nair et al., 2008a, 2009]; yet it did not explore the far-eastern and southeastern BoB, which is more anthropogenically affected. Our study aimed to bridge these gaps with concurrent measurements over the entire BoB region bound between 76.6°E and 97.5°E and 3.5°N and 21°N in a span of 34 days. The present study makes a synthesis of concurrent and collocated measurements of columnar AOD and mass size distributions in the marine atmospheric boundary layer (MABL) with simultaneous data on the vertical properties deduced from Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) to examine the heterogeneity in the vertical distribution of aerosols over the BoB and its distinctiveness over different parts of the ocean. Our study assumes significance because knowledge of the vertical distribution of aerosols is very important in estimating the vertical structure of aerosol warming in the lower atmosphere [Moorthy et al., 2009], which has important climate implications. To our knowledge, this attempt is done for the first time over the oceanic region around Asia.
2. Campaign, Instruments, and Data
 During the campaign, the ORV sailed off the port of Chennai on 27 December 2008 and after the expedition called at Kochi on 30 January 2009, following the track denoted by the solid line in Figure 1. During this period, continuous measurements of instantaneous spectral AODs were made on board the ORV using a Microtops sunphotometer (Solar Light Co, USA), at five wavelengths (440, 500, 675, 870, and 936 nm) at regular intervals of 10–15 min during cloud-free periods. The instrument was precalibrated at NASA for use on cruises as a part of the Maritime Aerosol Network (MAN). Latitude, longitude, and time information were fed to the instrument using a GPS receiver attached to it. During each measurement, three scans were made in quick succession after aligning the instrument with the Sun and the least of these values was taken as the AOD for that particular time. Details on Microtops measurement protocols are available in the earlier literature [e.g., Morys et al., 2001; Ichoku et al., 2002; Porter et al., 2001; Moorthy et al., 2009] and these protocols were strictly adhered to during data collection. The collected data were processed following the MAN protocol [Smirnov et al., 2009] and are posted on the MAN web site (http://aeronet.gsfc.nasa.gov/new_web/cruises_new/Sagar_Kanya_09.html).
 Size-segregated mass concentrations of composite (total) aerosols were measured using a 10-stage quartz crystal microbalance (QCM) cascade impactor (model PC-2, California Instruments, USA) having 50% lower size cutoff at each of its 10 size bins at >25, 12.5, 6.4, 3.2, 1.6, 0.8, 0.4, 0.2, 0.1, and 0.05 μm for size bins 1 to 10, respectively. It sampled the ambient air through the community inlet at a constant flow rate of 0.24 Lpm during the sampling interval of 300 s. Measurements were carried out at regular intervals of 30 min, around the clock. Measurements were restricted to the periods of relative humidity (RH) < 78% at the deck level, in view of the affinity of the quartz crystal to changes in RH for higher RH [Pillai et al., 2002; Nair et al., 2008b].
 Calculations of the particle growth factor with RH for typical maritime models [Hanel, 1976; D'Almeda et al., 1991] show that the growth is insignificant for an increase in RH up to 60%. Subsequently, although the particle size increases with increase in RH, the effect becomes substantial only for RH > 80. It is also seen that the growth factor is lower for accumulation mode particles than for coarse mode particles. For values of RH ∼ 78%, the growth factor is in the range ∼1.1–1.2 for particles with radius ≤ 1.0 μm; for coarse particles (r ≥ 1.0 μm), the growth factor is slightly higher. The ambient RH changes were continuously monitored using a hygrometer during the entire sampling period and the observations were discontinued if RH exceeded 78%, particularly in view of the affinity of the quartz to changes in RH for higher RH.
 The total mass concentration (MT) estimated from the QCM and total suspended particulate matter obtained using a high-volume sampler at several instances were compared at locations where these two instruments were operated simultaneously and an agreement in the total mass concentration of better than 80% was obtained [e.g., Moorthy et al., 2005; Nair et al., 2010]. Generally, the uncertainties in the measured MT values are in the range 10%–20% at very low mass concentrations (∼10 μg m−3), and the error reduces to <10% for the high mass concentrations (>30 μg m−3). More details of the instrumentation, analysis, and error budget are given elsewhere [Pillai and Moorthy, 2001; Nair et al., 2009].
3. Prevailing Meteorology
 The prevailing meteorology during the Winter Integrated Campaign for Aerosols, Gases and Radiation Budget (W-ICARB) comprised predominantly calm synoptic conditions with moderate winds and clear or partially cloudy skies. No major weather systems or cyclonic depressions were encountered in the study area during the cruise period, except perhaps the thunderstorm that occurred on the night of 28 January 2009 in the northern Indian Ocean. The prevailing synoptic winds at 850 hPa (derived by the National Center for Environmental Prediction) are superposed on the cruise track in Figure 1. An anticyclonic circulation, centered at 20°N over the Indian landmass, resulted in calm or weak prevailing northerlies and northeasterlies over the northern BoB. Moderate easterlies and northeasterlies prevailed over the middle and southern BoB (below 12°N), with higher wind speeds in eastern parts. This synoptic wind pattern revealed the prevalence of a continental airmass over the entire BoB during the campaign period; however, it originated from distinct geographical regions.
4. Results and Discussions
4.1. Aerosol Optical Depth and Angstrom Coefficients
 The spatial distribution of AOD at 500 nm over the cruise area is shown in Figure 2 (left). The implicit assumption in generating the map is that the temporal variations during the campaign period were far less significant compared to the spatial variations, the validity of which has been ascertained by examining the temporal variations of AOD from the island station Port Blair (PBR in Figure 1) in the BoB. These temporal variations (shown in Figure 2, right) showed that the AOD fluctuated in the range 0.35–0.55 with a mean of 0.43 ± 0.085 with no perceptible trends. The horizontal line in Figure 2 (right) depicts the monthly mean AOD at (500 nm) at Port Blair. The spatial distribution in Figure 2 (left) reveals significant heterogeneity. On the basis of the mean pattern and geographical distinctiveness, the study area has been divided into seven regions, R1 to R7 (as shown in Figure 2), within which the AOD was fairly homogenous but differed significantly between the regions, as will be seen subsequently. The details and criteria followed for these are given in Table 1. Upon examination, while region R1, close to eastern costal India and likely to be most influenced by the Indian mainland, was characterized by moderately high AODs (>0.4), and the highest AODs (going as high as 0.8 at 500 with an average of >0.5) were observed over region R2, in the northern or head BoB. The peculiar nature of this high that is rather detached from the mainland is quite similar to that seen in the earlier ICARB campaign of 2006 [Moorthy et al., 2008; Nair et al., 2008a]. In contrast, the lowest AODs (∼0.1) prevailed in the northeastern R3 region, east of R2. Moderately high values of AOD (∼0.5) prevailed over the entire R4 region, with a small patch of higher AOD toward the southern end, and close to the Malacca strait, a region that has hitherto remained unexplored. This high AOD is attributed to the higher winds that prevailed in this region during the campaign, which would have been conducive for the production of sea-salt aerosols. Moderate AODs prevailed over central BoB (R5), while lower values (<0.3) were observed over the southern BoB and equatorial Indian Ocean (R6) as well as over the coastal regions of peninsular India (R7).
Table 1. Grouping of the Oceanic Region and the Mean Values of AOD, α, and ν for Each Group
Eastern coastal region of the Indian subcontinent bound by the oceanic region ∼13°N–20°N and 80°E–87°E.
0.50 ± 0.05
1.30 ± 0.06
4.10 ± 0.06
Northern BoB, bound by the oceanic region of 15°N–21°N and 87°E–92°E.
0.56 ± 0.02
1.33 ± 0.07
4.06 ± 0.05
Northeastern BoB, the coastal region off Bangladesh and Myanmar, bound by the oceanic region of 14°N–21°N and 92°E–96°E.
0.19 ± 0.03
1.43 ± 0.08
4.12 ± 0.03
Eastern BoB, between the islands of Andaman Nicobar and East Asia, bound between the oceanic region of 6°N–14°N and 92°E–97°E. This region has remained unexplained until W-ICARB.
0.35 ± 0.02
1.33 ± 0.10
4.25 ± 0.04
Central BoB, bound by 6°N–14°N and 84°E–92°E
0.30 ± 0.02
1.09 ± 0.12
3.90 ± 0.04
Southernmost BoB and northern Indian Ocean bound between 3°N–6°N and 83°E–92°E.
0.27 ± 0.03
0.80 ± 0.17
4.04 ± 0.03
Coastal oceanic region near off-peninsular India and Sri Lanka bound between 3°N–9°N and 76°E–81°E.
0.34 ± 0.03
0.64 ± 0.20
4.06 ± 0.05
 The mean AOD spectra, representative of these seven regions, obtained by averaging the AODs within each region over the given period are shown in Figure 3 (left). Measurements of spectral AODs were also carried out continuously over the mainland coastal locations Trivandrum (TVM), Chennai (CHN), Visakhapatnam (VSK), Bhubaneswar (BBR), and Port Blair (PBR; Figure 1) concurrently with the cruise and the average AOD spectra (for the cruise period) for these regions are shown in Figure 3 (right). Five-channel Microtops instruments (compared with the one used on board) were used at Bhubaneswar and Chennai, and 10-channel multiwavelength radiometers (MWRs) were used at Trivandrum, Visakhapatnam, and Port Blair. The MWR provided mean AOD for forenoon and afternoon parts of the day at 10 channels (380 to −1025 nm) following Langley plot techniques, the details of which are available in several earlier papers [Saha et al., 2005; Gogoi et al., 2009]. The MWR-derived AODs agreed well with the mean AOD estimated from the concurrent Microtops measurements for the same station, as ensured during comparison. During the initial period of the campaign, an average AOD of 0.47 ± 0.03 (at 500 nm) was observed at Chennai, which was comparable to those observed over the nearby oceanic regions (Figure 2, left). At Visakhapatnam, the averaged AOD at 500 nm for the first half of the campaign was 0.73 ± 0.06. Examination of the cruise data over R1 (Figure 2, left) revealed a similar increasing trend in AOD from off the coast of Chennai to off the coast of Visakhapatnam. The mean AODs over the coastal location of Trivandrum (0.34 ± 0.01) and the island station Port Blair (0.38 ± 0.04) were also comparable to those measured over the nearby oceanic regions during the campaign period.
 Examining the average spectral AODs for the seven ocean regions (Figure 3), we note that even though high AODs occurred over R1 and R2, the steepest spectra were observed over R3, followed by R2, R4, and R1, implying significant contribution of accumulation mode aerosols over the eastern BoB, followed by northern-head BoB, eastern, and then western coastal regions of BoB. At the mid-ocean regions (R6 and R5) and at coastal Trivandrum (R7), the spectra tend to become flatter and depict high values of AOD at longer wavelengths. The spectral steepness over regions R1, R2, and R7 are comparable to that of the nearby coastal and island stations. By performing a regression analysis of the spectral AODs (in log-log scale) with the Angstrom relation τ = βλ−α, the Angstrom wavelength exponent α and turbidity coefficient β were estimated for each cruise-based measurement as well as for the data from the land stations.
 The spatial variation of α and β are shown in Figure 4. The regional mean values of AOD at 500 nm and α are given in Table 1. Over the entire oceanic regions that have proximity to the mainland or islands (regions including R1, R2, R3, and R4), α remained high (>1.2), with the highest values (as high as 1.5) occurring over R3 where, interestingly, the AOD values were very low. Another region of high α was observed at R1, off Bhubaneswar, where AODs were also high. Even over the mid-ocean regions of BoB, frequent occurrence of α > 1.0 were observed, except at a few patches centered at R5 and R6. The smallest values of α are observed over R7, except at coastal Trivandrum. Upon examination of the coastal landmass measurements, it is noticed that high values of α prevailed over Visakhapatnam, where α was higher than that seen over the adjacent oceanic region R1, whereas at Bhubaneswar the values of α over the mainland were lower than those prevailing over the adjacent oceanic region R2, showing mixing of different types of aerosols in these regions. On the other hand, at Port Blair and Trivandrum, the mean α values were comparable to those seen over the nearby oceanic region. The Angstrom turbidity coefficient β followed a pattern similar to that of AOD (Figure 4, right). Values of β remained generally high (∼0.2) for the entire R1 and R2 regions and the highest values (∼0.3) occurred in R2, indicating high columnar loading over the region. The advected coarse dust from northwestern India across the Indo-Gangetic plains (Figure 5) would enhance the coarse mode aerosol dominance over R1 and R2. It (β) also remained high partly over R5, R4, and R7; these patches of highs are found to be associated with regions of high wind speed. A similar pattern of isolated highs is seen also in the QCM measured coarse mode mass concentrations in the MABL (Figure 6, top). The smallest values of β occurred throughout region R3 (∼0.05), where the columnar loading was low and mainly by fine and accumulation mode aerosols, resulting in steep AOD spectra.
 With a view to delineating the role of transported (long-range) aerosols toward the observed spatial pattern, we examined the 5 day isentropic airmass back trajectories arriving at each of the seven regions at three height levels: 500 m (within the MABL), 1500 m (above MABL), and 3500 m (free troposphere). The most dominant pathways have been identified following a cluster analysis [Hafner et al., 2007] and the results are shown in Figure 5. The corresponding mean values of AOD, α, and β are also shown in Figure 5. The analysis brings out the following:
 1. The trajectory clusters arriving at R1 have very long continental overpass through the arid regions of western Asia and northwestern India before arriving at R1, across central India. These trajectories would thus be conducive for the significant advection of transported dust aerosols to the measurement region, which would significantly add to the existing aerosol abundance in the region (with significant anthropogenic components owing to the local human activities discussed earlier). On the basis of the observations from several land stations over the Indian subcontinent as well as over the island locations of Port Blair and Minicoy, Beegum et al.  have shown that the advected mineral dust aerosols from the arid western Asian regions lead to significant enhancement in the column AOD over the Indo-Gangetic plain (IGP). Furthermore, because the transported mineral dust is known to have a dominant submicron mode at ∼0.5 μm [Hess et al., 1998], it would contribute toward the accumulation mode aerosols also and thereby to the higher values of α. The region R1 adjoins the highly urbanized and industrialized eastern coastal India with three major harbors, Chennai, Visakhapatnam, and Paradweep. Consequently, significantly high anthropogenic aerosol concentration would prevail over these regions. The observed high values of AOD and moderately high α at R1 is the consequence of these factors.
 2. At R2, where the highest values of AODs are observed, the trajectories were found to have a wide spread over the Indian subcontinent, covering all of central India and the IGP and providing conduits for transport of continental aerosols, mostly of mixed type (anthropogenic + natural), to the measurement location, which would contribute significantly to the observed high values of AOD. The IGP, accounting for ∼21% of the land area of the Indian subcontinent, is a densely populated region with a high degree of anthropogenic activities [Nair et al., 2007] such as coal-based power plants, industries, transport, mining, and urban and agricultural activities. Several investigators have also reported the persistence of high AOD and large concentrations of aerosols over this region during winter months [Jethva et al., 2005; Ramana et al., 2004; Singh et al., 2004; Prasad et al., 2006]. The prevailing westerly winds, combined with the orography of the IGP that spatially confines the aerosols into a rather narrow channel and leads to the outflow into the northern and head BoB, results in the large loading of aerosols in that region [Girolamo et al., 2004; Nair et al., 2007; Niranjan et al., 2007].
 3. The trajectories coming to R3, where the lowest AOD and highest values of α were observed (during the cruise period), arrived here mainly from eastern coastal India and would be mostly devoid of significant natural components. In addition, the eastern coastal belt of India, comprising several ports and industries, is a hot spot of accumulation mode aerosols [Moorthy et al., 2005; Niranjan et al., 2005]. This is indicated by the very high α values observed over Visakhapatnam and Bhubaneswar during W-ICARB. Because the trajectories arriving at R3 have a long history of more than 4 days (∼1000 km) over the ocean, there would be significant reduction in the aerosol abundance by subsidence and dispersion, leading to a decrease in the loading and an increase in the spectral steepness.
 4. Even though moderate values of AOD are observed at R4, in the eastern BoB, α remained high (comparable to R2 and R3) and the trajectory analysis revealed significant advection from the continental locations of East Asia and south China. It suggests that fine and accumulation mode particles advected from these regions contributed significantly to the aerosols over R4. Several earlier studies from the BoB have demonstrated the role of advection from East Asia and south China in the enhancement of the AOD and its spectral steepness, as well as black carbon mass concentrations in the MABL over Port Blair [Moorthy et al., 2003; Moorthy and Babu, 2006; Nair et al., 2009; Vinoj et al., 2009]. Streets et al.  have reported that the anthropogenic aerosol emissions have increased in recent decades in East Asia because of the increase in urban activities, and this finding corroborates our inference.
 5. The trajectories reaching R5 were found to have originated from the western coastal regions of East Asia, which might transport anthropogenic aerosols along with the coarse sea-salt aerosols, because these trajectories have traversed across the vast oceanic regions where the wind speeds were moderately high (Figure 1). This would result in a mixed type of aerosols (coarse mode sea salt + accumulation mode anthropogenic aerosols) and explains, at least qualitatively, the moderate values of AOD and fairly low values of α.
 6. The trajectories arriving at R6 and R7 were mostly oceanic in nature. They would be mostly characterized by a smaller amount of accumulation mode aerosols and a larger amount of coarse mode sea salt because the marine aerosols generally have a coarse mode associated with sea spray. This would result in the moderate values of AOD and low values of α. As the southern BoB opens to the vast Indian Ocean, anthropogenic aerosol concentration would be insignificant.
4.2. Column Versus MABL Characteristics
 The shallow MABL in the northern BoB during winter (due to reduced solar heating, especially in the northern latitudes) favors the confinement of aerosols near the surface and, as such, these would contribute significantly toward the column AOD, whereas the situation would differ in the southern latitudes, where tropical conditions prevail. As such, it is important to examine the characteristics of MABL aerosols vis-à-vis those of the columnar aerosols. With a view to this, we examined the QCM data.
4.2.1. Concentrations From QCM
 From the QCM measured mass size distributions of composite aerosols in the MABL, we have estimated the coarse mode (particles with radius larger than 1 μm) mass concentration MC using the equation MC = mci and accumulation mode (particles within the radius range of 0.1 to 1.0 μm) mass concentration MA using the relation MA = mci, where mci is the mass concentration at the ith size bin of the QCM. From these values, we estimated the accumulation fraction (FA), which is the ratio of the accumulation mode mass concentration to the total: FA = . The spatial variation of MC, MA, and FA are shown in Figure 6 (top, middle, and bottom, respectively). The important findings are the following:
 1. Isolated pockets of high values of MC are observed over R1, R2, and R5 with values as high as ∼25 μg m−3 and moderately high values are observed (15–20 μg m−3) over the southern part of R4 (near the southern part of the Andaman Nicobar islands), which is in line with the earlier discussions.
 2. Very high values of MA are (reaching ∼50 μg m−3) observed over R1 and R2 and moderately high values (∼40 μg m−3) over the northern part of R4.
 3. Over R3, where the highest values of α were observed, the values of MA were found to be very low (∼5–15 μg m−3), which were comparable to or even lower than that over the mid-ocean regions of R5 and R6. This might be either because of the large difference in aerosol properties between the column and surface or because the size range contributing toward the total mass might be finer when compared with the accumulation mode size range. Because the column AOD itself remained low over there, the contribution of the latter might be more significant. This aspect is examined subsequently.
 4. The accumulation mass fraction was generally very high (>0.6) over most of the BoB. Extremely high values of accumulation fraction (>0.9) are observed over R1 and R2. It is interesting to note that, over R3 and R5, where MA values were low, the accumulation fraction is found to be fairly high (∼0.6); however, it is less than those observed over R1 and R2.
 5. Low values of MA prevailed over R7, except near coastal Trivandrum.In short, during the campaign, over the entire oceanic region, the anthropogenic aerosol concentration remained high in the MABL, and this feature is in conformity to the findings reported earlier over BoB [Nair et al., 2008a]. Over the regions R1, R3, and the northern part of R4, where α was high, MA and FA were only moderately high, implying the presence of very fine mode (r ≤ 0.1 μm) particles. To examine this, we have estimated the fine mode aerosol mass fraction (FF, which is the ratio of the sum of the mass concentrations at the lowest two size bins, 0.05 and 0.1 μm, to the total mass concentration), and its spatial distribution is shown in Figure 7. It reveals high fractions over the coastal regions of R1 (reaching as high as 0.8), the entire region of R3, as well as the northern part of R4. In contrast, the values of FF were very low over R2, as well as over the other mid-ocean regions (<0.3). The high values of α, low MA, and high FF clearly reveal the fine mode aerosol abundance over R3 compared to accumulation and coarse modes. These particles, although they contribute significantly to AOD at the shorter wavelengths (thereby leading to a steeper AOD spectrum and high α), would not contribute significantly to the aerosol mass (owing to their small sizes) and that explains, at least logically, the observed features in R3. The observations also indicate the prevalence of similar airmass types over R1 and R3, in line with the earlier finding of potential advection of aerosols from eastern India over R3 (as indicated by the trajectories in Figure 4).
 With a view to examining the association between the surface measured fine mode aerosol mass concentration and the spectral steepness of the column AOD, Figure 8 shows a scatterplot between the QCM-estimated FF and the Angstrom wavelength exponent α from the column spectral AOD. It reveals fairly good linear association with a correlation coefficient of 0.52. To delineate the association between the two at different regions mentioned above, we repeated the preceding analysis for the respective regions mentioned above, and the corresponding correlation coefficients are given in the Table 2. It is interesting to note that the correlation coefficient shows large spatial variation, with very good correlation over the eastern BoB (R4) followed by the northeastern BoB (R3) and R1 (the western coastal region of the BoB). On the other hand, over the head BoB (R2), as well as over central BoB (R5), the correlations were poor, suggesting significant vertical heterogeneity.
Table 2. Correlation Coefficients in the Scatterplots of Fine Mode Fraction (FF) Versus α for Different Regions
Number of Data Pairs
4.2.2. Number Size Distributions
 The number size distributions were deduced from each of the QCM samples and segregated in terms of the regions and averaged. The regional means are shown in Figure 9. Significant regional differences are seen in the particle size regimes of diameter <0.8 μm and >3 μm, whereas the differences are far less conspicuous in the intermediate size regime. The primary mode is not discernible in the size distribution, probably because it might be occurring below the lower size limit of the QCM (0.05 μm). The secondary mode is barely distinguishable in the winter season because the size distribution is dominated by fine and accumulation mode aerosols. But when there is a significant contribution from dust or sea salt, this mode becomes conspicuous, as was observed during an earlier campaign [Nair et al., 2008a]. In the present study, such a distribution is seen over R5, where the sea-salt aerosol concentration was significant. Similar size distributions are also reported over oceanic regions elsewhere [Iwasaka et al., 1996].
 From each of the QCM measured size distributions, we estimated the effective radius, Reff, and the ratio of the volume to area of the aerosol distribution [Pillai and Moorthy, 2001], considering spherical particles of density 2 g m−3 (intrinsically assumed in QCM-based mass measurements); the regional mean values of estimated Reff are also given in Figure 9. For maritime aerosols, the density varies between 1.8 and 2.2 g m−3 [Hanel, 1976; Pruppacher and Klett, 1978].
 It is interesting to note that the number concentrations at the lowest two size bins showed considerably higher values over R3 and R4, and the corresponding regional mean values of Reff were the lowest. Higher number concentrations also persisted over the shorter size bins of 0.05 and 0.1 μm in the region R1, and this led to the fairly low value of Reff, which was still higher than the value at R3. It is interesting to note that, even though the number size distribution followed a similar pattern over R3 and R4, a considerable increase is observed in the number concentration at ∼0.4 μm over R4, which led to the comparatively higher Reff over R4 than over R3. Over the mid-ocean regions, a considerable increase in number concentrations at larger size bins is observed, probably due to the increase in sea-salt aerosol concentrations.
 The average size distributions are parameterized using an inverse power law distribution of the form n(r) = n0r−ν, where n(r) is the number concentration of aerosols in an infinitesimal radius dr centered at r, n0 is a constant depending on the total aerosol concentration, and ν is the power law index. The regional average values of ν were estimated and are given in Table 1. The values of ν varied between 3.9 and 4.25, with higher values over the regions of R3 and R4. Even though both the parameters (Reff and ν) suggest anthropogenic aerosol dominance throughout the BoB, the relative dominance of the accumulation mode aerosols (0.1 to 1 μm) over the fine mode aerosol (0.05 and 0.1 μm) concentrations varied regionally. Among the seven distinct regions, the fine mode aerosol concentrations were found to be highest over R3, followed by R4 and R2. Even though a reduction in the values of ν and a consequent increase in Reff are observed at the mid-ocean regions, the values still depict significant amounts of anthropogenic aerosols in the MABL.
4.3. Latitudinal Gradients
 The latitudinal variations (values averaged over the longitude range 75°E–100°E for each latitude from 2°N to 21°N) of MT, MA, FA, α, and AOD are shown in Figure 10. The standard deviations of the means are represented by the vertical lines though the points (which represent the means) in each of the panels. All the parameters showed significant latitudinal gradients with an increasing trend from south to north. The increase, in general, appears to follow an exponential form from the near-pristine environments of southern BoB to the northern BoB; however, the steepness varied. The values of MA and MT showed the same nature of latitude dependence with exponential increase from ∼2°N to ∼20°N. For MA, a fourfold increase is observed toward the head BoB, whereas for MT, the increase is more or less gradual by a factor of 1.5. As such, we parameterized the latitudinal variation with an exponential growth function of the form
where x is the latitude in degrees, χ(x) represents the mean value of the parameter concerned at the latitude x, χ0 is the latitude-independent component of χ, and xD is the scaling distance for an e-fold increase in the value of χ in degrees latitude. While both MA and FA depicted steep gradients, in accordance with equation (1) with respective squared correlation coefficients of 0.96 and 0.92 and e1 scaling distance of ∼18°, the Angstrom exponent α showed a less conspicuous spatial variation. It increased from 3°N to 10°N, but the values remained more or less constant thereafter. The latitudinal variations of AOD (500 nm) showed low values (∼0.2) at the equatorial Indian Ocean region and a gradual increase to reach ∼0.6 at the head BoB (shown in Figure 10 by the scatterpoints). When we parameterized the latitudinal variation of AOD with equation (1), it revealed a squared correlation coefficient of 0.88 and an e1 scaling distance of 10.6°. The standard deviations of AOD also showed an increase toward the north, implying larger spatial (longitudinal) variation in the northern BoB.
4.4. Vertical Heterogeneity
 The dynamics of the MABL plays an important role in the vertical distribution of aerosols over any given region even though it is weaker over the ocean than over the landmass. If the dynamics associated with the regional-scale weather occurs over a large spatial extent, it could cause the columnar and MABL aerosols to behave similarly, leading to a vertical homogeneity. On the other hand, short-scale weather phenomena, long-range transport of aerosols above the MABL, and stratified turbulences would lead to vertical heterogeneity in the aerosol profile. With a view to examining these, we plotted the temporal variations of mass concentrations in the MABL (MT) with the column AOD over different regions, and the results are shown in Figure 11. Over R1 and R2, both the parameters were found to correlate poorly, with correlation coefficients of 0.42 and 0.37, respectively. This clearly indicates that, over these regions, the MABL and the column properties differed significantly and hence there exists vertical heterogeneity in aerosol properties. Earlier lidar observations off the coast of Bhubaneswar (during the premonsoon season of 2006) have clearly shown the presence of elevated aerosol layers in the altitude region of ∼2–3 km, extending ≥150 km offshore from the coastline [Babu et al., 2008; Satheesh et al., 2009] with the elevated layers contributing as much as 50%–60% of the mean to the column AOD. In light of the preceding observations, it appears that the vertical heterogeneity over these regions is a persistent phenomenon. However, interestingly, over R3 and R4 we have observed excellent linear dependence between MT and AOD with correlation coefficients of 0.94 and 0.98, implying a rather homogenous vertical distribution. We have calculated the exception p values using statistical methods [Fisher, 1970] for the correlation coefficients over distinct regions. For regions R1, R3, R4, and R5, p was better than <0.0001 (that is, the correlation coefficients are strongly significant with better than 99.9% confidence) and for R2, the p = 0.02 and is significant at the 98% confidence level. This finding, along with the low values of AOD and the vertical structure of convective mixing in the ABL in these regions as revealed from concurrent GPS-aided radiosonde ascents from the ship, suggested the absence of any distinct elevated layers over these regions; the aerosols were mostly confined within the ABL with very little abundance above. The observed lowest values of AOD, high α, as well as high fine mode fractions FF over R3 also corroborate this finding. Over R4, the observed MT values were high, reaching up to ∼60 μg m−3, and thus contributed significantly to the observed moderate values of AOD over that region. The regression analysis over the mid-ocean regions of R5, R6, and R7 together revealed a statistically significant linear dependence with a correlation coefficient of 0.63 between MT and AOD, implying a fairly good vertical heterogeneity. To vindicate these inferences based on concurrent vertical profiles of aerosols over BoB, during the study period, we examined the CALIPSO data.
 CALIPSO, one of the satellites in the A-train constellation [Stephens et al., 2002], provides the vertical distribution of aerosol backscatter and extinction. The extinction coefficients at 532 nm were estimated from the altitude profiles of extinction coefficients by weighting the integrated extinction coefficient with the corresponding AOD measured using Microtops (interpolated to 532 nm using the Angstrom equation) for each of the ship locations, and the area-averaged normalized extinction coefficient profiles for all seven regions were generated. These are shown in Figure 12 for the altitude regions of 0 to 5 km, where the features are distinct. It is interesting to note from Figure 12 that the profiles representing regions R1 and R2 are far separated in absolute magnitudes as well as in the vertical distribution from the profiles for the rest of the regions. Over R2, a prominent aerosol layer is observed at ∼0.5 km and another less prominent one at ∼2 km. Similarly, over R1, two distinct layer structures are noticed: one at ∼1 km and the other at ∼2 km, both being equally prominent. These are identified by the horizontal arrow marks in Figure 12. This feature is quite consistent with the aircraft-based observations taken off the east coast of India by Satheesh et al.  during ICARB 2006. Over the regions R3 and R4, the profiles did not indicate the presence of any elevated layer structure and the values of extinction remained nearly steady up to ∼2 km, vindicating the earlier conclusion drawn from Figure 10 of rather homogeneous vertical distributions there, compared to other regions of the BoB. Over the region of R5, even though the extinction coefficient values remained constant up to 2 km, layered structures were visible at higher altitudes (>3 km). These layer structures, examined in light of the back trajectory cultures (Figure 5), would be composed of different types of aerosols, characteristic of the region from which they are transported, thereby adding to the heterogeneity. These findings ascertain the role of elevated aerosol layers in bringing in heterogeneity in aerosol spatial distribution over the BoB.
4.5. Comparison With Earlier Observations Over the BoB
 As mentioned in the beginning of this paper, W-ICARB focused on the winter season, with a view to examining the seasonal distinctiveness by comparing it with the premonsoon observations of ICARB 2006. Such delineation did not exist for the BoB except perhaps for the observations from the island location of Port Blair [Moorthy et al., 2003]. The spatial variation of AOD at 500 nm during the premonsoon season of 2006 (ICARB 2006) is shown in Figure 13. It revealed the persistence of extremely high values of AOD in the northern BoB, especially due north of ∼15°N; the occurrence of a blob of very high AOD (with values as high as 1.0) at ∼17°N and 87°E and ∼2° away from the eastern coastal India; moderate values of AOD over the entire coastal region and central region of the BoB; and comparatively lower values over the southern and eastern BoB, all salient features. With a view to examining the temporal changes in aerosol properties from winter to premonsoon, we have examined the spatial variation of the difference in AOD (ΔAOD = AOD during W-ICARB − AOD during ICARB) and the similar change in the Angstrom exponent, Δα, between the two seasons. The results are shown in Figure 14 (left and right, respectively). Examining the 3 years of AOD (2006–2008) data from the island location of Port Blair (Figure 15), it is seen that, despite a gradual increase in AOD from 2006 to 2008, the mean AODs for winter seasons (December to February) are significantly lower than values for the premonsoon seasons (March to May). However, examination of Figure 13 (left) shows that a similar feature was noted during W-ICARB only in the northern (head) BoB and in the southern parts, ∼5°N. Over a large region at and south of PBR, the AODs were higher during W-ICARB than during ICARB. A small pocket of higher AOD occurred off Bhubaneswar during W-ICARB as well. Over the rest of the major portion of the BoB, the AOD during W-ICARB was nearly the same as that during ICARB of the premonsoon season of 2006. Positive values of ΔAOD (Figure 14, left) are observed in the southern region, especially near the Andaman Nicobar Islands, with a clear longitudinal gradient with an increasing trend toward the east. This is an indication of the advected accumulation mode aerosols from East Asia. High negative anomalies are observed off the Myanmar coast (over R3) as well as over part of R2 and R5. The size distribution, inferred from α, also showed large deviation from winter to premonsoon seasons, as evident from the spatial pattern of Δα (Figure 14, right). The spectra were much steeper during W-ICARB over the entire oceanic region east of 85°E, with a peak value of Δα of ∼0.8 near the Andaman Nicobar Islands and with a distinct longitudinal variation depicting a definite increasing trend toward the east. Over the entire eastern BoB, positive values with Δα > 0.3 are observed with more and more positive values toward south. This is quite in line with the earlier observations by Moorthy et al. , who reported an increase in the steepness of the AOD spectrum at Port Blair during winter, when the station was under the influence of increased advection from East Asia and south China regions. Satheesh et al.  also reported that the major source of aerosols over the northern BoB is the eastern coast of India and central India, whereas the transported aerosols from East Asia during winter and from the Arabian Sea during summer were found to modulate the aerosol properties over the southern BoB and Port Blair. Several investigators have reported that the aerosol properties over Port Blair are mostly influenced by the advection from East Asia, the Indian subcontinent, as well as from the Arabian Sea and the equatorial Indian Ocean [Moorthy et al., 2003; Satheesh et al., 2006; Moorthy and Babu, 2006]. During the winter season, anthropogenic aerosol dominance is observed over the region (high α) as the increased advection from the East Asian region dominates, whereas during the premonsoon season, mixed types of aerosol prevail because the trajectories were mainly from the Indian subcontinent and the Arabian Sea and equatorial Indian Ocean, leading to relative dominance of coarse mode aerosols. These cause comparatively lower values of α and higher AOD values.
 On the other hand, the western BoB, which is under the influence of advection mainly from the Indian mainland, showed lower values of α during winter, suggesting that this airmass transported more coarse, natural aerosols. Thus, the BoB is under the strong influence of concentrating source impacts on its eastern and western regions, which also undergo several changes. A comparison of the present observations with earlier reports is given in Table 3.
Table 3. Comparison of the Present Observations With Earlier Reports
Campaign and Cruise Number
Period of Campaign
Summary of Results
SK 161 B
NW Bay of Begnal (BoB)
Mean AOD of ∼0.6 over the western coastal BoB and decreases toward the central region to reach ∼0.2. High values of α (as high as 1.8) at the coastal region with a mean value of ∼1.20 [Satheesh et al., 2002].
AODs in the range 0.3–0.6 (500 nm), with an average value of 0.41. The average AOD for the western coastal BoB was 0.54 and was found to decrease toward the mid-BoB (0.30). The average value of α at the coastal region was ∼1.12 and at the mid-BoB was 1.10 [Vinoj et al., 2004].
The values of AOD (at 500 nm) were in the range 0.2–0.7 with a mean value of 0.43. The average value of α was 0.93.
Entire BoB and northern Indian Ocean
AODs (at 500 nm) in the northern BoB were in the range 0.3–1.2, with a mean value of 0.49 ± 0.01; over the south, these are in the range 0.1–0.9 with a mean value of 0.29, and over northern Indian Ocean values lie between 0.1 and 0.6 with a mean value of 0.28. AODs were extremely high at a large region of approximately 3° × 3° in size and centered about 17.4°N, 87.1°E with a mean value of 0.87. In general, the entire BoB exhibits moderate (≥1.0) values of α AOD except in the southeast BoB (0.5). The highest values (1.3) were observed over the eastern coastal region of India [Nair et al., 2009].
W-ICARB (Present Study)
Entire BoB including eastern BoB
Dec 2008 to Jan 2009
High values of AOD were observed over the northern and northwestern part of BoB with “two detached highs” in which AODs were as high as ∼0.8, while the lowest AOD values (∼0.1) were observed over the northeast BoB (Myanmar and Bangladesh coast). In the same region, the Angstrom wavelength exponent α showed the highest values of ∼1.5, even though generally high values prevailed over the eastern as well as northern coastal regions of India (present study).
 Continuous and collocated measurements of columnar AOD and surface mass size distributions were made on board the oceanic research vessel for the entire BoB during the period of W-ICARB (from 27 December 2008 to 29 January 2009). The observations revealed distinct spatial variations throughout the region. The main results are summarized as follows:
 • The spatial pattern of column as well as near-surface measurements of aerosol characteristics showed large variability. High values of AOD were observed over the northern and northwestern part of the BoB with “two detached highs” in which AODs were as high as ∼0.8, while the lowest values of AODs (∼0.1) were observed over the northeast BoB (Myanmar/Bangladesh coast). In the same region, the Angstrom wavelength exponent, α, showed the highest values of ∼1.5, even though generally high values prevailed over the eastern as well as northern coastal regions of India. Seven-day back trajectories were estimated to identify the origins and transport pathways of airmasses over different regions of the BoB and these revealed the significant role of the advected aerosols in the observed spatial pattern.
 • The near-surface measurements using QCM revealed high accumulation mode mass concentrations (MA) over the entire BoB with the accumulation fraction ranging from 0.6 to 0.95. The fine mode fraction was very high (∼0.8) over the northeast BoB as well as over eastern coastal India.
 • All the aerosol parameters (both surface and column measurements) in the MABL showed an exponential increase from the rather pristine equatorial Indian ocean region to the northern head BoB.
 • Homogeneous vertical distribution of aerosols were observed over the northeastern and eastern BoB.
 This study was carried out as a part of the ICARB project of ISRO-GBP. The authors gratefully acknowledge the CALIPSO teams for their effort in making the data available through the NASA Langley Research Center Atmospheric Science Data Center as well as the NOAA Air Resources Laboratory for the provision of the HYSPLIT transport and dispersion model and the READY website (available online at http://www.arl.noaa/gov/ready.htm).