Surface Current Observations in the Southeastern Tropical Indian Ocean Using Drifters

Abstract

The Southeastern Tropical Indian Ocean (SETIO) forms part of the global ocean conveyor belt and thermohaline circulation that has a significant influence in controlling the global climate. This region of the ocean has very few observations using surface drifters, and this study presents, for the first time, paths of satellite tracked drifters released in the Timor Sea (123.3° E, 13.8° S). The drifter data were used to identify the ocean dynamics, forcing mechanisms and connectivity in the SETIO region. The data set has high temporal (~5 min) and spatial (~120 m) resolution and were collected over an 8-month period between 17 September 2020 and 25 May 2021. At the end of 250 days, drifters covered a region separated by ~8000 km (83–137° E, 4–21° S) and transited through several forcing mechanisms including semidiurnal and diurnal tides, submesoscale and mesoscale eddies, channel and headland flows, and inertial currents generated by tropical storms. Initially, all the drifters moved as a single cluster, and at 120° E longitude they entered a region of high eddy kinetic energy defined here as the ‘SETIO Mixing Zone’ (SMZ), and their movement was highly variable. All the drifters remained within the SMZ for periods between 3 and 5 months. Exiting the SMZ, drifters followed the major ocean currents in the system (either South Java or South Equatorial Current). Two of the drifters moved north through Lombok and Sape Straits and travelled to the east as far as Aru Islands. The results of this study have many implications for connectivity and transport of buoyant materials (e.g., plastics), as numerical models do not have the ability to resolve many of the fine-scale physical processes that contribute to surface transport and mixing in the ocean.

1. Introduction

Surface circulation of the ocean (<15 m water depth) is primarily driven by winds and is a conduit for the transport of natural and anthropogenic buoyant material such as eggs and larvae of different organisms [1], plastics and dispersion of oil [2]. The Southeastern Tropical Indian Ocean (SETIO) is located between the Australian mainland and the Indonesian Archipelago (Figure 1a) and is also called the Indo-Australian Basin (IAB; [3]). It is a geographically important region for the transport of water from the Pacific Ocean through the Indonesian Throughflow (ITF) into the South Equatorial Current (SEC) in the Indian Ocean [4,5]. It also forms part of the global ocean conveyor belt and thermohaline circulation [6]. The region has been the subject of several oceanographic and meteorological studies [7,8,9,10,11,12] and experiences large sea surface temperature fluctuations, contributing to the Indian Ocean Dipole formation [13] (and is also influenced by the El Niño–Southern Oscillation (ENSO) variability [14,15,16]. The SETIO is affected by the Australasian monsoon and trade winds such that winds are southwesterly from December to March and southeasterly from April to September. This region also has a significant influence on rainfall in western and southern Australia through the generation of a northwest cloud band or atmospheric rivers [17]. SETIO is also the only known spawning ground for the southern bluefin tuna (Thunnus maccoyii) and a major fishery ground for yellowfin tuna (Thunnus albacares) [18,19]. Major currents in the region include (Figure 1a): the South Equatorial Current (SEC), Indonesian Throughflow (ITF), Eastern Gyral Current (EGC), South Java Current (SJC), and Holloway Current (HC).

Warmer, lower salinity ITF provides a tropical connection between the Pacific and Indian Oceans, with its main pathway through the Timor Passage (Figure 1a; [7,20,21,22]). Recent volume transport estimates of the ITF transiting through the Timor Passage have indicated an annual mean of 7.5 Sv, which comprises 50% of the total ITF (~15 Sv) to the Indian Ocean [22,23]. There is also flow through the passages between the different Islands, notably Lombok Strait, which amounts to 20% of the flow into the Indian Ocean. The ITF contributes to the SEC, the major east to west flow across the southern Indian Ocean basin, between 10° S and 20° S, which influence the mass and heat exchanges among the Pacific, Indian, and Atlantic Oceans (Figure 1a). The SEC transports warmer, lower salinity water originating from the ITF water into the Indian Ocean interior and to the western boundaries of the Indian Ocean (Figure 1a). The maximum volume transport > 17 Sv coincide with the peak of ITF transport and negative wind stress curl in July–September [24,25]. The eastward-flowing EGC is a continuation of the northern South Indian Counter Current (nSICC) (Figure 1a; [22,26,27]).

The bi-annually reversing South Java Current (SJC) flows along the Indonesian Islands, Sumatra and Java [28,29,30,31,32]. The SJC is driven by the monsoon winds and variation in freshwater influx from the Indonesian Archipelago [29,33]. From November to June, the SJC flows eastward along the southern coast of Java (between 100 and 120° E) with peak speeds up to 1.5 ms⁻¹. From July to October, the SJC flows to the west. Studies have revealed the rapid reversal in the SJC is associated with peak eastward flow in May–June and October–November from downwelling Kelvin waves associated with westerly wind bursts in the equatorial Indian Ocean.

Mesoscale eddies, both clockwise and anticlockwise, are formed in this region due to the shear/baroclinic instability/topographic interference between the three major currents (SEC, ITF, and SJC) [3,24,30,34,35,36,37,38]. The region where the mesoscale eddies are present is associated with a region of high Eddy Kinetic Energy (EKE > 0.1 m²s⁻²) to the south of Java coast bounded by 8–15° S and 105–120° E (Figure 1b,c). This high intra-seasonal EKE variability domain within the SEC was identified by many investigators [3,35,39,40,41,42,43,44,45]. However, the importance of this high EKE region for surface material transport has not been addressed to date.

Figure 1. (a) Schematic of the surface currents in the SETIO modified from Wijeratne et al. [22]. The Currents are as follows: ITF—Indonesian Throughflow; SEC—South Equatorial Current; SJC—South Java Current; nSICC—northern South Indian Counter Current; EGC—Eastern Gyral Current; HC—Holloway Current; and, (b) Climatology of the eddy kinetic energy in the study domain obtained from satellite altimeter data. Units are m²s⁻². (c) Drifter tracks superimposed on the eddy kinetic energy.

Figure 1. (a) Schematic of the surface currents in the SETIO modified from Wijeratne et al. [22]. The Currents are as follows: ITF—Indonesian Throughflow; SEC—South Equatorial Current; SJC—South Java Current; nSICC—northern South Indian Counter Current; EGC—Eastern Gyral Current; HC—Holloway Current; and, (b) Climatology of the eddy kinetic energy in the study domain obtained from satellite altimeter data. Units are m²s⁻². (c) Drifter tracks superimposed on the eddy kinetic energy.

Trajectories of water parcels in the ocean, defined as the Lagrangian description of the flow, are useful for both visualising ocean motion and determining its velocity characteristics and forcing mechanisms. Drifting buoys, or drifters, are one of the oldest instruments in physical oceanography and have been used to study near-surface horizontal currents throughout the global ocean [46,47]. Surface drifters track the moving water parcels (how, where and speed) and associated mixing. Superposition of numerous trajectories reveal different circulation patterns in both space and time. Surface drifters measure the transport of water parcels and have been used for as long as people have been going to sea [48]. The earliest measurements were made by taking visual sightings of natural and man-made floating objects within sight of land or from an anchored ship, which served as a reference [48]. The essential measurement of a drifter is the changing location with time. In early years of 1950s, visual sightings using telescopes, compasses, and sextants were used to measure bearings and locations, whist radio for direction finding, and radar observations were used to track the drifter location [48]. The automated use of surface drifters for ocean current measurements over large distances and time scales became possible through the development of satellite positioning systems in the 1970s and the Global Positioning System (GPS), which was fully operational in 1993. Removal of Selective Availability (SA) in the early 2000s allowed for the development of smaller cheaper standard GPS receivers that can fix the position within a few metres globally [49].

Satellite Lagrangian tracked drifters has been widely used to elucidate the upper ocean currents in different oceanic regions [46]. They have been used for many applications from the surf zone [50,51,52], estuaries [53], continental shelves [47,54,55,56,57] and in the open ocean [58,59,60,61,62]. Surface drifter data have also been used to construct transport matrices for long-term simulation of buoyant material [63,64]. Surface drifters have been used in the Indian Ocean to map surface currents [65,66,67,68,69,70,71]. The results of these drifter deployments revealed mesoscale variability resulting from eddies and meanders particularly in the SEC and seasonal changes in surface currents [68,69]. Statistical analysis of drifter tracks revealed regions of high currents and diffusivity in the Indian Ocean basin [70,71]. However, these studies did not include the whole SETIO region.

This paper describes a drifter deployment undertaken on 17 September 2020 where 15 identical drifters were released simultaneously at the same time and location and were tracked over a period of 8 months. The drifters were purpose-built and were undrogued as the primary concern was to define the transport of buoyant materials, such as plastics and larvae. Drifters were deployed offshore at 123.3° E, 13.8° S (Figure 1a) where the water depth was ~200 m. At the end of the experiment, after 250 days, the 15 drifters were separated by ~8000 km. As the drifters transited through the SETIO they experienced different forcing mechanisms such as tides, inertial currents, submesoscale, and mesoscale eddies. The aims of the paper are to analyse the drifter tracks over the 8-month period and identify the ocean dynamics with respect to changing forcing mechanisms of the surface currents with time and space across the region. The experiment also revealed very high separation of the drifters over a period of ~8 months.

The paper is structured as follows: Section 2 presents the methodology which includes details on the drifter and the data analysis techniques. Section 3 explains results of the experiment with details of the drifter trajectories along with description of the forcing mechanisms. Discussion and conclusions are given in Section 4 and Section 5, respectively.

2. Methodology

2.1. The Surface Drifters

Surface currents in the SETIO were observed using satellite tracked, easily deployable, purpose-built low cost (<U.S. $200 each) Lagrangian surface drifters. They were constructed at The University of Western Australia (UWA) based on two decades of developing low-cost drifters for the surf zone and the open ocean [49,50,51,56,57]. The drifter housing was purchased at a hardware store and built from standard cylindrical Polyvinyl Chloride (PVC) drainpipe parts, such as those used in residential houses; it consisted of two main sections of varying diameter (Figure 2). The total length of the drifter was 0.60 m, and its weight was <1 kg (Figure 2). The upper part of the housing contained the telemetry module, which was filled with expanding foam and had a positive buoyancy to keep the drifter upright in seawater. This upper section was 0.20 m in length and 0.10 m in diameter (Figure 2). The lower section was 0.40 m in length and 0.06 m in diameter and housed the battery pack along the whole length of the section acting also as the ballast. Lithium batteries in each drifter were designed to last for up to 8 months. Drifters were not designed to be recovered. Each drifter, including the data plan for satellite data transmission, costs < US $200 each. The telemetry module was a commercial, off-the-shelf GPS receiver SPOT Trace^TM device (Globalstar^®, Covington, LA, USA), which has been used in many recent studies [53,54,55,56,57,60,72,73]. The SPOT Trace^TM device was set to transmit its location nominally at five-minute intervals through the Globalstar^® constellation. The drifter units are neutrally buoyant so that only the upper surface covering the internal GPS received is above the water. Under calm seawater conditions, ~2 cm of the drifter body was exposed above the sea surface, such that windage was negligible. That is, the direct effect of the wind and inertia on the drifter was low (see also [56]). The windage coefficient R_windage is usually taken as follows [74]:

$$R_{windage}=\sqrt{{\displaystyle \frac{{\rho }_a}{{\rho }_w}}{\displaystyle \frac{A_a}{A_w}}{\displaystyle \frac{C_{Da}}{C_{Dw}}}}$$

(1)

where ρ is the fluid density, C_D is the object’s drag coefficient, A is the effective exposed area, and the subscripts a and w refer to air and ocean, respectively. Values of A_a (20 cm²) and A_w (560 cm²) for the drifters used in this study were determined in a tank filled sea water with a salinity of 35.5. For the UWA drifters, A_a/A_w = 0.036. The drag coefficient (C_Da/C_Dw) and density (ρ_a/ρ_w) ratios are assumed = 1 and = 1.17 × 10⁻³, respectively [56,75]. These values yield R_windage ~0.007 or windage to be 0.7% of the wind speed. This is also smaller than the value of 0.02 (or 2% of wind speed) reported by van der Mheen et al. [56].

We can also define wind slip as the horizontal motion of a drifter that differs from the motion of the surface currents [76]. Wind slip U_slip is defined as follows [54,77,78]:

$$U_{slip}={\displaystyle \frac{A}{R}}{W}_{10}$$

(2)

$$R={\displaystyle \frac{C_{Dw}}{C_{Da}}}{\displaystyle \frac{A_w}{A_a}}$$

(3)

where W₁₀ is the wind speed at a height of 10 m above the sea level and A = 0.07 [79], and R is the drag area ratio (DAR), which, using the values provided above, is 28. This results in U_slip to be 0.25% of the wind speed. This is a factor of 10 smaller than the 3% value usually quoted for wind induced surface currents in the ocean [56,80,81]. Also, as illustrated by Meyerjürgens et al. [54], the wind-induced slip resulting from wind speeds in the range of 1–10 ms⁻¹ would be 0.0027–0.027 ms⁻¹ in a downwind direction. This is much smaller in magnitude (<5%) compared to mean currents of 0.5 ms⁻¹ observed in SETIO during this study.

The drifter experiment described in this paper included the simultaneous deployment of 15 UWA surface drifters on 17 September 2020 at 02:05 UTC with all drifters released within 5 min at 123.3° E, 13.8° S, where the water depth ~200 m (Figure 1a). The drifter ID numbers were 4400419, 4400420, 4400657, 4402003, 4400425, 4400653, 4401668, 4401807, 4401995, 4400658, 4400898, 4401997, 4401998, 4400423, and 4401842. The SPOT Trace^TM device transmitted the longitude x(t) and latitude y(t) of the drifter position obtained from the satellite tracking system as a function of time. The drifter data had a temporal resolution of 3 to 6 min, and spatial gaps between 100 and 300 m. A total of 850,215 fixes was available for analysis before the QA/QC procedures (Section 2.2.1).

2.2. Analytical Approaches

2.2.1. Outlier Removal

Maximum current speeds of up to 1.5 ms⁻¹ for the SETIO surface currents has been reported in several studies [29,82]. Hence, a visual quality check based on a maximum current speed of 1.5 ms⁻¹ was set to identify the outliers and were removed manually. The manual examination also allowed for higher current speeds than this threshold to be accepted. A Hampel filter was also used to identify the outliers that were 3 standard deviations from the local mean (a MATLAB toolbox was used to apply the filter [83]). The measurement window for the application of the Hampel filter for each drifter was composed of the sample at time t and its six surrounding samples, three per side, i.e., a time window ranging between 35 and 65 min. This window of 7 (3 per side) samples was based on the sensitivity analyses that was undertaken that considered 5 (2 per side) and 18 (9 per side). The results indicated that window of 7 provided the optimum identification of the outliers in the drifter data set. For the entire drifter time series <0.2% of the data were marked as an outlier and discarded.

The operational procedure for the SPOT Trace^TM device was such that although the sampling interval was set to 5 min, if no position can be determined within 4 min—for instance, due to poor satellite reception—the SPOT Trace^TM device will try to determine a position at the next programmed tracking interval. In the case of no satellite connection, the device will store the coordinates until a successful transmission is made [54,60]. In this study, there were instances of no position location for up to 1 h. After removing the outliers, the east (u) and north (v) components of drifter time series were interpolated using a piecewise cubic interpolation method [84] to a common time base with a sampling interval of exactly 5 min. After interpolation, there were 922,379 data points available for analysis. The mean distance between each position fixes, after interpolation, was 120 m, i.e., for each drifter, the spatial resolution was 120 m.

2.2.2. Spectral Analysis and Wavelet Analysis

To identify the dominant frequencies present in the drifter time series records, Fourier and Wavelet transforms were used. Fourier analysis provided the frequencies present for the whole record, whilst the wavelet analysis included the time varying component. The individual drifter tracks were examined for different current patterns, and segments of the time series were extracted for analysis. Fourier transforms were used to construct auto-spectra with 32 degrees of freedom. Fast Fourier transform, Welch’s method, and the extracted time series were used to estimate the power spectra [85]. Wavelet transforms were used to analyse for non-stationary power at many different frequencies [86]. Time series of velocity components were decomposed using Torrence and Compo’s [86] continuous Morlet wavelets. The clockwise and anticlockwise components in the wavelet spectrum were calculated using MATLAB wavelet toolbox.

2.3. Kinetic Energy and Eddy Kinetic Energy

The drifter speed, direction, and east–west (u) and north–south (v) components were estimated using a central differencing scheme on each drifter position. All 15 drifters were binned into 0.5° bins, and the mean and standard deviation were calculated. The time series of daily mean speeds was used to calculate the total kinetic energy [87]:

$$KE={\displaystyle \frac{1}{2}}(u^2+v^2)$$

(4)

The altimeter delayed mode daily time series data over the period 1 January 1993 to 31 December 2020 (27 years), available through the Australian Ocean Data Network (https://portal.aodn.org.au/), were used for the analysis (accessed on 1 March 2025). The database includes Sea Surface Height Anomaly (SSHA) data (corrected using coastal tide gages) and derived geostrophic velocities in the east (u_g) and north (v_g) components. These were used to calculate daily values of Eddy Kinetic Energy (EKE) [87,88].

$$EKE={\displaystyle \frac{1}{2}}({u^{\prime }}^2+{v^{\prime }}^2)$$

(5)

where u′ = u_g − U and v′ = v_g − V. Here, we defined U and V as the daily mean climatology of zonal and meridional geostrophic velocity components, respectively. The altimeter delayed mode daily time series over the period 1 January 1993 to 31 December 2020 (27 years of data) were used to develop the daily climatology.

3. Results

The 27-year climatology mean EKE for the SETIO region obtained from geostrophic velocities derived from the satellite altimeter defined the major currents and current variability (Figure 1a). The South Equatorial Current (SEC), Indonesian Throughflow (ITF), Eastern Gyral Current (EGC), South Java Current (SJC), and Holloway current (HC) can all be identified through elevated EKE. However, the most dominating feature in the SETIO is the higher EKE in the region to the south of Java Island in the east of Cocos Keeling Islands (CKI) and Christmas Island (CI), bounded by 102–120° E and 8–15° S (Figure 1b). In this paper, we will define this zone as the ‘SETIO mixing zone’ (SMZ).

All 15 drifters were deployed on 17 September, and they remained close to each other as a single cluster until 2 October (15 days) when they crossed 120° E longitude and moved into the SMZ and 2 of the drifters were entrained into an anticlockwise eddy (Figure 3a). Subsequently, all the drifters were entrained into meanders and eddies and their movement was chaotic within the SMZ (Figure 3a,b). The maximum KE values were found in the SJC to the south of Java Island and in some of the eddy features (Figure 3b). When the drifters exited the SMZ, some moved to the west with the SJC and then to the east with the change in direction of the SJC, two drifters moved north through Lombok and Sape Straits and moved east to Aru Islands (Figure 1a). Others followed a westward track with the SEC. After 250 days the drifters were separated by distance up to 8000 km (4–21° S, 83–137° E) (Figure 3a).

The mean surface current speed of all the drifters, over the 8-month experiment (922,379 data points), was 0.37 ms⁻¹ (median = 0.34 ms⁻¹) (Figure 3c). Interestingly, if the mean was taken when the drifters were within the SMZ the mean speed was almost the same (0.36 ms⁻¹; median = 0.33 ms⁻¹). The mean speed outside the SMZ, after neglecting the higher currents in the island passages, was 0.36 ms⁻¹ (with median = 0.34 ms⁻¹). These results indicated that the currents within the SMZ were not significantly higher than the other regions, rather there was more variability through the presence of meanders, submesoscale, and mesoscale eddies. Maximum currents 2.5 ms⁻¹ (5 min averaged) were recorded in Sape Strait between the Sumbawa and Komodo Islands.

Time series of daily mean speed revealed that after deployment (maximum daily mean speed of 0.75 ms⁻¹), there was a linear decrease in current speeds to early October (Figure 3c). After entering the SMZ, all drifters travelled with mean daily speed of ~0.40 ms⁻¹ for the next 2 months. Some drifters (4400419, 4400420, 4400657, and 4402003) experienced relatively higher current speeds (>1 ms⁻¹) during November/December. This was due to the stronger currents in the SJC to the South of the Java Island and in the Indonesian channels during this time of the year. Similar current speeds were recorded by drifter 4400657 in the Aru Islands in the Arafura seas. Two drifters (4401842, 44001997) experienced speeds > 1 ms⁻¹ in April and May in the open ocean. Overall, all drifters had a mean speed of 0.37 ms⁻¹ over the 250-day deployment (Figure 3c).

Drifter-derived current speeds were binned to a 0.5° spatial grid, which provided information on the distribution of current speeds in the SETIO over the study period (Figure 4a,b). The spatially averaged currents in the SMZ were mainly > 0.5 ms⁻¹, and were also associated with higher standard deviations (Figure 4a,b). Current speeds were >1 ms⁻¹ to the south of Java Island in the eastward flowing SJC. The northward currents through the Lombok channel and in the Banda Sea (Figure 4a) also showed similar values. The westward flowing SJC and SEC experienced speeds 0.6 ms⁻¹ and 0.4 ms⁻¹, respectively, with low standard deviations (Figure 4a,b).

Immediately after deployment, all the drifters moved in a northwesterly direction and entered the SMZ and dispersed. However, interestingly, the 15 drifters then arranged themselves into groups that travelled together throughout the study period. These groupings included five pairs (Figure 5a–c,e,f) and a group of five drifters (Figure 5d). The first pair of drifters meandered through the SMZ and was entrained into the westward-flowing SJC and then retroflected when the SJC reversed direction (Figure 5a). Then, both drifters moved north, one through Lombok Strait (Figure 1a), and the other through channel between Sumbawa and Komodo islands and travelled east to Aru Islands (Figure 1a and Figure 5a). The second pair were very similar to the first pair in that they were entrained into the SJC after transiting the SMZ but remained to the south of Java Island (Figure 1a and Figure 5b). The third pair was almost within the SMZ, first travelling west and then retroflected to go east but passed through many meanders and eddies (Figure 5c). The next set of drifters consisted of 5 drifters which moved to the west, within the SMZ, they were entrained into a series of large eddies and then travelled to the west with the SEC (Figure 5d). The pair of drifters in Figure 5e were like the group of five drifters but were entrained into a single, large, clockwise eddy and then moved west following the SEC. The final pair initially moved to the west but was then always within SMZ for 4–5 months being entrained into different eddies (Figure 5f).Although all 15 drifters entered the SMZ at almost the same time, their behaviour inside the SMZ was very different. The paths of four drifters (4400657, 4402003, 4400419, 44004200) were relatively linear tracking with some meanders moving to the west and changed direction to east at the end of November and spent ~3 months within the SMZ in the northern section (Figure 5a,b). Drifter paths for 4400423 and 4401668 were like the above four drifters with a 180^o direction change around 1 January 2021 (Figure 5c) and located in the northern section. However, these two drifters remained within the SMZ for ~7 months without exiting and at which time the batteries were exhausted. During this period, drifters 4401842 and 4400423 were entrained into 12 and 3 clockwise mesoscale eddies, respectively, and spent up to 20 days entrained in a mesoscale eddy. The next set of drifters (Figure 5d) tracked to the west through a southerly transit spending ~4 months within the SMZ and were entrained into three clockwise and a single anticlockwise eddy. Drifter 4401997 was entrained into one clockwise and two anticlockwise eddies and was trapped for 450 h (~18 days) in the clockwise eddy (see Figure 6a). The drifter tracks indicated a dominance of clockwise eddies (17 vs. 3) and drifters in the northern section of the SMZ were entrained into more eddies when compared to those in the south.

3.1. Forcing Mechanisms of Surface Currents

Surface currents (in the top 0.6 m of the ocean), as measured by the UWA drifters, were primarily driven by the action of winds on the sea surface. Due to the absence of wind data at the resolution of the sampling rate of the drifters (5 min intervals or ~120 m), a direct comparison between wind stress and surface currents were not possible (see for example, van der Mheen et al. [56]). However, the drifters responded to forcing by tides, inertial, and mesoscale eddies. They were also influenced by topographic features such as straits and headlands. In this section, we use Fourier and wavelet analysis techniques to identify the dominant periods present in selected sections of the UWA drifter database.

Time series of currents from drifter 4401997 indicated that, over the period 14 January 2021 to 24 April 2021 (100 days), the drifter was close to Java Island, east of Christmas Island within SMZ (Figure 6a). Drifter current speeds showed a range of patterns (Figure 6b). Initially, the drifter was entrained into an anticlockwise eddy on 14–23 January 2021 (red line in Figure 6a) with relatively high frequency currents due to tides and also inertial currents with period of 55 h corresponding to the inertial period for this latitude (Figure 6c,d). These inertial currents were most likely generated by two tropical storms that were present in the region: (1) Tropical Cyclone Joshua (13–16 January 2021) and (2) Tropical low 10U (19–21 January 2021). After 13 February 2021, the drifter was entrained into two mesoscale eddies (purple line in Figure 6a) with a period of 450 h (~19 days). Current speeds > 1 ms⁻¹ were recorded in these eddies (Figure 6c). After exiting the mesoscale eddies in the SMZ, on ~10 April 2021 (red line in Figure 6a) the drifters were advected westward by the SJC. Throughout the period of 100 days, the drifters also experienced tidal forcing at semi-diurnal and diurnal frequencies (Figure 6d).

3.1.1. Semidiurnal Tides

The drifters were released ~500 km away from the eastern boundary of the SMZ. The first segment (5 days) of drifter 4401842 was representative of all 15 drifters after their deployment (Figure 7a). A northwestward current transported the drifters ~100 km in 2.5 days, with a mean speed of ~0.5 ms⁻¹ (Figure 7b,c). The drifters experienced maximum speeds (~1.5 ms⁻¹) on 19 September, and from 21 September onwards it reduced to a steady speed of 0.50 ms⁻¹. The time series indicated strong semidiurnal tidal forcing (~12 h), which was reflected in both spectral and wavelet analysis (Figure 7d–h). The wavelet analysis revealed both clockwise and anticlockwise velocity components at 12 h period, with slightly higher energy in the anticlockwise component (Figure 7e,f).

3.1.2. Mixed Forcing: Inertial, Semi-Diurnal, and Diurnal Tides

When the drifters entered the SMZ, they experienced a variety of forcing mechanisms. The meandering segment of drifter 4401842 (same drifter as in Section 3.1.1) over the period 20 October to 9 November 2020 (~21 days) whilst moving west with the SEC, experienced three major forcing mechanisms (Figure 8). The mean speed of the drifter was ~0.30 ms⁻¹ with a maximum of 0.69 ms⁻¹. Both the power spectrum (Figure 8d) and wavelet analysis (Figure 8e) confirmed that this segment was dominated by semidiurnal (12 h) and diurnal (24 h) tides and inertial currents (50 h). The inertial currents are clearly identified in the wavelet analysis, with energy only in the anticlockwise component (Figure 8e). The inertial currents were visible after the 29 October 2020 in the wavelet analysis.

3.1.3. Mesoscale Eddy

Drifter 4401842 was entrained into a clockwise mesoscale eddy within the SMZ. The segment from 2 April to 25 May 2021 showed that it initially moved to the northwest for 2 days and then to southeast and was trapped within an eddy south of Lombok Island and moved to the southwest (Figure 9). During the initial 10 days of this segment (2–12 April), the maximum currents reached 1.4 ms⁻¹ when the drifter was moving along a direct path to the east, i.e., 6–7 and 10–12 April (Figure 9a). The drifter was entrained into a clockwise mesoscale eddy 12–23 April during which time the currents speeds gradually decreased (Figure 9c). The mean rotating period of the eddy was 110 h (Figure 9d).

The drifter exited the eddy on 23 April, and after being entrained into a smaller eddy moved to the southwest into a region that was dominated by semi-diurnal tides from 27 April to 6 May with maximum tidal currents >1.0 ms⁻¹ (Figure 9b,c). Over the period 29 April to 25 May there was steady drift to the southwest in a ‘scallop’ type trajectory indicating possibly trapped in an eddy that was being advected to the southwest. The maximum currents reached >1.0 ms⁻¹ where there was a direct path (around 14 and 21 May).

3.1.4. Strait Flow

A segment from drifter 4400657 is used here to discuss the currents between two islands (Figure 10). The drifter travelled to the northeast after its deployment on 17 September and was entrained into the SJC. When the current reversed due to seasonal change, the drifter travelled parallel to coast to the east (Figure 5a). On 31 December, the drifter was located at the southern entrance to Sape Strait between Sumbawa and Komodo islands (Figure 10). On 17 January 2021, after 19 days, the drifter moved into the Flores Sea (Figure 10a). The time series revealed the strong semi-diurnal tidal currents in and around the channel. Over the period 31 December to 2 January, the drifter made 4 ebb-flood cycles to the south of Banta Island, with maximum northward velocities up to 2.5 ms⁻¹ (Figure 10b,c). This was the highest current speed recorded by all the drifters in this experiment. This was during spring tides. On 2 January, the drifter moved past the western side of the Banta Island and was located to south of Sangeang Island from 3 to 11 January (Figure 10a). During this period, the tidal currents were lower (up to 0.5 ms⁻¹) due to a combination of neap tides and lower tidal influence in the strait (Figure 10c). The mean tidal excursion was ~7 km. The drifter then was transported to the strait between Banta and Komodo Islands (Figure 10a). Spring tides were present with maximum tidal currents up to 2 ms⁻¹ with tidal excursion length of ~16–17 km (Figure 10c). Spectral analysis and wavelet analysis also confirmed that the dominant forcing was through semidiurnal tides (Figure 10d–h).

3.1.5. Headland Flow—Diurnal Tidal Currents

On 17 January 2021, drifter 4400657 left the Komodo Islands and continued eastwards. It reached as far as the Aru islands (~1700 km) on 28 March 2021 after 72 days (Figure 11a). The drifter remained in the northern region of Aru Islands for another 59 days (28 March 2021 to 25 May 2021). For majority of the time, the drifter was located to the northeast of the islands, and in the last 12 days it drifted to the west and to the north of the island, entrained in the headland flows (Figure 11a). The tides in this region were predominantly diurnal (Figure 11d), and the time series of drifter velocity captured four fortnightly tidal cycles (Figure 11b,c). The four fortnightly tidal cycles are clearly evident in the wavelet analysis for both clockwise and anticlockwise components (Figure 11e,f). The spectral and wavelet analysis also show the dominance of diurnal tides (Figure 11d–h). In the diurnal tidal system, the maximum tidal range, and, therefore, tidal currents, occurs during the tropic tides (when the moon is directly about the tropic of Cancer and Capricorn) whereas the minimum tidal range occurs during equatorial tides (when the moon is directly about the equator). The maximum tidal currents during tropic tides were up 1 ms⁻¹, and during equatorial tides they were ~0.25 ms⁻¹. During the last tropic tidal cycle, the drifter was entrained into the headland flow, with maximum tidal currents of 2.25 ms⁻¹ due to the topographic influence of the headland. The tidal excursion was ~13 km when drifter was in the east of the island and increased to 28 km, when the drifter was in the strong headland flow. Interestingly, although the tidal currents off the headland were extraordinarily strong, there was no flow separation on either side of the headland where the drifter movement was located due to the deeper water (~100 m; [89]).

3.1.6. Inertial Currents from a Tropical Storm

Drifter 4401842 (same drifter as described in Section 3.1.1, Section 3.1.2 and Section 3.1.3) continued its journey and here we describe the movement of the drifter over a one-month period from 1 to 30 January 2021. Initially, over the period 1–5 January 2021, the drifter moved southward (Figure 12a), with a relatively low mean speed of 0.25 ms⁻¹ and with the oscillations mainly in the semi-diurnal band (Figure 12b,c). The drifter then changed direction and moved to north-northeast over the period 5–30 January (Figure 12b,c). The change in the drifter direction coincided with the presence of tropical low 06U, which was active in the region during 5–10 January (path shown on inset in Figure 12a), with maximum wind speeds (from ERA5 re-analysis) reaching 10 ms⁻¹ on 7 January. During this period (6–8 January), the mean current speed gradually increased (Figure 12b,c). After the cessation of the tropical low on 10 January, the drifter continued to travel north-northeast with mean speed of 0.25 ms⁻¹. However, the ‘scallop’ type movement in the drifter tracks (Figure 12a) was also associated with large oscillations with a 55 h period, which is also the local inertial period for the latitude band 12–13° S. The inertial currents were superimposed on the mean currents. Maximum current speed associated with the inertial currents increased from 0.55 ms⁻¹ on 10 January to a maximum amplitude of 0.75 ms⁻¹ on 14 January and gradually decreased over time until 30 January after 10 inertial oscillations (Figure 12c). Spectral and wavelet analyses both indicated peaks at 55 h (Figure 12d–h), with wavelet analysis also showing energy only in the anticlockwise component, providing further evidence for the inertial currents (Figure 12e).

4. Discussion

We presented results from an ocean surface drifter deployment undertaken on 17 September 2020 in which 15 identical drifters were released simultaneously (within 5 min) at an offshore location (123.3° E, 13.8° S; Figure 1a) in a water depth of ~200 m. The low-cost drifters were purpose-built, undrogued, measured the top 0.6 m of the ocean, and provided location data and, therefore, the velocities, at 5 min intervals, over a period of 8 months. The mean spatial resolution was 120 m. At the end of the experiment, the 15 drifters were horizontally separated by ~8000 km within the latitude and longitude bands 6° S to 20° S and 83° E to 136° E, respectively (Figure 3a). After deployment, all the drifters remained close to each other as a single cluster for 15 days (2 October) when they crossed the 120° E longitude and moved onto the SMZ, a region reflected by higher EKE to the south of Java Island, bounded by 8–15° S and 102–120° E (Figure 1b). In the SMZ, the drifters were entrained into many meanders and eddies, and their movement was chaotic (Figure 3, Figure 4 and Figure 5). After many months in the SMZ, five drifters were transported to the west through the SEC (Figure 5d), others moved to the northwest through the SJC, which reversed in direction and were moved to southeast (Figure 5a,b). Two drifters that was entrained into SJC moved through two island passages (Lombok and Sape Straits) into the Banda Sea and were transported 2000 km to the east (Figure 5a).

The SMZ is in a region where three major currents converge: ITF, SEC, and SJC. In addition, part of the ITF also inputs to the region through the Lombok Strait (Figure 1a). Similarly, long-term model simulations by Wijeratne et al. [22] have shown a complex recirculation pattern of currents in the SMZ. Previous studies have shown that this region is the only eastern basin among the global oceans where strong eddy activity exists [90], and it is devoid of surface drifter data in global databases [58]. Mesoscale eddies, both clockwise and anticlockwise, are formed in this region and have been attributed to shear/baroclinic instability/topographic interference between the three major currents SEC, ITF, and SJC [30,35,36,37]. Nof et al. [91] named these mesoscale eddies ‘Teddies’. In this study, all 15 drifters were simultaneously released at a single point and after entering the SMZ, where all the drifters remained for several months. When they exited the SMZ, the drifters were entrained into the two main current systems in the region, SEC and SJC (Figure 3, Figure 4 and Figure 5). The drifters that were in the SEC travelled to the west until the batteries were exhausted (Figure 5d). The paths of the drifters through the eddies and meanders in the SMZ before moving west with the SEC is obvious in Figure 5d.

The SJC is driven by the reversing monsoon winds and variation in freshwater influx from the Indonesian Archipelago [29,33]. From November to June, SJC flows eastward along the southern coast of Java; then flows to the west from July to October. Quadfasel and Cresswell [29] using satellite tracked drifters drogued at 15 m measured a maximum current speed of 1.5 ms⁻¹ close to the coast of Java between December and March. Results from this study agree with these findings: the strongest currents measured from the drifters were along the shoreline of Java (Figure 3b and Figure 4a) in November/December, with daily mean currents up to 1.5 ms⁻¹ (Figure 3c).

In general, the major forcing of the surface currents is the direct action of wind on the sea surface. Many studies have shown that the wind drift current is ~3% of the wind speed, 10 m above the sea surface [56,80,81]. Direct impact of the wind on the wind drift currents is beyond the scope of this study mainly due to the large temporal (~8 months) and spatial (separated by ~8000 km) dimensions of the drifter movement. However, we have already noted the effect of monsoon winds, which reversed the drifter paths within the SJC. In addition to the direct wind effects, we identified that the drifter movement was subjected to forcing of astronomical tides (diurnal and semi-diurnal), inertial currents, submesoscale and mesoscale eddies, and large-scale ocean currents. They were also influenced by topographic features such as straits between islands and headland flows. Often a combination of these forcings impacted on drifter movement at any given time. As the drifters moved through different regions, the dominant forcing also changed both spatially and temporally. This is highlighted by the track of drifter 4401842, where the initial forcing was dominated by semi-diurnal tides (Figure 7), which transited to diurnal tides and inertial currents (Figure 8) and then entrained into a mesoscale eddy (Figure 9) over the eight-month period.

The measured currents using the drifters across the study region were relatively strong, with daily mean currents ~0.37 ms⁻¹ over the 8-month period, with maximum of 1.40 ms⁻¹ recorded in the SJC (Figure 3c). Maximum currents (1.40 ms⁻¹) in the open ocean were those associated with the SJC flowing parallel to the coastline of Java (Figure 3 and Figure 4a). However, the strongest currents >2 ms⁻¹ (5 min averaged) were recorded in the two regions where the currents were constrained by topography: semi-diurnal tidal currents within Sape Strait (Figure 10) and diurnal tidal currents north of the Aru Islands (Figure 11).

Since the advent of ‘modern’ oceanographic measurements initiated by the H.M.S. Challenger voyage scientists have been using the Lagrangian measurements of currents using surface drifters to define coastal and ocean circulation. Modern oceanographic measurements using Eulerian techniques and remote sensing approaches are restricted in both temporal and spatial coverage to capture small-scale dynamics over extended time periods. For this reason, Lagrangian measurements using drifters have provided new insights and will continue to do so into the complex ocean dynamics in estuarine, nearshore, coastal, and ocean dynamics [49,50,51,52,53,54,55]. This has been greatly supported by the global drifter programme [92,93].

5. Conclusions

The 15-drifter experiment reported here, for the first tome from the SETIO region, using low-cost surface drifters, sampling at 5 min intervals with a 120 m spatial resolution, revealed a highly complex ocean environment. Forcing by many different processes (tides, inertial, eddies and ocean currents) was evident, but more importantly, the drifters that were released simultaneously at a single location ended ~8000 km apart after 8 months. The drifters upon entering the SMZ were trapped for almost 5 months and when exiting were transported in different directions by the SEC (west), SJC (northwest and then southwest), and two drifters travelled through the Straits and to the east. This spread could only be achieved using Lagrangian drifters.

These results show the utility of high temporal resolution Lagrangian drifter experiments, which provide a completely different viewpoint of the surface ocean dynamics, that are not possible through other measurement approaches and numerical simulations. This has many implications for studies of connectivity and buoyant material transport (e.g., plastics, buoyant debris), as the numerical models do not have the ability to resolve many of the fine-scale physical processes that contribute to the surface transport in the ocean.