Variability of Arabian Sea surface circulation and chlorophyll distribution: a remote sensing estimation

High resolution mean surface velocity field of the Arabian Sea was derived by combining sea level anomalies based on satellite altimetry data with the surface drifter data. The mean current field exhibits the western boundary Somali Current, weak westward North Equatorial Current, weak West Indian Coastal Current. Besides strong currents, the Arabian Sea exhibits strong mesoscale eddy activity in the western side. The seasonal surface circulation changes of the Arabian Sea and variability associated with the Indian Ocean Dipole (IOD) events were determined. The Somali Current is northward during summer, with a speed above 1.8 m/s in July and it reverses towards south in winter. Significant changes were found in the flow pattern during IOD events. An anticyclonic gyre near the equator was a conspicuous feature during the positive IOD. Further, the Great Whirl shows an earlier formation and dissipation during the positive IOD. Notably, a delay of southward reversal of the Somali Current is found during the positive IOD. The southward flow of West Indian Coastal Current during summer monsoon is much stronger during the positive IOD and the northward flow in fall is prominent during negative IOD. The eastward Equatorial Jet is absent in fall and instead an opposite (westward) flow prevails during the positive IOD. Enhanced Chl-a concentration occurs during the negative IOD, compared to the positive IOD, especially in the western Arabian Sea. The wind-driven upwelling and mesoscale eddies are significantly modifying the Chl-a distribution along the Somali Coast. A high resolution mean surface velocity field is developed. Contrasting circulation changes are found during positive and negative IOD events. Mesoscale eddy activity is considerably influencing the Chlorophyll distribution. A high resolution mean surface velocity field is developed. Contrasting circulation changes are found during positive and negative IOD events. Mesoscale eddy activity is considerably influencing the Chlorophyll distribution.


Introduction
The Arabian Sea (AS) is a unique region of the world ocean, where the seasonally reversing monsoon winds drive one of the most energetic surface current systems in the world oceans (Fig. 1). It is the only ocean basin that fully reverses its circulation on a semi-annual basis (Schott 1983;Schott and McCreary 2001;Schott et al. 2009). The strong seasonal surface forcing cycle produces alternate converging and diverging flow fields in

Open Access
Terrestrial, Atmospheric and Oceanic Sciences *Correspondence: bennynpeter@gmail.com the coastal regions, and significantly controls the upper ocean characteristics of the AS. The AS surface circulation comprises of the zonal North Equatorial Current (NEC; also called Winter Monsoon Current), Southwest Monsoon Current (Summer Monsoon Current), coastal currents, gyres, eddies and is also influenced by low-frequency wave activity. Among them, the monsoon currents are the most prominent large scale currents. In summer, the Southwest Monsoon Current flows eastward as a broad continuous band from the western AS to the Bay of Bengal. In winter, the NEC flows westward between 8° N and the equator (Pickard and Emery 1982) from the eastern boundary of the Bay of Bengal to the western AS, transporting low salinity water from the Bay of Bengal to the AS. The NEC is comparatively weak in the annual mean circulation (Benny and Mizuno 2000). Several studies (Bruce et al. 1994(Bruce et al. , 1998Donguy and Meyers 1995;Shenoi et al. 1999;Murty et al. 2000) were carried out to understand the monsoon currents and their transports. Compiling surface buoy trajectories, Molinari et al. (1990) generated a monthly climatology of the tropical Indian Ocean.
In the AS, the Somali Basin had been well studied using both observations (Schott 1983) and models (Luther and O'Brien 1985). The seasonal reversal of wind system in the AS drives significant seasonal variations in the ocean circulation pattern, involving the total reversal in the Somali Current (SC). The SC flows northward during summer monsoon and southward during winter monsoon and is notable for its high speeds up to 2 m/sec and has a transport of about 65 Sv, mostly in the upper 200 m (Pickard and Emery, 1982). However, the current speeds in the western tropical Indian Ocean are of the order of 0.75 to 1 m/s (Bahulayan et al. 1997).
The East Arabian Current was observed to be very strong during summer monsoon (McCreary et al. 1993;Schott and McCreary 2001) and appears fully established by mid-May. Many studies (Shetye et al. 1990(Shetye et al. , 1991McCreary et al. 1993;Shankar and Shetye 1997;Amol et al. 2014) were made on the West Indian Coastal Current (WICC) in the AS, which also changes its direction seasonally; southward during summer monsoon and northward in winter monsoon. Wyrtki (1973) first identified the existence of Equatorial Jet (Wyrtki Jet) in the Indian Ocean and it was also simulated (O'Brien and Hurlburt 1974), using a two layer numerical model. The tropical Indian Ocean can support fast travelling planetary waves (Benny and Mizuno 2000) and the dynamics of the AS is significantly influenced by seasonal mesoscale eddies and planetary waves. Recently, Centurioni et al. (2017) carried out a comprehensive observational effort to understand the upper ocean processes and atmospheric feedbacks of the AS.
With the onset of the southwest monsoon, the currents in the AS evolve into a complex pattern of mesoscale eddies and gyres with associated horizontal temperature gradients and current shears (Bruce, 1983;Luther and O'brien 1985). In the south-western AS, the Southern Gyre and the Great Whirl (GW) develop as large anticyclonic eddies every year in response to the arrival of the Southwest Monsoon Current (Swallow and Fieux 1982). They are a part of the SC system, one of the most energetic regions of the world oceans, where surface currents of more than 3 m/s have been observed (Schott 1983; Page 3 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) Fischer et al. 1996). Most of the details concerning the generation, persistence, and decay or collapse of this eddy are still poorly understood (Schott and McCreary, 2001). Beal and Donohue (2013) studied the occurrence of the GW and its dynamics combining remote sensing observations. A large anticyclonic eddy had been recognized by Bruce et al. (1994) from in-situ observations, satellite altimetry and numerical experiment and named this as Lakshadweep High, as it forms near Lakshadweep Islands. The Indian Ocean Dipole (IOD) phenomenon has been recognised Webster et al. 1999) as the largest and unique interannual variability in the Indian Ocean. Besides SST anomaly variation, IOD is associated with other ocean-atmosphere interaction processes and temperature anomalies are also found in subsurface levels of the ocean (Vinayachandran et al. 2002;Horii et al. 2008). Moreover, the zonal wind anomalies during IOD events weaken or reverse the Equatorial Jet (Vinaychandran et al. 1999). Many studies Annamailai et al. 2003;Halkides et al. 2006;Meyers et al. 2007;Vinayachandran et al. 2007) had described the characteristics of different IOD events. The ocean-atmosphere coupling during IOD events involves atmospheric convection, winds, SST and upper ocean dynamics (Vinayachandran et al. 2009;Aparna et al. 2012). Lagerlof et al. (1999) introduced a new technique to combine the satellite altimetry and drifter observations to study the near-surface currents of the tropical Pacific Ocean. Further, global studies (Niiler 2001;Sudre and Morrow 2008;Maximenko et al. 2009;and Rio et al. 2011) have developed the mean dynamic topography and currents using satellite altimetry and drifter observations. Rio et al. (2014) and Laurindo et al. (2017) have updated the estimation technique of surface currents combining remote sensing observations.
As the AS lacks an organised observational network, much of the information is based on sparse historical data collected in different periods. Moreover, studies were conducted on regional wises and comprehensive information on its scales of motion and variability are not yet clear. Hence, the present study focused on achieving three important objectives, merging different remote sensing observations. First, to derive a high-resolution (1/3 ° latitude × 1/3 ° longitude) mean surface velocity field. Next, it aims to determine the effect of IOD on surface currents and mesoscale activity. Finally, to examine the influence of surface circulation features on Chlorophyll-a (Chl-a) distribution and the changes during IOD events. The rest of the paper is prepared as follows. Section 2 details the data used and the methodology followed. Section 3 presents the results from the analyses and validation of estimated fields in the present method with current meter observations. The influence of IOD on surface circulation and Chl-a was also shown in Sect. 3. The obtained results were discussed in Sect. 4. Lastly, the conclusions are summed up in Sect. 5.

Data and analysis
The present study employed remote sensing observations from different platforms to derive the mean velocity field. The surface drifter data used in this study comes from the Global Drifter Program (Surface Velocity Program, http:// www. aoml. noaa. gov/). The data used were qualitycontrolled and optimally-interpolated to uniform sixhour interval trajectories using the drifter's position data. The observational density of drifter data available in the period of 1993-2012 is given in Fig. 2.
The satellite altimetry data used in this study were Maps of Sea Level Anomaly produced by the Collect Localization Satellites, France for the period of 1993-2012. AVISO prepared this dataset by merging data from five satellites, i.e., Topex/Poseidon (October 1992 to July 2002), ERS-1 (January 1994 to March 1995), ERS-2 (June 1996 to June 2003), Jason-1 (December 2001 to July 2013), Jason-2 (June 2008 to October 2019) and Envisat (March 2002 to April 2012) (AVISO1997, http:// www. aviso. Ocean obs. com). Maps had been produced with a resolution of 1/3 ° latitude and 1/3 ° longitude in space and week in time. The weekly mean ocean surface wind fields derived from the scatterometers on board ERS-1/2, QuikSCAT and ASCAT, generated by "Centre ERS d' Archivage et de Traitement" (CERSAT), France (http:// cersat. ifrem er. fr/) for the period of 1993-2012 were used to estimate the wind-driven Ekman velocity.
Further, the monthly averages of Chl-a observations with 9 km resolution from SeaWiFS mission during 1997-2007 were used. The dataset was obtained from NASA's Ocean Color Web, supported by the Ocean Biology Processing Group (OBPG) at NASA's Goddard Space Flight Centre (https:// ocean color. gsfc. nasa. gov/).
The surface drifter and altimetry observations were combined to estimate the mean surface velocity following the method of Uchida and Imawaki (2003). At a location (x, y) and time t, the instantaneous geostrophic velocity "Vg (x, y, t)" can be written as, where, Vmg (x, y) is the mean geostrophic velocity over some observational period, Vg'(x, y, t) is the geostrophic velocity anomaly from the temporal mean, x is the distance in zonal direction and y is the distance in meridional direction. Therefore, Vg x, y, t = V mg x, y + Vg ′ (x, y, t), Page 4 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) 33:22 The instantaneous geostrophic velocity was estimated from the drifter observations, and the geostrophic velocity anomaly was computed from the altimetry sea level anomalies. Hence, the mean velocity was calculated by subtracting the altimeter-derived geostrophic velocity anomaly from the drifter-derived instantaneous geostrophic velocity measured at the same time and location. This method estimates almost unbiased Eulerian mean velocities which were free from the sampling tendency of the drifters. Thus, the unknown mean velocity can be estimated for the grid box where a drifter passed. The average of the calculated mean velocities in each grid gives a more accurate estimate by reducing the estimation error. Previous studies employing this method successfully derived the mean velocity field of the North Pacific Ocean (Uchida and Imawaki 2003), South Indian Ocean (Benny et al. 2014), South China Sea (Benny et al. 2015) and Bay of Bengal (Mridula et al. 2016).
The drifter trajectories are low-pass filtered by a 60 h running mean to remove the high frequency fluctuations before the calculation. Then the drifter data had been gridded into 1/3 ° latitude 1/3 ° longitude boxes, which are same as those of the Maps of Sea Level Anomaly. The surface velocity has been estimated in each grid from the drifter position data as shown in Fig. 3.
The wind slip correction was made using the wind data interpolated in drifter trajectories, employing the relation given by Niller and Paduan (1995). An additional correction was made for drifters, which had lost its drogues during its traverse, using the empirical relation given by  Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) Pazan and Niller (2001). The Ekman Current was estimated using the Ekman model by Ralph and Niller (1999) where, V E = (U E + i.V E ) is the Ekman velocity at 15 m depth, A = 7 × 10 -7 s −1/2 , f is the Coriolis parameter, and W is the wind speed. The term exp (i.θ) (θ = ± 54) represents the rotation of the Ekman velocity to the righthand-side of the wind vector in the northern hemisphere. The geostrophic velocity component was computed from the drifter velocity by subtracting the Ekman component. The zonal and meridional components (u g ' , v g ') of geostrophic velocity anomaly Vg' (x, t) are computed from altimetry sea level anomaly data using the conventional geostrophic relation.
where, g is the acceleration due to gravity and h is the sea surface height anomaly. Since geostrophic approximation is not valid near the equator, the estimation of velocity anomaly was limited down to 2° N. In this study, mean velocities were calculated only at locations where eight or more drifter observations were available. The time series of geostrophic velocity was obtained by adding the mean geostrophic velocity to the geostrophic velocity anomaly derived from time-series satellite altimetry data.

Results
The Eulerian mean velocity field obtained by combining the satellite altimetry and drifter data reveals the details of surface flow pattern in the AS (Fig. 4a). The average of altimeter-derived velocity anomalies at the time of drifter measurements ( Fig. 4b) serves as a correction to the simple average of the drifter-derived velocities (Fig. 4c).

Mean velocity field
The mean surface geostrophic velocity field (Fig. 4a) illustrates a weak basin wide anticyclonic gyre in the AS. The flow is quite weak everywhere except at the western boundary where northward-flowing SC is present. The SC is intense between 2° N and 12° N, where the speed reaches 1 m/s. A part of this current forms a strong anticyclonic eddy at 7.6° N, 52.6° E and the current further weakens as it moves northward.
In the southern AS, the weak broad westward flow of NEC is visible between 53° E and 70° E and between 2° N and 5° N with a speed between 0.25 and 0.5 m/s. The westward-flowing NEC turns northward at around 55° E, merges with the anticyclonic flow and joins the northward-flowing SC. Traces of the eastward flow is present in the south-western part between 2° N and 3° N up to 52° E and no regular current pattern is observed in the south-eastern region.
In the eastern AS (West coast of India) the southward flow is observed in the southern part with a maximum speed of 0.6 m/s. When it reaches the southern tip of the Indian peninsula, the current joins with the westward flow of NEC to complete the anticyclonic circulation. In the northern part of the AS, the flow is quite weak, but the northward flow prevails along the western side up to 19° N with an average speed of 0.4 m/s. The central AS exhibits dull circulation with velocity less than 0.2 m/s. The simple averages of the drifter-derived velocities ( Fig. 4b) are corrected by the altimeter-derived velocity anomaly (Fig. 4c). Although the large-scale flow pattern is not much different from the simple average of drifterderived velocities, the above-mentioned correction by the altimeter-derived velocity anomaly shows significant values along the peripherals, especially in the western part and near the equator. Maximum anomaly of about 0.7 m/s is found along the western boundary, whereas the eastern and northern AS show comparatively small (< 0.3 m/s) anomaly. Anomaly field is also prominent in the zonal eastward flow near the equator, where the variations up to 1 m/s are found, especially in the southwestern part.
The mean velocity field is more aligned to the flow pattern during southwest monsoon. This is because the circulation in the AS is strongest during southwest monsoon and hence the mean velocity field holds a fair similarity to the pattern during that monsoon.

Validation of estimated velocity field
The velocity field obtained by combining satellite altimetry and surface drifter observations was compared with in-situ observational data: shipboard acoustic Doppler current profiler (ADCP) data and moored recording current meter (RCM) observations in the western AS, which cover the annual cycle from April 1995 to October 1996. Thus, the comparison was carried out on spatial and temporal scales to validate the estimated field.

Comparison with ADCP observations
The ADCP data collected during the World Ocean Circulation Experiment (WOCE), September to October 1994 (cruise number 237) in the AS was used for qualitative validation of flow pattern. The estimated absolute geostrophic velocity field in September 1994 and the ADCP velocity fields at depth 50 m were presented in Fig. 5a.
The estimated instantaneous velocity field and the in-situ field data from ADCP observations show consistent flow patterns. In both estimated and observed Page 6 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022)  fields, the maximum speed (1 m/s) was found in the Bay of Bengal region at approximately 5° N, 82° E and the flow was westward. Velocity close to 0.75 m/s can be seen at 6° N, 79° E and the flow was directed towards the Sri Lankan Coast.
In the central AS, the velocity field attains a maximum of only 0.5 m/s. At around 14° N and 8° N along 65° E, the speed increased to 0.5 m/s. Hence, the overall current pattern (the magnitude and direction) from ADCP observations were in good agreement with the estimated velocity field over the expanse of the AS.

Comparison with moored current meter observations
Moored recording current meter observations in the AS acquired as a part of WOCE (Schott and Fischer 2000) during April 1995-October 1996 was also used for validation of the estimated velocity field. Data from five current meters, namely; IK-19, IK-18, IK-17, IK-14 and IK-12, located between 6° N and 12° N ( Fig. 5b) were employed for comparison. The estimated meridional and zonal velocity components compared separately with the observed components at 50 m depth (Fig. 5c). Good correlation was obtained for all locations without any regional biasing. Highest correlation was with IK-19 located around 7ºN, where the correlation coefficient is 0.93 and 0.88 for zonal and meridional components respectively. The lowest correlation of 0.73 is found for the meridional component of IK-18. The IK-14 and IK-12 are located around the GW region, where the current speeds are higher and show a better correlation of above 0.8. The high speed zonal component velocity is underestimated at IK-12 and IK-14, is may be due to the influence of the GW (the location is closer to the GW region). Thus, the present estimation is reliable for both weak and strong current regions and seasonal changes. Also, it is obvious that the estimation is consistent with the observed annual cycle (shown for IK-19, Fig. 5d). Hence, consistency of the present estimation is validated against in-situ observations in spatial and temporal scales.
In January, active circulation prevails in the southern and western part of the AS. The westward flow of the NEC is the most prominent current with speed up Fig. 6 Maps of mean surface geostrophic velocity fields (m/s) in January, April, July and October, averaged over 20 years (1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012) Page 10 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) 33:22 to 1 m/s. The NEC enters from the Bay of Bengal as a mighty flow and flows between 2° N and 7.5° N up to 60° E, where it takes an anticyclonic turn and forms a weak gyre between 60° E and 75° E. At the western boundary, the current is weak and not continuous. A southward flow starting at around 5° N and 49° E turns eastward at around 2° N with speeds up to 1 m/s and extends up to 55° E, where it merges with the westward flow. A weak northward flow along the western boundary is visible from 7.5° N with speeds up to 0.6 m/s and extends up to 16° N, where it takes a cyclonic circulation in front of the Gulf of Aden, north of Socotra. Along the west coast of India, the weak northward flow of WICC with a speed less than 0.3 m/s is visible. The central and eastern AS displays sluggish circulation with current speeds below 0.2 m/s. Significant changes have occurred in April. The crossbasin westward NEC is weakened and confined south of 3.5° N. The flow is irregular but reaches up to the western end. Northward-flowing SC is the dominant one in April with a maximum speed of 1 m/s. When it reaches around 10° N, the major part turns eastward and forms an anticyclonic eddy centred at 7.6° N, 52.6° E. Another part exits towards the Gulf of Aden and a minor part of the current continues to flow northward. A weak cyclonic eddy is observed at 15° N, 53° E. The WICC is southward and is visible only from 15° N with a speed less than 0.4 m/s and the major part of the WICC joins with a westward flow at the southern end.
In July, a narrow westward flow is visible near the equator with a speed of 1 m/s. The SC is the prominent flow with a maximum speed of 2 m/s during this month. A strong anticyclonic eddy (GW) is formed on its right (7.6° N, 52.6° E) with a speed of 1-2 m/s and the westward flow near the equator is partially merging with this eddy. To the left of this eddy, a weak dual core anticyclonic eddy is observed with a speed of 0.2-0.5 m/s. A minor part of the SC continues northward up to 22° N. On the eastern side, the southward-flowing WICC is visible from 15° N with a speed of 0.3-0.6 m/s. As it reaches the southern tip of India, with the contributions from the southeastern AS, the flow becomes wider and turns towards the Bay of Bengal. Overall, the circulation is active in the western AS with the strong anticyclonic eddy.
Conspicuous changes are followed in October from July. The westward flow near the equator is replaced by an eastward flow prominent from 60° E. The eastward flow near the equator with a speed of 0.3-0.6 m/s is a part of the Equatorial Jet and broadened at the eastern end. The northward-flowing SC is weak compared to July, but present with a speed of 1.2 m/s. The anticyclonic eddy centred at 7.6° N, 52.6° E still persists. Another anticyclonic eddy (Socotra Eddy) is observed northeast at 12° N, 57° E with speed of 0.3-0.6 m/s. A part of the SC exits into the Gulf of Aden and another part continues to flow northward. A weak cyclonic eddy is present near Socotra at 15° N, 52.7° E with speed less than 0.5 m/s and towards north the circulation is weak. The southward flow of WICC has almost ceased and no definite flow pattern is visible along the west coast of India.

Seasonal cycle during IOD events
As the IOD is the well-known dominant climate mode influencing the AS, it is important to derive its impact on surface circulation. The IOD index (Fig. 7) can be defined by the difference in SST anomaly between the western and south-eastern equatorial Indian Ocean. The positive (negative) value refers to the positive (negative) IOD event. The changes in mesoscale surface circulation during positive and negative IOD events were identified by comparing the instantaneous velocity fields.
The surface geostrophic velocity fields in January, April, July and October of 1995and 1998 were analysed to understand the variations in surface flow during normal, positive and negative IOD years. Among these years, 1995 was a normal year, 1996 was a strong negative IOD year, 1997 was the strongest positive IOD year, and 1998 was a slightly weak negative IOD year.

January
In January 1995, the westward-flowing broad NEC was the most prominent flow, which extends northward up to 5° N and strong between 60° E and 70° E with a maximum speed of 1 m/s (Fig. 8a). The influx from the Bay of Bengal was clearly visible at the eastern end. In 1996, tremendous intensification was found in the westwardflowing NEC, as a broad strong flow with speeds up to 1.5 m/s between 60° E and 75° E. After crossing 60° E, the westward flow was suddenly weakened, merged with the eastward flow from the African Coast at around 55° E and turned north-westward. In 1997, the NEC was slightly weakened compared to 1996 and its speed was decreased to 1 m/s. When it reaches 60° E, it takes an anticyclonic turn and forms a clockwise gyre between 60° E and 70° E. The eastward flow from the African Coast is strengthened and reaches 55° E. Then it turns north-westward and feeds the western boundary current. In 1998, the flow pattern was entirely different; the westward-flowing NEC was not clearly visible and weak eddies were observed near the equator. The eastward flow on the western side was clearly visible in 1998.
In 1995, the weak narrow northward-flowing boundary current was present from 3° N to 12° N and multiple eddies were observed in the north-western part along the western boundary. A part of northward flow intruded into the Gulf of Aden and the major part was turning eastward between 5° N and 10° N. A similar flow pattern was observed in 1998 at the western boundary. In 1996, flow pattern along the western boundary was a bit different; a current divergence region was found at around 6° N, and the northward flow was strong and well organised. A major part of the northward flow intruded into the Gulf of Aden, where it took a cyclonic circulation and flows towards the Red Sea. In 1997, also the current divergence was at around 6° N but the northward flow was embedded with eddies. In 1998, the current divergence was again shifted southward at around 3° N and an organised eastward flow was initiated at around 7° N. In 1995, the northward-flowing WICC was observed along the west coast of India with speed of 0.2 m/s. The northward WICC was more prominent in 1996 and 1997. In 1998, the coastal flow was similar to 1995. An anticyclonic eddy was observed at the southern end (7° N, 74° E) of the WICC in 1995. Another weak anticyclonic eddy was detected at 9° N, 69° E. The anticyclonic eddies were more prominent in 1996. The anticyclonic eddy at 9° N, 69° E was absent in 1997. The condition in 1998 was same as in 1995.

April
In April, flow near the equator was weakened and flow pattern was similar in 1995, 1996 and 1998, where no organised flow was observed (Fig. 8b). In 1997, a well organised westward flow was present between 2° N and 4° N, and between 50° E and 60° E in the western region with a speed of 1 m/s. In 1998, no organised flow was found near the equator and a weak eastward flow was initiated at the western end. The western boundary current was further strengthened in April with maximum speeds above 1 m/s. A part of the current escaped into the Gulf of Aden and the major part continued northward and embedded with eddies. In 1995, the GW was developing at 7.5° N, 51° E. It was shifted to around 54° E and comparatively weak in 1996. Another weak anticyclonic eddy was observed south of this eddy at 6° N, 52° E. In 1997, the main eddy was found to be in its more matured state, compared to other years and a weak anticyclonic eddy was observed at 13° N around 58° E. In 1998, the anticyclonic eddy was just in its formation stage. The southward-flowing WICC was strong in 1996, compared to other years and the central AS was slightly active with multiple weak eddies, compared to January.

July
In July of 1995 and 1996, the surface circulation near the equator was not well organised and embedded with weak eddies on the eastern side (Fig. 8c). A westward flow prevailed in 1997 and 1998, and was much stronger in 1997. The northward-flowing western boundary current was further strengthened in all years, compared to April. It attained a maximum speed up to 1.9 m/s near the Somali Coast and the GW was shifted slightly to the north. An anticyclonic eddy was found near Socotra at 13° N, 53° E with speed of 0.7 m/s in 1997. The western boundary current was fed by both the flow from south of the equator and the westward flow near the equator. Multiple  1995, 1996, 1997 and 1998. b Maps of mean surface geostrophic velocity fields (m/s) in April 1995April , 1996April , 1997April and 1998. c Maps of mean surface geostrophic velocity fields (m/s) in July 1995July , 1996July , 1997July and 1998. d Maps of mean surface geostrophic velocity fields (m/s) in October 1995and 1998Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022  mesoscale eddies were present at the western boundary, especially in the northern part. The southward-flowing WICC was clearly visible even from 15° N in 1997, but weak in 1995. At the southern end, the WICC merged with flows from the central AS and turned towards the Bay of Bengal, except in 1997.

October
In October, an eastward flow towards the Bay of Bengal was found near the equator and more organised at the eastern end with maximum speed of 1 m/s in all years, except in 1997 (Fig. 8d). In 1997, the flow near the equator was westward. The northward-flowing western boundary current was slightly weakened with a speed between 1 and 1.3 m/s. The anticyclonic eddy was also weakened. In 1995, a broad northward flow was observed at the western boundary with a jet-like flow between 10ºN and 15° N and many mesoscale eddies were found in the northern part. A cyclonic eddy was observed northeast of the GW at 13° N, 54° E near Socotra (Socotra eddy). A weak northward flow was observed along the eastern side, except in 1997. In 1997, northward flow was more prominent with a speed of 0.2-0.5 m/s.

Circulation in fall during IOD events
Analysing the circulation features of IOD years, it was found that major changes were occurring in fall. Hence, composite maps of surface geostrophic velocity fields in October, November and December for positive and negative IOD years were presented. Further, detailed views of the western, eastern and equatorial regions in October and November of 1996 and 1997 were also presented. The monthly-mean velocity composite maps (Fig. 9a) of October, November and December for positive IOD years (1994, 1997 and 2006) and negative IOD years (1996, 1998 and 2005) clearly illustrated distinctive circulation patterns in the AS. The equatorial circulation was significantly different during positive and negative IOD events. The composite maps visibly show the absence of the Equatorial Jet in fall, instead weak anticycolonic flow was found.
As it observed in individual years, the western boundary shows more intense flows during negative IOD years; the GW was very powerful in October and weakened in November, but still traceable in December. During positive IOD years, the GW was present in October, but much weakened in November and not noticeable in December.
Another interesting feature was the reversal (southward flow) of the SC during negative IOD years from south of around 7° N in December, whereas during positive IOD years no regular southward flow was found. The southward flow was supporting the eastward flow near the equator between 2º N and 3° N. The eastward current was strong up to 70° E, where the major part took a cyclonic turn and a minor quantity was moving eastward. A cyclonic circulation was present near the equator between 2° N and 10° N during negative IOD events and an anticyclonic circulation prevailed during November-December between 2° N and 6° N during positive IOD years.
Along the west coast of India, the flow pattern depicted significant changes between positive and negative IOD events. The coastal current was sluggish during October-November during both positive and negative IOD events. But in December, an organised northward flow of WICC was present during negative IOD events, while no systematic flow pattern was found during positive IOD events. The NEC was bifurcated at 75 ° E around 5 ° N; its northward branch fed the WICC and southward branch took a cyclonic turn and merged with the eastward equatorial flow.
A detailed view of the equatorial region in October and November of 1996 and 1997 is given in Fig. 9b. In October 1996, no regular flow was observed in the western part. But the overall flow was towards east from 60° E and attained a speed above 1 m/s near the equator at 75° E. The equatorial flow was broad and strong even from 50° E in November 1996. Thus, the eastward flow near the equator was traceable during the negative IOD event (1996).
In October 1997, irregular flow patterns with weak eddies were found in the western region. But a westward flow was evident from the east up to 57° E, where it took an anticyclonic turn and formed a gyre around 6° N. By November, the equatorial anticyclonic gyre was strengthened and occupied the region between 2° N and 6° N, and between 55° E and 72° E. The westward equatorial flow even reached up to Somali Coast and maintained a speed above 1 m/s between 52° E and 72° E.
The direction and speed of the zonal components along different longitudes (60° E, 65° E and 70° E) between 2° N and 10° N in November of 1996 and 1997 portrayed the major variations that took place near the equator (Fig. 9c). During the negative IOD (1996), the eastward current with speeds above 0.5 m/s was found near the equator up to 4° N. Whereas during the positive IOD (1997), the zonal current is directly opposite (westward) and very strong as a speed up to 1.0 m/s at 70° E. Remarkably, a strong eastward flow occurred between 4° N and 6° N during the positive IOD. Thus, a strong anticyclonic gyre circulation was found in 1997 near the equatorial region. Beyond 6° N the zonal current was weak and almost similar in both years.
Page 15 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022)  Page 18 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) 33:22 The detailed view of the velocity field of the western and eastern AS in December of 1996 and 1997 is shown in Fig. 9d. The northward-flowing WICC was observed during the negative IOD event (1996). Also, the southward flow of the SC was observed in 1996 while no organised flow was found in 1997. The wind pattern over the AS in November and December of 1996 and 1997 also showed corresponding changes (Fig. 9e). Significant variations in wind pattern were found in the equatorial region in November. Along the equator, the easterly wind was dominating in 1997, whereas it was westerly in1996. Strong wind was observed in the south-western AS in December of 1996, compared to 1997. Along the west coast of India, it was easterly in 1996, while the wind was weak and northerly in 1997.

Circulation and Chl-a distribution
The Chl-a distribution in the AS was analysed along with circulation features that reaffirms the precision of the velocity estimation using the present method. The monthly-mean Chl-a in January, April, July and October from SeaWiFS observations during ten years were displayed in Fig. 10a. Enhanced Chl-a of about 1 mg/m 3 was found in January almost over the whole AS, except in the south-eastern region, where it was below 0.3 mg/ m 3 . Chl-a concentration was maximum along the Arabian Coast and northern part of the Indian Coast. In April, Chl-a was significantly diminished (below 0.3 mg/ m 3 ) near the equator to about 15° N region. A tremendous increase in Chl-a distribution was evident in July (Chl-a observations were missing due to cloud over black shaded region), the summer monsoon period. The strong wind-driven upwelling along the western boundary coast as well as the mesoscale eddies were significantly controlling the Chl-a concentration. Also, the generated coastal upwelling enhanced Chl-a along the Indian Coast. By October, the southwest monsoon circulation had ceased, upwelling was weakened and Chl-a concentration was slightly diminished.
As major changes in surface circulation were found in the later periods of IOD events, the Chl-a distribution in  Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) 33:22 October and November during the positive IOD (1997) and negative IOD (1998) were compared (Fig. 10b).
Positive IOD (1997): In October, the conspicuous feature was the high Chl-a concentration along Arabian Coast. As the northern arm of the anticyclonic eddy at 8 °N limited the intrusion of Chl-a rich waters towards Somali region, low Chl-a concentration (below 0.2 mg/ m 3 ) prevailed along the Somalia Coast. Chl-a distribution was moderately high along the west coast of India. The WICC was clearly visible as a northward flow and aligned enhanced Chl-a waters in the coastal region without spreading much too offshore. In the western AS, Chl-a concentration was less than 0.2 mg/m 3 along the Somali Coast, while off the Arabian Coast Chl-a was greater than 1 mg/m 3 . In November, the Chl-a concentration was significantly reduced both along Arabian Coast and Indian Coast. Analogous changes were found in circulation features; the mesoscale activity was much diminished in the western part. The anticyclonic gyral circulation near the equator was prominent and the WICC was not visible during this period.
Negative IOD (1998): In October 1998, enhanced Chl-a concentration was found in the whole basin, except near the equator, compared with October 1997. A high Chl-a tongue extending from the west coast of India was found at 12° N, 70° E. A high Chl-a patch was found around 60° E with a latitudinal extent between 12° N and 16° N. In November 1998, the Chl-a concentration was further diminished over the AS. The coastal circulation was weakened and Chl-a concentration was also reduced.
The equatorial region also displayed substantial influence of flow pattern on Chl-a distribution. In November 1997, low Chl-a band was found; a strong anticyclonic gyre circulated the low Chl-a water to the central part of the basin. The flow pattern was almost reversed in 1998; a cyclonic pattern near the equator carried water from the west to east.
Detailed examination of the Chl-a distribution in the western AS in October 1997 (Fig. 10c) indicated that the currents along the Somali Coast were much weakened and the northward flow from the south brought low Chl-a waters to the Somali Basin. In addition, the anticyclonic eddy between 5° N and 10° N reduced the upward nutrient flux to the surface layers. Even the weak anticyclonic eddy north of 10° N also influenced Chl-a concentration in November 1997. A series of cyclonic eddies modified the Chl-a distribution off the Arabian Coast. The Chl-a concentration was almost 0.6 mg/m 3 over the north-western region, except a slight decrease between 10° N and 15° N where an anticlockwise circulation occurs.
Enhanced Chl-a distribution was found in November 1998, compared to 1997. The cyclonic eddy (at 5° N) neighbouring the anticyclonic eddy had significantly modified the offshore Chl-a concentration. Further, Chl-a above 1 mg/m 3 found around 15° N was due to the strong cyclonic eddy and extended more into the offshore region.
Further, the wind pattern in October of 1997 and 1998 was significantly different (Fig. 10d). Strong south-westerly wind was present during the negative IOD (1998), whereas the wind was weak and easterly during the positive IOD (1997). The south-westerly wind in October 1998 was favourable for upwelling and enhanced Chl-a in the western AS (Fig. 9c). In October 1997, upwelling was absent and diminished Chl-a concentration.

Discussion
The AS undergoes large seasonal changes in response to the monsoon winds, leading to a reversal of the AS surface circulation. The present study derived mean and seasonal variability of near-surface geostrophic flow by combining surface drifter and satellite altimetry data. The spatial and temporal changes due to IOD also identified. The estimated geostrophic flow near the equatorial region is westward during winter monsoon, weakly westward during summer monsoon and strongly eastward in fall, which matches with earlier observation. Ship drift climatology of Cutler and Swallow (1984) indicates that surface currents in the equatorial Indian Ocean reverse direction four times a year, flowing westward in winter, weakly westward in the central and western oceans in summer and strongly eastward in spring and fall.
The climatological-mean surface currents for the winter monsoon show the westward-flowing Northeast Monsoon Current (NMC; Tomczak and Godfrey 2003) south of India and Sri Lanka, which carries water from the Bay of Bengal towards the AS. Along its way, some fraction of the NMC splits off north-eastward toward the Indian Coast, flowing anticyclonically around the Laccadive High with its centre around 10° N, 70° E (Bruce et al. 1994). A small anticyclonic eddy was found at 7.5° N, 75° E in the present study. The NMC then continues westward at low latitudes toward the African coast, bifurcating off Somalia in the near-surface layer at about 8° N (Schott and Fischer 2000). The southward-flowing branch supplies the SC. The northward branch splits south of Socotra, with a part of the flow continuing towards the Gulf of Aden through the passage between Socotra and the African continent, while another fraction supplies the interior northern AS around Socotra to the east (Schott and Fischer 2000). This feature of the northward flowing branch of the SC is clearly visible in the present estimation.
Ship-drift observations (Cutler and Swallow 1984;Shetye et al. 1990) show the weak southward WICC along the west coast of India, with a magnitude of 0.2-0.3 m/s, during summer monsoon. These observations are consistent with the present estimation. In the present analysis, with the onset of the southwest monsoon, the circulation in the AS evolved into a complex pattern of mesoscale eddies and gyres. This matches with observations of Bruce (1983), andLuther andO'brien (1985).
The seasonal development of the SC system was described by Schott et al. (1990). According to them, before the onset of monsoon during March-May, the southern SC is an extension of the East African Coastal Current that flows northward across the equator to 3° N with a magnitude of 1 m/s. There it turns offshore forming a cold wedge along its shoreward side. The present estimation of the SC was northward with a magnitude of 1 m/s from the southern end. With the monsoon onset in June, the cross-equatorial flow strengthens to the order of 1.8 m/s and leaves the coast south of 4 ° N, where it partially turns eastward and partially flows back across the equator in a circulation pattern referred to as "Southern Gyre (SG)". Another circulation, the "Great Whirl (GW)", develops between 4 ° N and 12 ° N with a second cold wedge where it turns offshore (Schott et al. 1990). A third gyre, the "Socotra Eddy (SE)", was seen northeast of Socotra during summer monsoon (Schott and McCreary 2001). The GW starts developing in April and an eastward flow is observed from the right arm of the GW. In July, the fully developed northward-flowing SC and GW were visible in the present maps. Speed of the SC is found to be 1.8 m/s and the position of the GW was between 4° N and 12° N. A small anticyclonic eddy is observed north of the GW near Socotra at 13° N, 53° E with a speed of 0.7 m/s. During October-November, when the southwest monsoon dies down, the cross-equatorial SC turns offshore at 3° N with a magnitude of 1.3 m/s. The GW is even discernible underneath the developing northeast monsoon circulation well towards the end of the year (Schott et al. 1990;Schott and McCreary 2001). In October, the speed of SC was 1.3 m/s and the GW moved southward, compared with July. The eddy was still strong and fully developed. Using satellite altimetry data, Trott et al. (2018) tracked the GW and found that the location of GW development and propagation shows significant variability. Melzer et al. (2019) suggests a high-frequency periodicity in the size and strength of the GW and interannual variability of the life cycle reveals a later termination date than previously thought.
During the northeast monsoon (winter), the winds blow away from the Indian subcontinent and the surface SC reverses to flow southward (Schott et al. 1990;Schott and McCreary 2001). Moored array observations during WOCE (Schott and Fisher 2000) showed that the northern SC during this time was characterized by an inflow from the east, causing a divergence at the coast near 6-8° N, with a northward surface flow north of these latitudes and equatorward surface flow south of them. The westward upper-layer flow into the SC system north of 7° N in winter was also confirmed by the Expendable Bathythermograph data analysis of Donguy and Meyers (1995). A consistent flow pattern was obtained in January, illustrating southward flow of the SC south of 5° N and northward flow north of 5° N. Also the westward surface flow into the SC occurs north of 5° N.
The AS shows significant spatial and temporal changes in the flow pattern. Visible change is taking place in the equatorial region during fall, where the eastward Equatorial Jet is absent and instead reverse flow is taking place during the positive IOD event. As the easterly wind is strengthened during the positive IOD, the surface eastward flow is replaced by the westward current . The strong eastward Equatorial Jet is directly forced by the semi-annual component of winds (Han et al. 1999), which was not happening during the positive IOD. Both in spring and fall, the Equatorial Jet was absent in 1997 and instead a westward current was present. The wind pattern in November and December evidently displayed significant changes in 1996 and 1997. Along the equator, the easterly wind field was found in 1997, whereas it was westerly in 1996. Moreover, the strong westward flow constitutes an anticyclonic gyre near the equator with its eastward limb around 6° N. This equatorial anticyclonic gyre was a conspicuous feature found during the positive IOD event in the fall.
Significant changes were found in the south-western AS during the IOD events. During the negative IOD event, strong mesoscale activity was occurring with the presence of multiple eddies. However, the circulation was weak during the positive IOD event. The SC reversal was also consistent with the strong north-easterly wind along the Somali Coast (around 6° N) in 1996. Furthermore, the wind pattern along the west coast of India was favourable for the organised northward flow of WICC in December 1996, while the wind was weak and southward in1997.
The GW also shows conspicuous changes in its size, formation and existence. Intra-seasonal fluctuations in the south-western AS are reported by Brandt et al. (2003) and the instabilities of the flow in the region between the SG and GW are likely the sources of these fluctuations. Interannual variability of intra-seasonal fluctuations are possibly related to the GW dynamics and especially its decay and collapse (Brandt et al. 2003). During the positive IOD event, the GW was moderately developed in April. Whereas, multiple weak eddies occupies during negative IOD. Beal et al. (2013) noted the presence of a weak anticyclonic eddy in April as the initial development of the GW. It seems that the GW was decaying early during the positive IOD event, compared to the negative IOD event. The GW was even stronger in October and was traceable in December as a weak anticyclonic eddy during the negative IOD. Whereas, the GW was fairly weak even from November during the positive IOD event. Hence, the earlier formation and dissipation of the GW was found during the positive IOD event, compared to the negative IOD event.
Furthermore, northward flow was found along the western boundary during the positive IOD event and a part of it grows as the GW. But, during the negative IOD event, the westward equatorial flow that takes an anticyclonic turn between 50° E and 60° E was imputing to the GW. Thus, during the positive IOD, the GW was maintained by the northward coastal current and the equatorial contribution is significant during the negative IOD event. Substantial differences were also found in the north-western AS, where a broad north-eastward flow (Beal et al. 2013) was occurring during the negative IOD, while weak eddies were prevailing during the positive IOD. The southward WICC was stronger during southwest monsoon during the positive IOD, whereas the northward WICC was much stronger in fall during the negative IOD.
The mesoscale circulation features were consistent with the Chl-a distribution in the AS. The Chl-a distribution is largely controlled by the upwelling systems along West Indian and Somali coasts. During the southwest monsoon, huge quantities of nutrients are dumped into the sea along with river discharge, increasing the Chl-a concentration along the West Indian Coast. Wind-driven upwelling is significantly influencing the Somali Coast. In addition to the strong northward SC and upwelling off the Somalia Coast, the southwest monsoon also drives Open Ocean Ekman pumping in summer (Murtugudde et al. 1999). The nutrient-rich waters upwelled off the coast are advected into the interior and add significantly to the offshore blooms in the AS, which is driven mainly by mixed-layer entrainment of nutrients (McCreary et al. 1996).
The high Chl-a concentration is confined to coastal regions all around the basin. The Chl-a concentration in the northern AS is significantly high in all seasons. Tang et al. (2002) reported that in the northern AS, the rate of phytoplankton cell division is controlled by nutrient availability rather than light, while light inhibition of photosynthesis near the surface is negligible. The boundary and open ocean processes in the AS is influenced by upwelling in summer and cooling in winter; the upwelling brings high amounts of nutrients into the upper ocean, enhances primary productivity (Madhupratap et al. 2001).
Enhanced Chl-a concentration was found in the western AS during the negative IOD, where the wind-driven upwelling and intense mesoscale activity were occurring. Impacts of the 1997 IOD event induced a decrease of surface Chl-a in the AS (Sarma 2006), which was attributed to anomalous north-easterly winds. The change in wind pattern, surface circulation and the mesoscale eddies are largely influencing the Chl-a distribution in the AS. Wiggert et al. (2009) identified the contrasting biological response of the western AS to the 1997/1998 and 2006/2007 IOD events, with an overall decrease of productivity during the 1997/1998 event and a slight increase during the 2006/2007 event. While the AS experienced changes of − 9 and + 15% in basin-wide primary production during 1997 and 2006 events respectively. The monthly-mean surface circulation pattern estimated in this study is explaining the different Chl-a distribution. During the 1997/1998 event, the anticyclonic GW (5° N-10° N) was strong and prevails near the Somali Coast and associated downwelling limits the nutrients supply. There is a significant change found in mesoscale dynamics, during the 2006/2007 event. The surface circulation in October 2006 illustrates the strong SC (1.5 m/s) extending up to 4° N where it turns offshore and flow up to 58° E and forms a strong cyclonic eddy (a conspicuous feature) in the western AS, which occupies the region between 3° N and 8° N, and between 50° E and 58° E. This cyclonic eddy maintained strong upwelling that added more nutrient to surface levels and hence, enhanced the Chl-a concentration during 2006/2007 (Fig. 11). The weak GW was also evident north of the cyclonic eddy. Another strong cyclonic eddy found at 15° N, 55° E was also enriching with nutrients and the Chl-a concentration. Thus, the presence of strong cyclonic eddies and weakening of GW in 2006 were the reason for the Chl-a enhancement.
Enhanced Chl-a was around the southern tip of India in fall during the negative IOD (1998), while the Chl-a was much diminished during the positive IOD (1997).
The monthly-mean circulation shows that the presence of multiple cyclonic eddies in this region, which was partly enhancing the Chl-a through upwelling during the negative IOD. Wiggert et al. (2009) found negative influence of IOD on Chl-a around the southern tip of India in fall (October-December).

Conclusions
Combining satellite altimetry and surface drifter observations the present study derived the mean surface velocity field and brings out the details of magnitude of various currents in the AS and their mean and seasonal variability. The major currents observed in the mean field are western boundary Somali Current, weak westward NEC and southward WICC.
Seasonal circulation displays a westward flow in the equatorial region except in fall. The zonal eastward-flowing Equatorial Jet is visible in fall. The SC flows northward in summer and southward in winter and is the strongest current in the AS with mean speed above 1.8 m/s in July. Seasonal changes are also found in the WICC, which flows northward in winter and southward in summer. Besides strong currents, mesoscale eddy activity is intense in the western AS. The GW and Socotra Eddy are prominent during the summer monsoon.
The AS surface circulation shows significant variations during IOD events. The major changes are occurring along in the equatorial zone and western (Somali Coast), and eastern boundaries. The significant change in the equatorial region is the absence of the Equatorial Jet during the positive IOD in fall; instead an opposite (westward) flow occurs. The SC is strong but narrow during the positive IOD. Whereas, the SC is broad and the GW extends more to the north in the negative and neutral IOD years, compared to the positive IOD. An earlier formation and dissipation of the GW was found during the positive IOD event, compared to the negative IOD event. Further, a delay in the reversal of the SC is found during the positive IOD event. In the eastern AS, the southward WICC was stronger during the southwest monsoon during the positive IOD, whereas the northward WICC was much stronger in fall during the negative IOD. Another conspicuous feature observed is the anticyclonic gyral circulation near the equatorial region during the positive IOD event in fall.
The Chl-a distribution in the AS shows significant influence of eddy activity; the cyclonic and anticyclonic eddy locations were well consistent with the high and low Chl-a concentrations, respectively. Higher Chl-a in the western part is caused by the strong wind-driven upwelling along the Somali Coast and the mesoscale eddies contributing to the regional highs and lows. The Chl-a distribution in the AS shows significant influence of IOD events. Enhanced Chl-a is found during the negative IOD in fall, especially in the western AS, compared to the positive IOD. The difference of Chl-a distribution during positive IOD years between 1997 and 2006 in the western AS is due to the change of mesoscale eddy activity.
The AS surface circulation is highly varying with respect to time and space. Monsoonal circulation is the dominant pattern in the annual cycle, whereas interannual variability is modulated by the IOD events. Extensive mesoscale activity is occurring near the Somali Coast and near the equator. Significant changes are induced in surface circulation and Chl-a distribution in the AS by IOD activity. Page 25 of 26 Peter et al. Terrestrial, Atmospheric and Oceanic Sciences (2022) 33:22