Making use of aerosol optical depths (AOD) derived from MODIS (onboard TERRA satellite) and winds from NCEP, and the fact that sea-salt optical depth over ocean is determined primarily by sea-surface wind speed, we examine the contribution of sea-salt to the composite aerosol optical depth (AOD) over Arabian Sea (AS), by developing empirical models for characterizing wind-speed dependence of sea-salt optical depth. We show that at high wind speeds, sea-salt contributes 81% to the coarse mode and 42% to the composite AOD in the southern AS. In contrast to this, over the northern AS, share of sea-salt to coarse mode and composite optical depth is only 35% and 16% respectively. Comparison of the sea-salt optical depth and coarse mode optical depth (MODIS) showed excellent agreement. The sea-salt optical depth over AS at moderate to high wind speed is comparable to the anthropogenic AOD reported for this region during winter.
 Natural aerosol production rate is the highest for sea-salt [Blanchard and Woodcock, 1980]. Blanchard and Woodcock  and Fitzgerald  have shown that bursting-bubbles associated with breaking waves is responsible for the production of sea-salt aerosols. The abundance and size spectrum of these aerosols strongly depend on the over-ocean wind speed, besides the temperature and salinity of seawater [Fitzgerald, 1991]. In this context, the Arabian Sea region has significance due to its tropical nature, which keeps the water warm all through the year, and the wind pattern that reverses direction and undergo changes in amplitude seasonally [Moorthy and Satheesh, 2000]. During December to March the winds over the AS are, in general, weak north easterlies while during June to September winds are strong south westerlies. There have been several ship-borne investigations on the relationship between sea-salt aerosols and wind speed over AS and Indian Ocean [Moorthy et al., 1997; Moorthy and Satheesh, 2000; Vinoj and Satheesh, 2003] and empirical relations have been deduced to model the wind-speed dependence of aerosol optical depth (AOD) over the ocean. Based on a similar study by Smirnov et al.  from Midway Island, Kaufman et al. [2005a, 2005b] estimated the anthropogenic aerosol fraction by combining empirical model relationships with AOD derived from MODIS (onboard TERRA) data. In this paper, we examine the contribution of sea-salt to the composite AOD over the AS using empirical models for characterizing the sea-salt contribution and the latitudinal dependence of wind-independent AOD (non-sea-salt) component. This is a first step toward separating the natural aerosol contribution to the composite and to assess the anthropogenic impact.
2. Estimation of Sea-Salt Optical Depth
 The impact of changes in the wind speed on the columnar AOD (due to changes in sea-spray production) over the AS has been studied by various investigators [Moorthy et al., 1997; Moorthy and Satheesh, 2000; Vinoj and Satheesh, 2003]. These investigations have shown that an increase in the wind speed (U) causes an exponential increase in aerosol optical depth (τa), which could be expressed as,
where τ0 is the optical depth at zero wind condition (wind-independent component) and b is the index for wind speed dependence, the reported values of which are in the range 0.08 to 0.12. Keeping this relation as the base, and making the following assumptions, we attempt to estimate the contribution of sea-salt to composite AOD. (i) Major contributors to coarse mode optical depth over the ocean are sea-salt and dust, (ii) Anthropogenic aerosols contribute mostly only to the fine mode optical depth, (iii) In the absence of dust, coarse mode optical depth over ocean is primarily due to sea-salt aerosols, and (iv) Over open ocean the only wind dependent aerosol component is sea-salt. Even though O'Dowd and Smith  have reported that both sea-salt and sulfate (from di-methyl sulfide) depend on wind speed, the contribution of natural sulfates to optical depth over the open ocean is much smaller compared to that of sea-salt [Fitzgerald, 1991] and hence the above assumptions have a general validity.
 We show in Figure 1 (top panel), the MODIS-derived AOD (at 550 nm) as a function of wind speed over the southern ocean (30°S–40°S; 60°E–70°E) where continental influence is negligible. The monthly mean AODs are obtained from the level-3 product at 1° × 1° resolution for the period 2001–2003 and monthly mean wind speed from NCEP/NCAR reanalysis data for the same period. Figure 1a shows a general conformity to equation (1) as indicated by the solid line, which is regression fitted. The correlation and regression coefficients (ρ, τ0 and b) are also calculated and reported in the figure legend. The value of b thus obtained (0.09), lies well within the earlier reported values from cruise measurements.
 Following equation (1), for a given wind speed (U), sea-salt optical depth is given by,
When U = 0, τSS = 0 and τa = τ0, the non-sea-salt component of AOD.
 Results of a similar analysis performed for the northern (10°N to 15°N) and southern (5°S to 10°N) AS are shown in Figure 1 (bottom panel), along with the respective values of ρ, τ0 and b. The results show that, in the Arabian Sea, the index b of wind-speed dependence has a mean value of 0.09 ± 0.03 and is latitudinal independent (which is logical because the production of sea-salt by wind is expected to be the same irrespective of location), while τ0 has a strong latitudinal variation. This observation is examined in juxtaposition with the earlier observations during ground-based (from islands and ship-borne) investigations; (i) Minicoy island (8.5°N; 77°E) [Moorthy and Satheesh, 2000], (ii) Shipborne measurements over equatorial Indian Ocean [Moorthy et al., 1997], (iii) observations at Kaashidhoo Climate Observatory (KCO; 4.9°N, 73.5°E) as part of Indian Ocean Experiment (INDOEX) [Satheesh et al., 1999], and (iv) measurements over northern and central AS as part of Arabian Sea Monsoon Experiment (ARMEX) [Vinoj and Satheesh, 2003]. These observations indicated a consistent increase in τ0 (from 0.07 to 0.3) with latitude from the equatorial Indian Ocean to the northern AS. Because ship-borne measurements cover only limited area and period, we used MODIS-derived optical depths for the latitude range from 10°S to 15°N and the longitude belt from 60° to 70°E; for the period 2001–2003 and plots (similar to Figure 1) of optical depth versus surface wind speed were made for every latitudinal bin of 5° width, from which τ0 and b were estimated. The latitudinal variation of τ0, shown in Figure 2, revealed an exponential increase with latitude; from a value of ∼0.03 at 10°S and 0.06 near the equator to as high as 0.20 over northern AS (∼17°N). Comparison of τ0 derived from the island/ship-borne measurements at specific locations (described earlier) with the current estimates (Figure 2) shows a good agreement between the two for the corresponding latitudinal bins. The latitudinal variation of τ0 can thus be expressed as,
where Λ is the latitude (in degrees to the north of 10°S). However, the index b did not have any significant latitudinal variation as was seen earlier in Figures 1.
 Using the above parameterizations, we have derived the sea-salt optical depth (τSS) at 550 nm (following equation (2)) at 1° × 1°, independently using the NCEP wind speed data and we have produced monthly maps (averaged for 2001–2003). Here we have used τ0 values at 1° interval estimated using equation (3). This is because the wind independent AOD (which is mainly the anthropogenic component) is known to have an exponential latitudinal variation into the oceans [e.g., Satheesh et al., 1998]. Only a few of these are presented in Figure 3. Here, we used daily-mean wind speed values (NCEP) and derived sea-salt optical depths for each day (for 2001 to 2003) and then averaged the daily values to obtain monthly mean values. We have also overlaid the monthly mean wind vectors on to the sea-salt AOD maps in Figure 3. High values of τSS occur over regions and in seasons of high wind speeds. The τSS values are now compared with MODIS-derived composite aerosol optical depth (τa) as well with the MODIS-derived coarse mode optical depth (τCM). In view of the large seasonal changes in the wind speed and the large spatial variation of τa, and the potential transport of mineral dust during summer season from Arabia to northern AS, we considered two distinct seasons [namely JFMND (January to March, November, and December) when prevailing winds are predominantly from the northeast and JJA (June to August) when prevailing winds are predominantly from southwest/northwest] and two distinct oceanic regions [southern AS (0° to 15°N) which is less prone to continental dust transport and northern AS (north of 15°N), which is influenced significantly by dust transport from west Asia]. An example is shown in Figure 4, for southern AS for the JFMND season, as a scatterplot of τSS versus τCM. A significant correlation is seen between the two (with a coefficient of 0.55 for 1576 data pairs) with τSS contributing to as much 64% to the coarse mode optical depth. Results of similar analysis for the other seasons and different oceanic regions are given in Table 1, which reveals several important points:
Table 1. Share of Sea-Salt to Composite Optical Depth
Share of τSS to τa, %
Share of τSS to τCM, %
 1. During the JFMND period, sea-salt contributes ∼30% to τa (at 550 nm) almost over the entire AS. Its share to τCM is ∼60%.
 2. In total contrast to this, during the JJA period, when the winds are of opposite direction and stronger in magnitude, the sea-salt becomes the most dominant contributor (∼80%) to the τCM and significant contributor to the τa (∼40%) in the southern AS. Not only its share, but the absolute magnitude also significantly increases during this period.
 3. In the northern AS the scenario is somewhat different. Even though the sea-salt AOD increases here also during JJA period, its share to the total, as well as coarse mode optical depths are considerably low (∼20% and 40% respectively).
 Extensive investigation of aerosol optical depths from Minicoy [Moorthy and Satheesh, 2000] have shown that during any year the AOD over the island maximizes in JJA period, despite the strong and extensive monsoon rains over the island and mainland, and that the effect is more conspicuous at the longer wavelengths. They suggested the increased sea-alt production at this time as the possible mechanism to explain this. It may be noticed that the island is located in the southern AS and the results in Table 1 provide a quantitative support to the above argument. Similar observations (of high AOD over AS in JJA period) have also been reported by Satheesh and Srinivasan  and Li and Ramanathan , and attributed this mainly to the transported mineral dust during the summer season. It should also be noted that these observations pertain more to the northern AS. In the absence of dust, sea-salt is the major contributor of coarse mode aerosol, and as such, it dominates both the oceanic regions (north and south AS) during JFMND periods, even though the winds are comparatively weaker and are mainly directed more from India and adjoining regions to the AS. During the JJA period, the winds over the northern AS are strong, and are directed from the west Asian regions, which experience the warmest conditions during the year. Regionally, the winds are strongest around Somali region (Somali jet) and are thus conducive for generating significant amount of dust. As such, a significant fraction of the AOD over this region would come from the mineral dust during this season. Nevertheless, the contribution due to sea-salt continues to be important (Table 1). The INDOEX observations [Satheesh et al., 1999; Ramanathan et al., 2001] have shown that anthropogenic optical depth over northern Indian Ocean is in the range 0.13 to 0.25 during winters of 1998 and 1999. Our study suggests that over the AS natural aerosols contribute substantially to the total optical depth during most part of the year and during the JJA period they might be even dominating the anthropogenic counter part.
 1. We have characterised the contribution of sea-salt to composite AOD over Arabian Sea using empirical models.
 2. At high wind speeds and over southern Arabian Sea, sea-salt contributes as much as ∼80% to the coarse mode optical depths and ∼40% to the composite AOD.
 3. In contrast to this, over the northern Arabian Sea, shares of sea-salt to coarse mode and composite AOD are only ∼40% and ∼20% respectively.
 4. Comparison of the sea-salt optical depth and coarse mode optical depth (derived from MODIS) showed excellent agreement.
 5. The sea-salt optical depth over Arabian Sea at moderate to high wind speed is comparable to the anthropogenic aerosol optical depth reported during winter.
 Authors thank Indian Space Research Organisation's Geosphere Biosphere Programme (ISRO-GBP) for supporting this work. We thank Y. J. Kaufman, NASA/GSFC for valuable suggestions.