Modulation in Direct Radiative Forcing Caused by Wind Generated Sea-Salt Aerosols

Sea-salt aerosols, prominent natural aerosols over the ocean, play a vital role in direct and indirect radiative forcing. Since surface winds are the prime cause of sea-salt generation, we have developed an empirical relationship between aerosol optical depth (AOD) and sea-surface winds over the study region (60–70°E; 40°S–20°N). The latitudinal variation of background aerosol optical depth (τ0) and the wind index (b) are estimated as they are essential inputs for the estimation of the spatial variation of sea-salt aerosols and are used over the Arabian Sea (AS) to generate spatial heterogeneity of seasalt aerosols. The latitudinal variation of τ0 and b show a nearly exponential and linear increase, respectively, as we moved towards the north. We used an empirical-cum-model approach to construct an aerosol system to reproduce the observed AOD and aerosol optical properties. Utilizing this information as input to a radiative transfer model, we worked out direct radiative forcing (DRF) over the study region in the short wave (0.2–4 μm) and long wave (8–14 μm) region at the surface, top of the atmosphere (TOA) and in the atmosphere. Short wave cooling at the surface, TOA and heating in the atmosphere are estimated as 40 W m, 32 W m and 8 W m, respectively. Long wave heating due to the sea-salt aerosols estimated at the surface, TOA and in the atmosphere is about 9 W m, 6 W m and 15 W m respectively. Long wave forcing (LWF) partly counterbalances the effect of short wave forcing (SWF) and the cooling at the surface and at TOA. The highest value of such an offset at the surface was observed over the AS (~23%) and that at TOA was ~19%, obviously at high wind conditions. This implies that over the AS sea-salt aerosols (in coarse mode) contribute significantly to LWF compared to other adjacent oceanic regions.


INTRODUCTION
The most omnipresent natural aerosols over the oceanic region are sea-salt aerosols (Blanchard and Woodcock, 1980;Winter and Chylek, 1997;Haywood et al., 1999;Randles et al., 2004;Satheesh and Moorthy, 2005).Aerosols are one of the decisive components for the radiative forcing of the Earth system (IPCC, 2013), and sea-salt aerosols constitute one of its major components.Extensive measurements reveal that processes associated with the bursting of the white-cap and wave breaking primarily generate sea-salt aerosols.
Concentration of sea-salt aerosols strongly depends on sea-surface wind speed and its mechanism of generation is discussed by several investigators, though this dependence showed large spatial and temporal variability (Stuhlman, 1932;Kohler, 1936Kohler, , 1941;;Nair et al., 2005;Grythe et al., 2014).Sea surface temperature is also an important factor in sea salt aerosol production, and investigators have reported the impact of sea surface temperature on its production (Jaeglé et al., 2011;Spada et al., 2013;Grythe et al., 2014).
For an accurate estimation of the radiative effects of seasalt aerosols, information about their abundance is required.Several campaigns and satellite data analyses have been performed to study the concentration and radiative effects of aerosols (Moorthy and Satheesh, 2000;Vinoj and Satheesh, 2003;Wai and Tanner, 2004;Satheesh and Moorthy, 2005;Satheesh et al., 2006).Various relationships between these parameters have been reported as the outcome of these studies, which include exponential, linear and power-law type relations (Woodcock, 1953(Woodcock, , 1957;;Monahan, 1968;Lovett, 1978;Kulkarni et al., 1982;Gras and Ayers, 1983;Exton et al., 1985;Marks, 1990;O'Dowd et al., 1997;Moorthy and Satheesh, 2000;Vinoj and Satheesh, 2003;Wai and Tanner, 2004;Satheesh et al., 2006).Over oceanic regions adjacent to the Indian subcontinent also, the dependence of sea-salt aerosols mass/number concentration on wind speed has been scrutinized by several researchers (Moorthy and Satheesh, 2000;Vinoj and Satheesh, 2003;Wai and Tanner, 2004;Satheesh et al., 2006).Jaeglé et al. (2011) combined station-based observations, satellite-estimated parameters and model simulations to estimate the global distribution of sea salt aerosols.Some investigators have used various realtime measurements and modeled sea salt aerosol distribution in chemical transport models to estimate the global sea-salt aerosol distribution (Spada et al., 2013).Several investigators also estimated the sea-salt aerosol mass burden with various model simulations, and an inter-comparison of these results showed large inter-model variations (Textor et al., 2006;Jaeglé et al., 2011).
Model simulations include various parameterization schemes, making inter-comparison exercises a challenging task (Jaeglé et al., 2011;Jiménez-Guerrero et al., 2011;Spada et al., 2013;Struthers et al., 2013;Tsigaridis et al., 2013;Gythe et al., 2014).Fan and Toon (2011) attempted to develop a global parameterization scheme for sea-salt aerosol mass/number concentration and optical properties, and they also verified the quadratic dependence of sea-salt aerosol optical depth on wind speed.In recent work, Grythe et al. (2014) used a Lagrangian particle dispersion model, comprised of observed sea-salt aerosol concentration from various sources, and examined/evaluated various source functions.They also demonstrated that wind speed in the 5-14 m s -1 range is responsible for ~80% of global sea-salt aerosol generation and that sea-salt aerosol production exhibited a power law dependence on wind speed.
Sea-salt aerosols can absorb solar radiation depending on the wavelength (short wave or long wave) and can cool or heat the atmosphere accordingly (Satheesh and Moorthy, 2005).Sea salt aerosols can have a direct radiative effect in the visible and infrared regions, and, at the same time, can have a role in indirect radiative effect by acting as cloud condensation nuclei (CCN) (Twomey 1977;Twomey et al., 1984;Charlson et al., 1992;Ayers et al., 1997).With the increase in number concentrations of sea-salt aerosols, CCN and cloud droplet number concentration also increase, which leads to an enhancement in the reflectivity of the clouds.Furthermore, for a given amount of moisture, an increase in CCN also reduces the effective radius of droplets and causes an increase in the lifetime of clouds.
The magnitude of the direct forcing of aerosols depends on their scattering and absorbing nature.The cooling and warming effects of aerosols are governed by their optical properties, which depend primarily on the composition of the aerosol (Chylek and Cookley, 1974;Crutzen and Andreae, 1990;Ramanathan et al., 2001;Randles et al., 2004).It has been mentioned that change in wind speed causes a change in sea-salt aerosol concentration in the aerosol system.A change in aerosol composition will change the refractive index of the aerosols and, consequently their radiative forcing.Our incomplete knowledge about spatial and temporal variations in sea-salt aerosol concentration limit our exact estimation of its direct and indirect forcing.Along with the direct and indirect radiative effect and impact of sea salt aerosols; their importance has been also noticed in heterogeneous chemistry (Lewis and Schwartz, 2004;Jaeglé et al., 2011;Spada et al., 2013).
Oceanic aerosols include maritime sources as well as those from continental regions.Transported continental aerosols perturb the composition of the oceanic aerosol system, causing spatial and temporal heterogeneity.Consequently, the radiative impact changes, and these changes have been widely addressed by several investigators (Johansen and Hoffmann, 2003;Bates et al., 2004;Moorthy et al., 2005;Zhu et al., 2007).Most investigators have investigated direct as well as indirect radiative forcing primarily owing to anthropogenic aerosols (e.g., sulfate and black carbon aerosol), while some have focused on the radiative effects of natural aerosols (Charlson et al., 1987;Crutzen and Andreae, 1990;Kaufman et al., 1990;Charlson et al., 1991Charlson et al., , 1992;;Penner et al., 1992;Hegg et al., 1993;Kiehl and Breigleb 1993;Satheesh et al., 1999;Satheesh and Moorthy, 2006).
Studies of the radiative effect from natural aerosols (i.e., sea-salt aerosols) are rather sparse over the Indian domain (Moorthy and Satheesh, 2000;Vinoj and Satheesh 2003;Vinoj and Satheesh 2004;Nair et al., 2008).Satheesh et al. (2002) investigated sea-salt aerosol radiative forcing over the tropical Indian Ocean, reporting that radiative forcing at the surface and top of atmosphere (TOA) were -6.1 W m -2 and -5.8 W m -2 respectively.Vinoj and Satheesh (2003) studied the change in radiation due to the presence of aerosols over the Arabian Sea during the summer monsoon season, and reported a decrease of 21 W m -2 at the surface and an increase in TOA reflected radiation by 18 W m -2 .Vinoj and Satheesh (2004) studied indirect radiative forcing due to sea-salt aerosols and demonstrated that its value was higher than direct radiative forcing (DRF) at TOA.Large uncertainties are associated with sea salt aerosol estimates, and thus their unrealistic representation in climate models can cause significant uncertainties in TOA radiative forcing estimates (Textor et al., 2006;Fan and Toon, 2011;Struthers et al., 2013).
Most prior work was conducted as part of field campaigns from specific locations, and hence were constrained by spatial and temporal coverage.In this work, we study the wind dependence of sea-salt aerosols over a large oceanic region and the variance of related parameters (i.e., wind independent component of aerosol optical depth τ 0 ; and the wind index b) with latitude.When the wind direction is from the continent, an increase in wind speed causes an increase in the concentration of transported continental aerosols along with an increase in sea-salt aerosols due to local production.Hence, it is important to remove the continental influence when we isolate sea-salt aerosol radiative effects.As mentioned earlier, several studies have been carried out on the modulation of sea-salt aerosol generation by the wind where they discuss continental aerosols influence but do not take into account their impact.In this work, we have removed the continental influence using back-trajectory analysis.The use of satellite data has improved the spatial and temporal coverage, using MODIS satellite-derived aerosol optical depth (AOD, τ) with an accuracy of ± 0.03 ± 0.05τ over global oceans (Remer et al., 2005;Yu et al., 2006).
We first establish an empirical relation between τ and seasurface wind speed.Thereafter, we estimate the latitudinal variation of the wind index (b), which relates wind speed dependence (with a unit of s m -1 ) and a wind-independent component of aerosol optical depth (τ 0 ).This variability has been used to generate regional maps of sea-salt aerosol optical depth (τ s ) and its fraction in τ.We use the empiricalcum-model approach for the estimation of DRF of sea-salt aerosols at the short wave and long wave regions.One notable difference of the present work from previous ones is that the estimation of radiative forcing in the long wave region as long wave forcing (LWF) partly counterbalances the effect of short wave forcing (SWF) and cooling at the surface and at TOA.Thus, the present work has several significant differences and advantages.
After establishing the wind-τ speed relationship for the various regions mentioned above, we investigate the modulation in DRF owing to a change in sea-salt aerosol concentration caused by the wind.To isolate DRF due to a change in sea-salt aerosols alone, we need to remove the continental aerosol influence.The hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model run in back trajectory mode for 7 days, has been used for this purpose (Draxler et al., 2003).A hybrid approach connecting the Lagrangian and Eulerian modes has been used for the estimation of simple parcel trajectories, complex dispersion and depositions.A predictor-corrector advection scheme was used for the forward and backward calculation.We used the 7 day back trajectory method for the removal of transported dust and other anthropogenic aerosols, assuming a life time of aerosols as a week.We generated aerosol trajectories for each grid of MODIS data falling in our study region and those data were removed when the corresponding trajectories intersected a significant land mass (continental history).
In rapidly changing meteorological conditions, HYSPLIT may incorporate significant error.Other sources of error may come from the temporal and spatial interpolation scheme used in the model.Removal limits the influence of continental aerosols, which may increase due to the effect of windborne transport and could contribute to columnar optical depth.Thus, over the open ocean, the only aerosols being investigated relative to wind speed are the sea-salt aerosols, and these can be assumed to be the major contributors to AOD that can be modulated with wind speed (Fitzgerald, 1991;Satheesh et al., 2006).
Several studies have reported the wind dependence of τ/mass/number concentration in various forms (i.e., linear, power-law and exponential type) over different parts of the ocean (Moorthy et al., 1997;Moorthy and Satheesh, 2000;Jennings et al., 2003;Smirnov et al., 2003;Satheesh et al., 2006;Mulcahy et al., 2008;Glantz et al., 2009).Studies over the Arabian Sea and Indian Ocean regions have shown an exponential relationship between τ and wind speed of the form, where τ is AOD at wind speed U, τ 0 is the same at U = 0 (i.e., wind independent component of AOD) and 'b' is defined as the index for wind speed dependence (s m -1 ; Moorthy et al., 1997;Moorthy and Satheesh, 2000;Vinoj and Satheesh, 2003;Satheesh et al., 2006).It has been shown that the value of b depends on the wavelength.For example, b = 0.12 for λ = 0.5 µm and b = 0.18 for λ = 1.02 µm (Hoppel et al., 1990;Moorthy et al., 1997).In ideal conditions, the index of wind, b, which indicates the dependence of the generation of sea salt aerosols on wind, should depend only on wind (i.e., the generation of sea-salt aerosols should be same over all latitudes for the same wind condition).But in the practical scenario, this parameter is affected by several meteorological and terrain parameters, such as salinity, viscosity, friction, sea surface temperature and proximity to land.Researchers have reported a linear relationship also between τ and wind speed (Smirnov et al., 2003;Jennings et al., 2003).Mulcahy et al. (2008) showed a power-law relation between these two parameters for the west coast of Ireland.Glantz et al., (2009) have studied the τ and wind speed relationship over the North Pacific Ocean and found a power-law relationship.Since sea-salt aerosol is the only wind-dependent aerosol, its contribution to AOD is estimated by eliminating background aerosol concentration.Following Eq. ( 1), sea salt aerosol optical depth can be derived using the relation, where at U = 0, τ s will be zero and τ will be τ 0 (i.e., optical depth due to background aerosol; Satheesh et al., 2006).
To incorporate the effect of a change of wind on AOD owing to a change in the mass/number concentration of aerosols, we estimated τ at different wind speeds from low to high (0-15 m s -1 ).Following the approach discussed by different investigators (Satheesh and Srinivasan, 2002;Satheesh and Lubin, 2003;Markowicz et al., 2003;Babu et al., 2004;Moorthy et al., 2005;Satheesh et al., 2006;Nair et al., 2008), we have also used the empirical-cum-model approach for the estimation of the optical properties of aerosols and DRF.
The τ values estimated from the wind-τ equations are incorporated in a Mie scattering model, Optical Properties of Aerosols and Clouds, OPAC, (Hess et al., 1998) by changing the sea-salt aerosols concentrations in various pre-defined aerosol models.The purpose of such an exercise is to estimate the optical properties of an aerosol system in which sea-salt aerosol is the only component whose concentration changes with wind speed.South of the equator, we adopted clean marine aerosol models proposed in the OPAC model, while to the north of the equator different aerosol models were used in which the relative abundance of various species at no wind condition are based on previous observations over those regions (Satheesh et al., 1999).Over the southern oceanic regions (i.e., SO1-3), we selected the Maritime Clean model, which is formulated specially for unperturbed remote oceans with no soot and where sea-salt aerosol concentration depended on wind speed.The fourth region (SO4) maritime tropical model was used.
The simulated results from OPAC were used as input parameters to a radiative transfer model, SBDART (Santa Barbara DISORT Atmospheric Radiative Transfer Model; Ricchiazzi et al., 1998) in conjunction with observed τ, for the calculation of DRF at TOA, the surface and in the atmosphere.Fluxes at TOA and the surface were estimated for the calculation of radiative forcing and forcing was estimated using (Satheesh 2002;Nair et al., 2008) where F No Aerosol is flux at TOA/surface assuming zero aerosol concentration and F Aerosol is flux at TOA/surface with aerosol.The difference between TOA and surface forcing is the atmospheric forcing.

RESULTS AND DISCUSSIONS
Wind Dependence of AOD Fig. 1 shows the scatter plot of daily MODIS τ as a function of surface wind speed over the six study regions starting from the Southern Ocean (40°S-30°S, 60°E-70°E) to the Arabian Sea (10°N-20°N, 60°E-70°E).For SO2-4 and the AS2, τ shows a clear exponential growth with an increase in wind, though over regions SO1 and AS1 this relationship is not as prominent.Still, an exponential relationship between τ and wind speed appears constant.This result is in agreement with earlier studies performed over the Indian sea and the Arabian Sea region (Moorhty et al., 1997;Vinoj andSatheesh, 2003, 2004;Satheesh et al., 2006).The correlation coefficients shown between τ and wind speed were significant at a 95% significance level for all the regions.
From these exponential fits, we estimated the values of τ 0 and b for each region, which were subsequently used to derive the value of τ at different wind speeds (0-15 m s -1 ) for the respective regions (Fig. 2(a)).Here, the contribution of wind-generated sea-salt aerosols was not prominent in an increase in τ.But when we eliminated τ 0 and estimated the contribution of τ s in τ with the help of Eq. ( 2), then its contribution was seen to be significant ~50%.The increase in sea salt aerosol concentration discussed here is only that fraction of sea salt aerosols which increased due to the increase in wind and not the pre-existing sea salt aerosol concentration as background aerosols.Also, the presence of a large amounts of phytoplankton over SO is known and may be a source of oceanic sulfate.The percentage contribution of τ s in τ, increased gradually with increase in wind as we moved towards north (Fig. 2(b)).Contribution of τ s in τ showed a variation from 40% (SO1) to 80% (AS2) at high wind speed (~15 m s -1 ) conditions (Fig. 2(b)).

Latitude Variation of τ 0 and Wind Index b
The relationship between τ and wind speed indicated an increase in τ 0 as moves to the north.Therefore, to estimate the exact pattern of increase in τ 0 with latitudes, we studied the latitudinal variation of τ 0 and wind index b for all the regions (Figs. 3 and 4).This analysis was also important because variations in these parameters were essential in order to derive the spatial heterogeneity in sea-salt aerosols.The latitudinal variation of τ 0 revealed an exponential increase as moving the north, which can be justified by the fact that, continental aerosol concentration increases due to the proximity to land.The latitudinal variation of τ 0 has been parameterized as, where Λ indicates latitude.
Near the southern latitudes in the study region, the value of τ 0 was observed to be 0.09 while over the northern AS it was observed to be 0.33.This latitudinal variation is consistent with various island and ship-borne investigations (Vinoj and Satheesh, 2004;Satheesh et al., 2006;Vinoj et al., 2007).The present result was more robust as compared to previous ship-based and island measurements, which had limited spatial and temporal coverage as compared to satellite data.This relationship was subsequently utilized in estimating the background AOD (i.e., the wind independent component of τ).
In conjunction with τ 0 variation, we also studied the latitudinal variation of b.Other studies which incorporate cruise, island measurements and satellite data analysis have shown different results regarding the latitudinal variation of the wind index.Vinoj and Satheesh (2004) show an exponential decrease in the wind index as moving the north, though this result was based only on a few point measurements.Satheesh et al (2006), with the help of satellite data analysis, observed no significant variation in the wind index as they moved from the south to the north.In our study, we observed a gradual linear increase in b as we move from the southern ocean towards the north, which can be expressed as, B = 0.0015 × Λ + 0.096 (5) Out of the several factors discussed earlier in section 2, the one which causes this change is not clear.Thus, the exact reason for the increase in the wind index as we move towards the north is still an open question.

Regional Distribution of Sea-Salt Aerosols
In Fig.
(2), we observed that among all the six regions,  4) and ( 5)) in Eq. ( 2) with NCEP surface wind speed.The monthly mean τ 0 and b values were calculated for each latitude starting from the equator to 20°N to preserve their latitudinal variations.These monthly latitudinal variations of τ 0 and b were used in conjunction with monthly NCEP wind speed to generate monthly regional maps of τ s at 1° × 1° resolution.With the help of these calculations, we estimated the seasonal variations in sea salt aerosol distribution.
Fig. 5 shows the seasonal maps of sea-salt aerosols for summer (March, April and May) and winter (January, February, November and December).The first panel in Fig. 5 (top to bottom) shows MODIS-derived composite τ, estimated τ s and the fraction of sea salt aerosol (caused only by an increase in sea surface wind) in the total aerosol load for the summer season, and panel 2 is the same for the winter season.In the summer, high τ values were observed near the Indian west coast and the northern part of AS, while high τ s values were found over the Somalia region and the highest (~0.2) at the northern edge of the AS.A high concentration of sea salt aerosol was observed over these high wind regions as during the summer season, when a south westerly wind prevails (Gadgil, 2003;Satheesh et al., 2006).During summer, sea salt aerosol fractions in the total aerosol load were the highest over the Somalia current region and the south east corner of the Arabian Sea.In winter, when wind directions are from the north east, a limb of high τ was observed from the west coast of India with the highest values near the coast (Satheesh et al., 2006).A substantial decrease was noticed moving away from the coast to the far ocean since the continental influence decreases.As compared to the summer season, in winter, aerosol optical thickness was less, and similarly low τ s values were observed with its peak value (~0.1) over the Somalia region (Fig. 5(b) for winter).In the winter season the maximum contribution of sea-salt AOD in the total τ was about 50% while in summer, this contribution was more than 70% at some locations.Satheesh et al. (2006) have also studied the spatial variation of sea-salt aerosols but on a monthly time scale.

Modeled Aerosol Optical Properties
The prime objective of this work is to estimate the change in atmospheric DRF in short wave as well as in long wave regions due to the modulation in sea-salt aerosol concentration by surface wind speed.To achieve this, we used the empirical-cum-model approach, as explained earlier in Section 2. The calculation of radiative forcing from the SBDART model needed the optical properties of the aerosol system, which have been set based on the OPAC model.Depending on the geographical location of the study regions, we selected aerosol models from OPAC.These models provide the concentrations of the various components of the aerosol system, which were used to derive the optical properties of that specific aerosol model.
Regions SO1-3 are situated far to the ocean (i.e., they are the least anthropogenic affected).Hence, we selected a maritime clean aerosol model for these regions.For SO4, we chose the maritime tropical aerosol type.Water-soluble and sea-salt aerosol (accumulation and coarse modes) are the components of the aerosol model for the southern ocean region, but the concentration of water soluble aerosols differs in SO1-3 and SO4.AS1-2 are in Indian Ocean/Arabian sea region and observational information regarding aerosol concentrations available in the individual aerosol components prevailing over these regions.Therefore, we used aerosol models based on earlier observations (Satheesh et al., 1999).Aerosol components for these models are soot, water soluble, dust and sea-salt aerosol (in accumulation and coarse modes).
Using these aerosol types for each of the respective regions, we reproduced τ, derived from the wind-τ relation at different wind speeds, and, simultaneously, other optical properties.Concentrations of the aerosol components, other than sea-salt aerosol, were kept constant with wind speed according to the model.We increased the concentration of sea-salt aerosols (accumulation and coarse modes) alone to generate an estimated τ from the empirical relation.We also modeled τ at 11 µm at different wind speeds over the respective regions (Fig. 6(a)).Over the SO (SO1-4), we observed low values of τ at 11 µm, but at the same time for AS1-2 relatively high values were observed.An increase in the number concentration of sea-salt aerosol in the coarse mode showed the same variation as τ at 11 µm.At high wind conditions, τ at 11 µm is enhanced about 2-20 fold from its low wind values over different study regions, whereas τ at 0.55 µm values showed a maximum 6 fold increase in its values from low wind conditions to high wind conditions.Fig. 6(b) shows the contribution of τ s at 11 µm to the total τ at various wind speeds, estimated using Eq. ( 2).We noticed that the percentage contribution of sea-salt aerosol optical depth to the total τ at 11 µm reached about 100% at moderate wind speed (~5 m s -1 ) while it was only ~80% in the total τ at 0.55 µm contribution at 15 m s -1 over AS (AS1-2).In addition to τ, we studied the variation of single scattering albedo (SSA) with an increase in wind speed for all regions at 0.55 µm and 11 µm (Fig. 7).Extinction represents the composite effect of scattering and absorption, and SSA represents the fraction of scattered light in the total extinction.The SSA of an aerosol depends on its chemical composition and varies from 0 (i.e., completely absorbing) to 1 (i.e., completely scattering).A change in wind will change sea-salt aerosol concentration and hence the composition of the total aerosol system.Thus, a change in the contribution of sea-salt aerosol to a particular aerosol model will influence SSA values.
The SSA of an aerosol system is the weighted average of the individual components of aerosol species.Fig. 7(a) shows the variations of SSA at 0.55 µm and Fig. 7(b) shows the same at 11 µm.Over the SO (SO1-4), SSA was close to 1 for all wind speeds as the sea-salt aerosol concentration did not change drastically.Over the AS (AS1-2), it showed a gradual increase from 0.85 to 0.97 and from 0.75 to 0.96 respectively.When the number concentration of the seasalt aerosol showed an exponential growth (for AS1-2), then the concentration of the other components of the system (i.e., water soluble, soot etc.) was constant.As seasalt aerosols are of scattering type, the SSA of the composite system approached 1 at 0.55 µm.Over SO (SO1-4), the SSA at 11 µm was constant (~0.23) for all wind speeds, while over the AS (AS1-2) the SSA showed a moderate increase with wind speed where it increased from 0.27 to 0.5.In the infrared region, sea-salt aerosols act as partial scatterers and other components like soot as perfect absorbers.Thus, a moderate increase in SSA at 11 µm was observed with wind speed (AS1-2).

Direct Radiative Forcing in Short Wave and Long Wave Region
In this section, we discuss the results for DRF in shortwave and long wave regions.We estimated LWF and SWF caused due to a rise in sea-salt aerosol concentration owing to increased wind.After estimating aerosol optical properties, we estimated the modulation in DRF at the surface, TOA and in the atmosphere (Figs. 8 and 9).The modeled aerosol properties were used as input in the SBDART model for the estimation of DRF.Fig. 8 represent DRF at the surface, TOA, and atmospheric forcing, respectively, in the short wave region.
As the wind speed increased from low to high (~15 m s -1 ), a change in short wave forcing at the surface was observed about (3-10 W m -2 ) over regions SO1-4 (Fig. 8 (a)).The highest change in surface radiative forcing from low to high wind conditions was observed over the Arabian Sea (AS1-2) near ~40 W m -2 .Over the AS (AS1-2), SWF at the surface was several-fold higher than in the other regions.Similar to surface SWF, TOA SWF also showed the highest increase over the AS (~32 W m -2 ) as wind speed changed from low to high wind conditions.Over the other regions this increase was from (~2 W m -2 to 9 W m -2 ).These results are in agreement with other studies where radiative forcing in the shortwave range has been estimated in the same range (Vinoj and Satheesh 2003;Vinoj et al., 2004;Satheesh and Moorthy, 2005).Atmospheric forcing modulation owing to an increase in sea-salt aerosol is shown in Fig. 8(c).A change in the atmospheric heating over the SO (SO1-4) with a change in wind speed was about ~1 W m -2 , as it varied from 0.4 W m -2 to 1.4 W m -2 with wind speed.Over AS1 and 2, atmospheric forcing changed to about 6 W m -2 and 8 W m -2 , respectively from low to high wind conditions.
LWF at the surface varied between 0.2-0.7 W m -2 over the SO region (SO1-4), but over the AS it was again several time higher than in other regions.At high wind conditions (~15 m s -1 ), it was observed to be as high as ~9 W m -2 .At TOA also, comparatively little heating due to LWF was observed for the SO region, which varied from 0.1-0.4W m -2 at 15 m s -1 .Over the Arabian Sea it was ~5 W m -2 at 15 m s -1 .Atmospheric forcing in the long wave region was estimated as 0.3-1 W m -2 over SO.Over AS, it observed several-fold higher than in other regions, and increased approximately 15 times from calm wind conditions to high wind conditions.The highest modulation of radiative forcing was observed over the AS (AS1-2), as the highest rate of generation of sea-salt aerosols (in both modes) was observed over this region.High values of LWF observed over the AS indicate the dominance of sea-salt aerosols in coarse mode over this region.
SWF and LWF have a counter effect at the surface and TOA, as SWF produces a cooling effect while LWF causes a heating effect at the surface and TOA, though the magnitude of LWF is usually much smaller than SWF.It is important to notice that the rate of increase in LWF is more than the corresponding increase in SWF over AS (AS1-2), which indicates the dominance of the coarse mode at higher wind speeds.LWF counter balanced SWF at the surface and TOA by about 7% and 5%, respectively, in SO (SO1-4) from calm wind conditions to high wind conditions.The maximum offset of SWF at the surface was observed over AS (AS1-2) where it was about 23%.At TOA, this offset ranged from 4-5% over SO (SO1-4) and over AS (AS1-2), this offset was about 19%.

CONCLUSIONS
The generation of sea-salt aerosols by the action of the wind is a well-established fact.But the dependency of the mass/number concentration or aerosol optical depth (AOD, τ) on wind speed does not follow a universal pattern over all  ocean, and remain a topic to be explored.Sea-salt aerosols are the prime components of the aerosol system over oceanic regions and significantly affect the radiation budget.Any change in sea-salt aerosol concentration due to wind will produce a considerable change in radiative forcing, and hence, it is essential to study the wind modulation of radiative forcing.
Here, region of 10° × 60°, starting from the mid-southern ocean to the Arabian Sea, was selected to derive a wind-τ relationship and, subsequently, to study modulation in direct radiative forcing (DRF) in short wave and long wave regions.The study region was sub-divided into six regions, and an exponential relationship between τ and wind speed was observed.An actual increase in sea salt aerosol optical depth (τ s ) caused due to wind speed was estimated by eliminating the background aerosol optical depth from the total τ.The maximum increase in sea-salt aerosol concentration and its contribution to the total τ was observed over the Arabian Sea (AS1-2) where it was nearly 80% at 15 m s -1 .In conjunction with τ and τ s variations, we also observed a latitudinal variation of the wind-independent component of AOD (i.e., τ 0 ) and wind index b.These were essential inputs to derive the spatial variation of sea salt aerosol and were used over the Arabian Sea.The latitudinal variation of τ 0 showed an exponential increase as we moved from the southern to the northern latitude while there was a linear increase for wind index b.The precise reason for the increase in b is not clear, and needs to be explored.But, several factors such as temperature, salinity and proximity to land play a role in the modification of b.
We have then studied DRF over the study region and to model DRF, the optical properties of aerosols were generated with the help of Optical Properties of Aerosols and Clouds, i.e., OPAC model simulations and the wind-τ relationship.Along with other optical properties, τ at 11 µm was also modeled and a several-fold increase was noticed in its value from low to high wind conditions as compared to that of 0.55 µm.The contribution of τ s at 0.55 µm in the total τ was ~80% over the Arabian sea and the contribution of τ s at 11 µm was nearly 100% over the same region.In the long-wave region, sea-salt aerosols in coarse mode primarily contributes to optical depth.These modeled optical properties from OPAC were used in the SBDART ((Santa Barbara DISORT Atmospheric Radiative Transfer) model for radiative forcing calculations.DRF was calculated in shortwave and long-wave regions.SBDART simulations show, at high wind conditions, that the sea-salt aerosolinduced short wave cooling at the surface, top of the atmosphere (TOA) and heating in the atmosphere were 40 W m -2 , 32 W m -2 and 8 W m -2 respectively.Long wave heating due to sea salt aerosol observed at the surface, TOA and in the atmosphere were about 9 W m -2 , 6 W m -2 and 15 W m -2 , respectively.The highest value of this offset at the surface was observed over the Arabian Sea; i.e., ~23% and at TOA this offset was observed to be about 19%.This implies that over the Arabian Sea sea-salt aerosols in coarse mode contribute significantly to long wave radiative forcing compared to other regions.

Fig. 2 .
Fig. 2. (a) Variation of total AOD at 0.55 µm as a function of wind speed (b) percentage contribution of sea salt AOD in total AOD.

Fig. 6 .
Fig. 6.(a) Variation of total AOD at 11 µm as a function of wind speed (b) percentage contribution of sea salt AOD in total AOD.

Fig. 8 .
Fig. 8. Variation of radiative forcing in short wave range (0.2-4 µm) a function of wind speed (a) at surface (b) at TOA (c) in atmosphere.

Fig. 9 .
Fig. 9. Variation of radiative forcing in long-wave range (8-14 µm) a function of wind speed (a) at surface (b) at TOA (c) in atmosphere.