Ocean Turbulence and Mixing Around Sri Lanka and in Adjacent Waters of the Northern Bay of Bengal BAY OF BENGAL: FROM MONSOONS TO MIXING Oceanography

USAGE Permission is granted to copy this article for use in teaching and research. Republication, systematic reproduction, or collective redistribution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: info@tos.org or The Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA. Oceanography THE OFFICIAL MAGAZINE OF THE OCEANOGRAPHY SOCIETY

fresher surface water.These eddies degenerate into smaller (submesoscale) features, forming sharp salinity fronts and filaments (Sengupta et al., 2016).Lateral mixing associated with submesoscale dynamics affects sea surface temperature (SST) and sea surface salinity (SSS), and hence the overlying convection and propagating disturbances in the atmosphere.Small-scale mixing is central to turbulence, which can be generated by wind stress, shear, collapsing fronts, baroclinic instabilities, convective overturning, and internal wave breaking.Integrative understanding of BoB dynamics from large to dissipative scales is a challenging puzzle.The smaller the scales of the processes, the more difficult they are to sample, observe, and interpret.Until now, no systematic measurements of turbulence in the northern Indian Ocean existed, and as a part of the US Air-Sea Interactions Regional Initiative (ASIRI), we had an opportunity to obtain turbulence measurements in the BoB and in the coastal waters of Sri Lanka.This paper describes the findings of this study.
During the boreal winter, the East India Coastal Current (EICC; Figure 1) develops under the influence of the northeastern monsoon and flows southward along the eastern and southern coasts of Sri Lanka (Shetye, 1993;Wijesekera et al., 2015).During the summer monsoon, southwestern winds drive the Summer Monsoon Current (SMC; Figure 1), which is directed along the southern coast of Sri Lanka into the BoB.During transition periods (April-May and September-October), the EICC reverses or weakens along the Sri Lanka coast, and the SMC gradually reverses westward, becoming the winter monsoon current (WMC; de Vos et al., 2014).These current systems, and their interactions with nearby waters, initiate a rich variety of turbulence-generating processes, which cascade energy down to dissipation scales (Lee et al., 2016, in this issue).The convergence/divergence of currents, as well as their reversals during monsoon transitions, can be expected to produce remarkable spatial variability of mesoscale and small-scale phenomena near the eastern and southern coasts of Sri Lanka (Shetye et al., 1996;Mukherjee et al., 2014).However, despite their importance, very little information has been obtained about small-scale processes in the BoB (Kunze et al., 2006), and no direct measurements of small-scale turbulence there and in adjacent waters have been captured until now.Dissipation rates and vertical diffusivities inferred from Argo float strain profiles in the 250-500 m depth range indicate that the central and western parts of the BoB have anomalously low turbulence (Whalen et al., 2012).
The first direct microstructure data and associated hydrodynamic variables (stratification, currents) were collected in 2013-2015 in the BoB and to the south of Sri Lanka (Figure 1) during research cruises conducted onboard R/V Roger Revelle and R/V Samuddrika (a regional research vessel of the National Aquatic Resources Research and Development Agency of Sri Lanka).At many of the sites in 2013-2014, the duration of microstructure measurements was limited to no more than several hours at a time, so it was not possible to deduce details of turbulence generation and dissipation there; longer time series were obtained during 2015.Here, we present snapshots of the turbulent structure in a handful of

INTRODUCTION
The variability of key atmospheric and oceanic processes in the Bay of Bengal (BoB) over a range of scales-from monsoons to mixing-significantly affects regional (Indian Ocean) and global weather and climate.As a result of heavy rainfalls and enormous river discharge into the northern BoB, mainly during the summer monsoons, a very sharp density interface known as a barrier layer is formed between low-salinity surface waters and more saline deep-ocean waters (Girishkumar et al., 2011).The barrier layer is a very distinct feature of BoB stratification, and small-scale mixing across this layer is crucial for heat, mass, momentum, and biogeochemical fluxes in the bay (Venyachandran et al., 2013;Akhil et al., 2014).
Large-scale currents along the BoB boundaries, as well as eddies propagating toward the center of the bay, transport the ABSTRACT.As a part of the US Air-Sea Interactions Regional Initiative, the first extensive set of turbulent kinetic energy dissipation rate (ε) measurements from microstructure profilers were obtained in the Bay of Bengal (BoB) and around Sri Lanka during 2013-2015.The observations span almost 1,200 km meridionally, and capture the dynamics associated with a variety of mesoscale and submesoscale features.High freshwater input in the northern part of the basin leads to regions of intense near-surface stratification, which become weaker moving south.The thin layers trap mechanical energy input from the atmosphere, often confining turbulence to the surface boundary layer.These thin layers can form shallow fronts, which at times resemble turbulent gravity currents (Sarkar et al., 2016, in this issue), and are associated with high levels of mixing.Away from the local frontal zones, turbulence in the surface low-salinity layer appears to be decoupled from the underlying pycnocline, where turbulence occurs only in rare and sporadic breaking events.A striking feature common to all of the data acquired is a dearth of turbulent mixing at depth, a condition that appears to be pervasive throughout the basin except during the passage of tropical storms.It is likely that the strong near-surface stratification effectively isolates the deeper water column from mechanical penetration of atmospheric energy.obtain precise estimates of temperature, salinity, and potential density.The data processing followed the methodology of Roget et al. (2006); additional information can be found in Liu et al. (2009) and Lozovatsky et al. (2015).Data above 5-10 m have been discarded due to the potential for ship-wake contamination.

STRATIFICATION AND TURBULENCE IN THE NORTHERN BAY OF BENGAL Surface Layer Turbulence and a Weak Interfacial Mixing
In this section, we compare two series of microstructure measurements collected in the same location in the BoB on November 18 and 19 around noon local time.During the 23-hour time period of VMP measurements, the wind speed periodically changed from W a ~ 10 m s -1 down to ~4-8 m s -1 and up to ~14-16 m s -1 .Four periods of wind increase were registered, each lasting approximately two hours.The air temperature fluctuated between 27.5°C and 24°C.
Under moderate winds (November 18), a shallow (z <10-15 m) low-salinity (32.2-32.6 psu) surface mixed layer (Figure 2a) was effectively decoupled from the water below by a sharp thermohalocline, where N 2 > (6-8) × 10 -4 s -2 (Figure 2c).Relatively small but distinguishable horizontal variability of T/S and specific potential density σ θ = (ρ θ − 1,000) in the mixed surface layer (Figure 2a,c) indicate that the measurements were in an area of a weak local frontal zone.On November 19, after several relatively short periods of higher winds, the mixed layer deepened only slightly, still decoupled from the waters below by an even stronger barrier layer where N 2 > 2 × 10 -3 s -2 (Figure 2b,d).Wind mixing, however, erased the  Under mild winds, the turbulence intensity, characterized here by ε, gradually decreased from ε ~ (3 × 10 -5 −10 -3 ) W kg -1 at z = 5 m to ε ~ (10 -6 -10 -8 ) W kg -1 between z = 10 m and z = 15 m.Thereafter, a sharp drop to ε ~ 10 -9 W kg -1 was followed by an approximately constant value of ε(z) with increasing depth.The horizontal differences of T, S, and σ θ in the middle of the surface layer (z = 7 m) were Δ x T ≈ 0.25°C, Δ x S ≈ 0.4 psu, and Δ x σ θ ≈ 0.22, respectively, over a ~2 km separation (Figure 2a,c).An increase of the dissipation rate to ε ≈ (10 -6 -10 -7 ) W kg -1 across the entire mixing layer at the second mini-transect most likely is associated with periodic, but short-lived, segments of wind stress intensification up to ~0.5 N m -2 .Thus, simple formulae for mixed layer deepening based on a constant friction velocity at the sea surface (e.g., Pollard et al., 1972; see the review by Zilitinkevich et al., 2007) is expected to fail in predicting the observed changes of MLD from 10-15 m on November 18 to 22-25 m on November 19.
The enhanced turbulence in the surface layer may also break down spatial gradients by lateral stirring, as the horizontal thermohaline differences along the second mini-section reduced to Δ x T ≈ 0.017°C, Δ x S ≈ 0.02 psu, and Δ x σ θ ≈ 0.008 over approximately the same distance as they did in the first section.On the other hand, the observed increase of the mixed layer depth to z ≈ 22 m could be associated not only with localized wind-induced mixing but also with lateral advection of a deeper mixed layer to the measurement site.
The observation that the surface layer and pycnocline were effectively decoupled from each other suggests that internal sources of turbulence in the interior (z ≈ 25-120 m) were weak (ε <10 -8 W kg -1 ; Figure 2a,b) under mild and even relatively strong sporadic winds observed before and on November 18 and 19.Internal wave radiation and breaking of the waves below the barrier layer appear to be damped, possibly due to local wave breaking in the pycnocline (where a slight increase of ε could be seen).

Upper-Ocean Response to a Moderate Storm
A 48-hour time series of almost 700 vertical profiles at 18°N, 89.5°E provides an opportunity to assess the BoB's upperocean turbulent response to a moderate wind event (Figure 3).Initially, winds were light (5 m s -1 ) and the near-surface boundary layer was capped by a cooler, low-salinity layer about 5 m thick and delineated from the fluid below by stratification associated with both salinity and temperature steps.As the winds picked up to 10 m s -1 , the surface boundary layer gradually deepened, but remained less than 10 m thick.Mixing at the base of the surface boundary layer reached 10 -6 W kg -1 , but such directly forced turbulence never penetrated deeper than 15 m, even during the peak of the wind forcing.Below this depth, significant mixing events were weaker (10 -8 -10 -7 W kg -1 ) and relatively infrequent, presumably because the strong near-surface stratification limited downward energy transfer (see also MacKinnon et al., 2016, in this issue, for a discussion).
During this period, three notable patches of enhanced turbulence can be identified beneath the surface layer.One is a region of elevated mixing that occurred at 06:00 on September 13 at 20 m depth as the surface mixed layer deepened but was clearly distinct and separated from the surface mixed layer by a layer of weaker turbulence.The other two patches occurred around 12:00 on September 14 following an increase in wind speed at  -10 12:00 12:00 00:00 00:00 18:00 18:00 06:00 06:00 12:00 September 12-14, 2015

Weak turbulent response
Time series at 18°4'N, 89°27'E about 09:00; any connection between this wind event and turbulence is speculative.However, neither of these patches significantly exceed 10 -8 W kg -1 , such that the time-averaged deep dissipation is 10 -9 W kg -1 (Figure 1e), which is quite weak for upper-ocean turbulence.In addition to these localized events, there is also a notable band of slightly enhanced turbulence at 30 m depth that is persistent over the entire record and appears in conjunction with a thermal inversion there.

Surface-Layer Turbulence Affected by Strong Salinity Fronts
Frontal zones between saltier oceanic waters and lenses or filaments of fresher waters of riverine and/or rainfall origin strongly influence turbulence and mixing in the upper layer of the northern BoB.Horizontal dimensions of the low-saline features may vary from a few to tens and hundreds of kilometers.Many frontal zones were detected in the northern BoB during R/V Roger Revelle ASIRI cruises (Lucas et al., 2014;Wijesekera et al., in press).Turbulence measurements across one such front were made on November 21; this is the same front shown in Figure B1c in Box 1 in Sarkar et al. (2016, in this issue).Two minisections (550 m and 720 m long, each with six and eight approximately equally distributed casts) were obtained while crossing approximately perpendicular to the front from saltier warmer ambient waters to the fresher colder side.The origin of this specific pool of low-salinity water is most likely an eddy or filament associated with southbound low-saline flow to the east (within the Sri Lankan Exclusive Economic Zone not sampled at that time).A very sharp salinity change (Δ x S ~ 0.71 psu) was observed at the first crossing over a distance of ~100 m (Figure 4a); the temperature and density at the fresher side of the front (not shown here) decreased by Δ x T ~ 0.52°C and Δρ x ~ 0.33 kg m -3 , respectively.Over the course of 10 hours, R/V Roger Revelle crossed the front 10 times while towing a thermistor chain from its bow.These data permit us to determine that the front was very sharp at times, dropping by as much as 0.5 psu in less than 5 m horizontal distance.These data also permit us to determine that the feature's propagation speed (0.15-0.2 m s -1 relative to the fluid ahead of it) was close to √ g'H, consistent with internal gravity current speed (Turner, 1973).Here, g' = gΔρ/ρ = (3.2-3.5)× 10 -3 m s -2 is reduced gravity and H = 10 m is the thickness of a colder, low-salinity layer.From the VMP-500 data, the feature appears turbulent, especially closer to the sea surface (compare the regions delimited by dashed lines in the upper and lower panels of Figure 4).Detailed inspection of T, S, and ρ θ profiles closest to the front revealed many density inversions in the upper 10 m layer (not shown here), pointing to active turbulent mixing by the frontal flow.Below the low-salinity layer, turbulence was sharply reduced, with ε < 10 -8 W kg -1 , but only in a limited depth range (~25-35 m) shown by blue-green regions in the ε(z,x) contour plots in the two lower panels of (yellow-green strips in the ε(z,x) plots) due to persistently high vertical shear.The squared shear was close or exceeded Sh 2 ~ 10 -4 s -2 , ensuring a Richardson number of the order 0.1 for N 2 ~ 10 -5 s -2 .Below the MLD, stratification increased in the pycnocline with N 2 ~ 3 × 10 -4 s -2 on average, while the mean shear substantially and continuously decreased to less than Sh 2 ~ 2 × 10 -6 s -2 , preventing shear-induced instability.As a result, the dissipation rate in the pycnocline drops below 10 -10 W kg -1 (Figure 4).In addition to these 2013 observations associated with the propagation of a gravity current, regions of strong horizontal gradients were also sampled at two separate locations in 2015 and are summarized in Figure 1f,g.In both cases, turbulence profiling spanned a region of strong horizontal velocity gradients that were associated with ageostrophic fronts, some of which were as sharp as those observed in 2013.While the strongest turbulence was confined to the surface boundary layer, in both of these cases, dissipation rates below the near surface were also enhanced, but not to the same extent as the 2013 event (compare Figure 1 panels f and g with d); dissipation rates below 40 m depth remained close to background levels.

STRATIFICATION, CURRENTS, AND TURBULENCE TO THE SOUTH AND EAST OF SRI LANKA IN THE SMC AND EICC
The VMP measurements to the south and east of Sri Lanka along the WS and TS lines (Figure 1) revealed substantial differences in stratification and turbulence between the EICC and SMC branches of the near-coastal currents that reflect seasonal and/or spatial variability.The TKE dissipation rate profiles along the WS and TS (color strips of log 10 ε) are shown in Figure 5a,b overlaying the contour plots of potential density (the corresponding temperature and salinity panels are presented in Wijesekera et al., in press).During the intermonsoon season (April; Figure 5a), the depth of the relatively well-mixed surface layer in the blue water (>22 km from the coast) appears to be about 30-40 m across the entire southern branch of the current, deepening slightly toward the open sea.According to the VMP measurements made in February 2014 at two WS stations near the shelf break (not shown), the depth of the surface homogeneous layer was ~60 m, indicating the possibility of substantial convective cooling and/or strong wind mixing in the upper layer south of Sri Lanka during the winter monsoon, and gradually relaxing forcing toward the transition period.Lenses of slightly fresher and cooler water near the sea surface did not affect the near-surface density structure very much, as evidenced by  the almost uniformly mixed upper layer and the sharp, but not very narrow, pycnocline.In general, the pycnocline deepens toward the south, suggesting predominantly eastward geostrophic transport at the end of April.This is well supported by the contour plot of geostrophic velocity (Figure 6a) calculated using deepwater Sea-Bird CTD profiles with a zero velocity reference level at 550 db.An interesting dynamical feature appears in Figure 6a closer to the Sri Lankan coast, where the eastward-directed along-slope current flows next to the westwarddirected offshore current (dark blue contours).Although it is possible that the observed westward current is a remnant of the seasonal WMC during the transition season (April) between winter and summer monsoons, it is even more likely that we observed a well-developed (down to ~320 m) clockwise-rotating anticyclonic mesoscale eddy approximately 50 km in diameter with a core (~zero velocity) located about 45-48 km from the coast.Note that the regular SMC usually does not extend below ~200 m depth (Jensen, 2001).Sea surface elevation maps retrieved from the Aviso archive (http://eddy.colorado.edu/ccar/ssh/nrt_global_grid_viewer) show a mesoscale feature to the south of Sri Lanka, with a positive height (up to 6 cm max), which could be associated with an anticyclonic eddy that separated from westward flow near the Sri Lanka shelf break.
In September, toward the end of the summer monsoon, the thermohaline structure along the TS to the east of the Sri Lanka coast (Figure 5b) is much more complex than that of the WS.It exhibits a sharp thermohalocline in the depth range z = 30-40 m near the shelf break.However, approximately 80-90 km offshore, the pycnocline bends toward the sea surface, forming a striking baroclinic front that separates the fresher BoB surface water (S <33.8 psu, σ θ <21.4) moving southward along the Sri Lanka coast from the saltier water of Arabian Sea origin (S > ~35.2-35.4psu at z ≈ 40-50 m, corresponding to the specific density range σ θ ~ 22.4-22.5 in Figure 5b) that is moving north-northeastward at the eastern end of the section.The calculation of geostrophic velocities across this section (Figure 6b) and ADCP measurements (Figure 6c) support this notion.Interestingly, in September, the southward-flowing branch of the EICC is present (green strip in the lower panel of Figure 6c), although it is quite narrow, extending from the coast only ~40-45 km offshore.It has two cores, at ~30 m and 90 m depths, with maximum velocity of about 0.4 m s -1 .Simulations by Jensen (2003) indicate that the southward branch is perennial in the northeastern part of Sri Lankan waters.On the contrary, the northward branch of the SMC is still a powerful current, with a maximum longitudinal velocity component v max of ~1.5 m s -1 in the mixed layer at a distance of 90-110 km from the coast (intense red  area in Figure 6c).The zonal component u is much weaker (u max ~ 0.2 m s -1 ) compared to v max ; it is always directed eastward in the upper 75 m (pink segments in the upper panel of Figure 6c), showing oscillations with approximate wavelengths of ~35-40 km.Below 75-80 m depth, the u component, which is still associated with the SMC because of the strong northward flow in the same depth range (v ~ 0.5-0.7 m s -1 ), is, however, directed westward (blue segments in the upper panel of Figure 6c), having approximately the same maximum value as its eastward counterpart.(Comparing Figure 6c and b, it is clear that the core of the SMC is composed of very salty Arabian Sea water [S = 35.3-35.4 psu] and that the EICC water flowing along its east side originates in the BoB and exhibits salinity as low as 33.7-33.8psu.) The main features turbulence in the SMC somewhat resemble the observations in the northern BoB (Figures 2-4), where turbulence was mostly confined to the surface mixed layer, which is detached from interior water by a strong barrier layer.In the BoB pycnocline, however, only a small number of patches with ε exceeding 10 -8 W kg -1 were observed, but there was patchy intermittent turbulence in the pycnocline along the entire WS section above z ~ 75 m (Figure 5a).At distances of 20-40 km from the Sri Lankan coast, the dissipation patches were present at all depths where turbulence is likely to be sustained by the instability of radiative internal waves generated at the shelf break.In situ measurements of internal waves in Sri Lankan waters are currently underway (Wijesekera et al., in press).A more speculative possibility is that advection and regeneration of shear-induced turbulence in the interior water could also be in play.This is similar to the shelfgenerated patches of turbulence as exemplified by Phillips et al. (1986) using laboratory experiments.Note that Lozovatsky et al. (2012) reported topographically induced turbulence at a distance greater than 15 km from the source; this turbulence was advected downstream by strong currents and sustained along the way by shear instability (flow behind Baker Island in the western Pacific).
The spatial structure of the dissipation rate appears to be quite different along the TS (Figure 5b).The turbulent patches with ε >10 -8 W kg -1 occupy the EICC at all depths (up to ~40 km from the coast).East of the EICC, the high turbulence is mostly confined within a very narrow and sharply sloping upper pycnocline, which starts 16 km from the coast at z = 35-40 m and crops out at the sea surface about 80 km offshore (Figure 5b), and in the lower secondary pycnocline at depths of z = 70-80 m.The most probable source of this turbulence is strong shear instability at narrow interfaces (e.g., Strang and Fernando, 2001).Indeed, we detected a layer of strong shear (Figure 5c), Sh >(2-3) × 10 -2 s -1 , that coincided with the secondary pycnocline centered at z = 75 m, ensuring sporadic generation of the high-level of turbulence shown in Figure 5b.

CONCLUSIONS: SPATIAL PATTERNS OF MIXING IN THE BAY OF BENGAL
The first measurements of TKE dissipation rate ε in the northern BoB and adjacent waters around Sri Lanka provide an initial glimpse into mixing in the BoB.From these data, we gather a consistent picture of the patterns of mixing, as summarized by the profiles in Figure 1.Most striking is the general finding that the dissipation rate is extremely weak (~10 -9 W kg -1 ) below ~20-30 m depths, except during rare storm events.We also observe a general trend that the deep dissipation rates are weaker to the north, where surface stratification is strongest, and increases southward, where nearsurface stratification is reduced.These observations lead to the general conclusion that turbulence in the northern BoB is largely influenced by details of the very thin, near-surface layer that is controlled by monsoon rainfall and river inflow and the complex meso-and submesoscale motions that advect it (MacKinnon et al., 2016, in this issue).More specific conclusions of this study are: 1. Very strong stratification in the sharp BoB pycnocline can damp windinduced mixing almost completely, preventing penetration of turbulence below a thin, lower-salinity mixed surface layer (MLD <15-20 m).Under moderate winds (W a ~ 11-12 m s -1 ), the surface layer is effectively decoupled from the thermohalocline, but horizontal/temporal gradients of T and S still exist above the MLD.Under stronger winds (W a ~ 16-18 m s -1 ), the homogeneous mixed layer deepens only slightly (Figure 1c), but is still largely decoupled from the pycnocline.The horizontal/temporal gradients of T, S, and specific potential density in the surface layer almost completely vanish, possibly due to enhanced wind stirring.2. The northern BoB is characterized by shallow and very sharp density fronts, which at times resemble thin gravity currents with strongly turbulent heads.
Turbulence observations across fronts made in both 2013 and 2015 indicate enhanced mixing associated directly with gravity current shear.Below the low-salinity near-surface frontal features, turbulence is enhanced, possibly as a result of strain induced by the overlying dynamics, leading to dissipation rates as high as 5 × 10 -8 W kg -1 down to depths of almost 50 m in 2013.In the pycnocline, turbulence is sharply reduced, with ε reduced to less than 10 -9 W kg -1 .3. Substantial deepening of the surface layer south of Sri Lanka during the winter monsoon was detected (the MLD deepens to 60 m in February compared to ~30-40 m in April).The forcing gradually relaxes toward the transition period.4. The spatial structure of the dissipation rate is quite different along meridional and zonal transects made to the south (WS) and to the east (TS) of Sri Lanka that cross the SMC and the EICC.The main features of turbulence in the SMC resemble the observations in the northern BoB, where turbulence is mostly confined to the surface mixed layer, which is detached from waters below it by a strong pycnocline.However, coastal bathymetry influences pycnocline turbulence in the northern part of the WS, where turbulent patches may be related to internal waves generated at the shelf break. 5.In contrast to turbulence in the SMC, high-level dissipation that occurs along the TS (east of the EICC) is mostly confined within a very narrow and sharply sloping upper pycnocline, which starts at z = 35-40 m near the coast and crops out at the sea surface about 80 km offshore.The most probable source of such turbulence is strong shear instability at narrow interfaces, which is documented by ADCP data in the lower secondary pycnocline at the depths z = 70-80 m.

FACING
PAGE. (inset) Oceanographers from Sri Lanka's National Aquatic Resources Research and Development Agency working aboard R/V Samuddrika launch a vertical microstructure profiler for turbulence and stratification measurements in the Bay of Bengal.Photo credit: B.M.D.H. Kumarasiri.(background) A US scientist deploying a smaller version of the same instrument off the back of R/V Roger Revelle, using a fishing reel and rod instead of the A-frame.Photo credit: Gualtiero Spiro Jaegerlog 10 ε (W kg -1 ) log 10 ε (W kg -1 ) log 10 ε (W kg -1 ) log 10 ε (W kg -1 ) log 10 ε (W kg -1 ) log 10 ε (W kg -1 ) log 10 ε (W kg -1

FIGURE 1 .
FIGURE 1. Summary of vertical microstructure profiler (VMP) measurements in the northern Bay of Bengal, with a sea surface temperature composite from December 14, 2013, shown in color.Counterclockwise from bottom left are average vertical profiles of dissipation (red shading) and salinity (solid line) from: (a) Weligama section (WS) from R/V Samuddrika (April 24, 2014), (b-d) 16°N for November 23, 16°N for November 19, and bore sections from R/V Roger Revelle (November21, 2013), and (e-g) 18°N, "front" and "jet" sections from R/V Roger Revelle(August-September 2015).Not shown are the Trincomalee section (TS) data, which were not averaged because of substantial spatial variability.

FIGURE 2 .
FIGURE 2. Salinity (upper) and the dissipation rate (lower) contour plots along two mini-sections taken in the Bay of Bengal on November 18 (a) and 19 (b) and the corresponding vertical profiles of T(z), S(z), N 2 (z), σ θ (z), and ε(z) in the upper 30 m layer (c) and (d), respectively.
horizontal thermohaline and density gradients in the surface layer almost completely, as Figure2c,d, shows in detailed profiles of T(z), S(z), and σ θ (z) and ε for the upper 30 m.

FIGURE 3 .
FIGURE 3. Upper-ocean response to a moderate strength wind event in the northern Bay of Bengal as captured by 699 vertical profiles spanning a 48-hour period near 18°N.(a) Wind stress from the Woods Hole Oceanographic Institution mooring at 18°N, (b) turbulent kinetic energy (TKE) dissipation rate, (c) salinity, and (d) temperature.Beneath the surface boundary layer, only a few patches of turbulence have dissipation rates approaching 10 -8 W kg -1 (circled in red); the time-averaged dissipation below 20 m is less than 10 -9 W kg -1 , as summarized in Figure 1e.

Figure 4 .FIGURE 4 .
FIGURE 4. Salinity (upper) and TKE dissipation rate (lower) contour plots along two frontal cross sections with six (left) and eight (right) VMP profiles, respectively, taken on November 21, 2013.The heavy dashed lines are approximate lower boundaries of the low-salinity lenses.The ε(z, x) plot on the left is overlaid by u(z) and v(z) current components measured in the front at the first (circles) and second (diamonds) sections.Examples of the mean squared shear profiles Sh -2 (z) are in the lower right panel.Distance (m)

FIGURE 5 .
FIGURE 5.The TKE dissipation rate (ε, colored strips) overlaid on a background of specific potential density σ θ along (a) the WS (80.4°E), and (b) the TS (8°N).The enhanced vertical shear (c) centered along z = 75 m coincides with the layer of high dissipation rate within the secondary pycnocline shown in (b).

FIGURE 6 .
FIGURE 6. Geostrophic velocities (the reference level is 550 db) across the (a) WS and (b) TS.Positive velocities are to the (a) east and (b) north.(c) ADCP velocity components u and v along the TS.