The gliders cover an east–west transect across the regions of the SLD and SMC (Fig. ), providing time series measurements of chlorophyll. Surface layers remained weakly productive during most of the observational period, however, events of enhanced chlorophyll were observed at all of the four glider locations (Fig. a–d) as well as in the CTD data (Fig. e). Surface chlorophyll concentrations from gliders and the CTD are shown in Fig. . During the beginning of the observational period, concurrent occurrence of elevated chlorophyll levels was observed within the SLD and along the path of the SMC, as recorded by SG579, SG534 and SG532. At SG620 (TSE), two events were recorded with relatively weaker magnitudes of chlorophyll compared to the other glider locations. CTD measurements captured the surface chlorophyll events in the region of the SMC on 1–2 July and at TSE on 6–8 July, consistent with the glider observations. The ship and glider were about 10 km apart during most of the observational period at TSE, and hence an exact agreement in chlorophyll time series is not expected.

Within the SLD. Summer chlorophyll blooms in the region of the SLD have been reported earlier using satellite images of ocean colour (). The daily evolution of SLAs and currents from archiving, validation, and interpretation of satellite oceanographic data (AVISO) show the intensification of the SLD during the early part (29 June–3 July) of the BoBBLE field programme (Fig. ). Observations from SG579, which falls right inside the dome, revealed the development of a surface chlorophyll bloom during the same period (Fig. a; 30 June–2 July). Chlorophyll concentration at the surface was ∼0.3 mg m-3 on 30 June, increased to ∼0.7 mg m-3 on 1 July and reduced to ∼0.4 mg m-3 on 2 July (Fig. ). CTD observations were available within the dome on 28–29 June, before the ship started moving eastwards from TSW. Until 29 June, surface chlorophyll values were much lower (<0.1 mg m-3), with higher concentrations mostly confined to a depth of about 30–60 m (Fig. e). Hence, it can be inferred that the surface chlorophyll bloom within the dome probably commenced on 30 June, peaked on 1 July and started decaying on 2 July. There were no glider observations of chlorophyll before 30 June to corroborate the CTD data.

The observed increase in surface chlorophyll at SG579 coincided with the intensification of the SLD, characterized by negative SLAs embedded within the cyclonic circulation to the east of Sri Lanka (Fig. ). Time series of minimum SLAs in the region of the dome show that the SLD attained its peak by the end of June (Fig. a). Sea level anomalies decreased to about -0.3 m on 30 June. The thermocline was shallow, located at a depth of about 70 m, during the peak phase of the surface chlorophyll bloom (1 July; Fig. ). The doming of the thermocline indicates dynamical uplifting of the nutricline and enhanced nutrient concentrations in the euphotic zone (; ). The chlorophyll bloom event was characterized by lower surface temperatures (28.6 ∘C) and higher surface salinities (33.95 psu) with upsloping isotherms and isohalines (not shown), compared to the period when the surface chlorophyll concentrations were weak. The decay of surface chlorophyll bloom after 2 July (Fig. ) followed the weakening of the dome (Fig. a). Surface temperature increased by 0.7 ∘C, and surface salinity decreased by ∼1.5 psu on 3 July. The weakening of the dome indicates reduced upwelling of the subsurface nutrients. Nutrient limitation restrains the growth of phytoplankton, leading to the decay of surface blooms, when the biological loss terms dominate. CTD observations within the dome until 29 June, when the ship was at TSW, show that the subsurface chlorophyll concentrations were weak (<0.5 mg m-3) just before the surface chlorophyll event (Fig. e). This indicates that the surface chlorophyll bloom is probably not a result of the vertical redistribution of subsurface phytoplankton. On the other hand, the vertical transport of subsurface nutrients to the near-surface layers can favour the growth of phytoplankton in the given timescales (), leading to the intensification of surface chlorophyll. Though the evolution of observed chlorophyll follows the dynamics of the study region, the concurrent role of biological loss terms, including grazing, mortality and sinking rates, cannot be ignored, which requires additional data sampling.

The southern BoB was characterized by cyclonic wind stress curl, inducing Ekman suction during the field programme. The vertical transport of nutrients to the surface sunlit layers through Ekman suction favours the generation of phytoplankton blooms (; ). Spatial distribution of Ekman vertical velocities, calculated using ASCAT winds and averaged for the BoBBLE observational period (24 June–23 July), indicates widespread upwelling in the southern BoB (Fig. b). Time series of Ekman vertical velocities in the location of SG579 show that Ekman suction peaked to about 2–3 m day-1 by mid-June (Fig. a). Ekman vertical velocities remained to be favourable for upwelling (0.4–0.7 m day-1) during the period of surface bloom (30 June–2 July), though the magnitudes were relatively weaker. Strong upwelling in the second half of June, prior to the surface chlorophyll event, is presumed to provide a favourable preconditioning by lifting the nitracline towards the surface.

During the decaying phase of the SLD in July, Ekman vertical velocities were positive, with peak values of about 2 m day-1 (Fig. a). This indicates the dominant influence of remote effects propagating from the eastern boundary of the BoB (; ; ; ; ). A time–longitude Hovmöller diagram of SLAs from AVISO during May–July along 8∘ N, between 80 and 100∘ E, is shown in Fig. c. The decay period of the SLD coincides with the arrival of positive SLAs from east, representing the westward propagation of downwelling Rossby waves (). Rossby waves propagating from the eastern boundary of the BoB can influence the depth of thermocline (nitracline) in the study region. This shows that, despite the Ekman suction, remote forcings contributed to the weakening of the SLD and hence the chlorophyll distribution. As far as surface chlorophyll is concerned, the proximity of nutricline to the surface is of primary concern. Results from the ecosystem model have been used to identify the dominant forcings controlling the vertical displacement of nitracline (see Sect. 3.3.1).

Along the path of the SMC. Increased surface chlorophyll levels were observed at SG534 and SG532 on 1–2 July (Fig. b) and 2–4 July (Fig. c) respectively. Both gliders were located along the path of the SMC, with SG532 in the region of the subsurface high salinity core (Fig. b). Surface chlorophyll concentration peaked to about 0.35 and 0.4 mg m-3 at SG534 and SG532 respectively (Fig. ). This increase in chlorophyll was associated with lower temperatures (28.7 and 29.1 ∘C for SG532 and SG534 respectively) and higher salinities (34.4 and 34 psu for SG532 and SG534 respectively) at the surface, compared to the period when the chlorophyll levels were weak.

Along the path of the SMC, when the surface chlorophyll levels were high, the thermocline was deep (∼100–130 m at SG534 and ∼160–180 m at SG532), which is 40–100 m deeper than that in the region of the dome (Fig. a–c). The spatial variability of thermocline is evident from the CTD observations as well, showing a shallow thermocline during the beginning (27–30 June) and end (20–21 July) of the field programme, when the ship was in the west, and a deeper thermocline farther east (2–18 July; Fig. e). A deeper thermocline generally indicates a deeper nitracline and stronger nutrient limitation in the surface layers. At the same time, the region of the SMC is also subject to an additional supply of biologically rich waters advected from the coasts of India and Sri Lanka (). In addition, the possibility of lateral advection of nutrients and chlorophyll generated within the SLD to the nearby glider locations cannot be ignored (see Sect. 3.3.2).

Mixing events. Chlorophyll distribution observed outside the dome, farther east at TSE, differed from that in the region of the SLD and SMC, in terms of intensity as well as the vertical structure. SG620 captured two events of enhanced surface chlorophyll: the first on 3 July and the second on 6–8 July (Fig. d). The surface chlorophyll concentrations were ∼0.3 mg m-3 during both the events (Fig. ), characterized by low surface temperatures and high surface salinities. The observed SST from SG620 was about 28.7 ∘C on 3 July and 28.8 ∘C on 6 July. Surface salinities were about 34.5 psu and 34.7 psu on 3 July and 6 July respectively. Temporal coverage of the first event is insufficient in explaining its evolution, since the chlorophyll bloom decayed immediately after 3 July, when the sampling began. Wind speed measured by the shipboard automatic weather station (AWS) was 5–9 m s-1 on 3 July (Fig. ). A deeper mixed layer depth (MLD) of about 60 m during this period indicates that vertical mixing is presumably the primary factor which favoured the increase in surface chlorophyll. The second event was captured by the CTD measurements as well (Fig. e), consistent with the glider data. This event coincided with a phase of increasing wind speed of about 6–11 m s-1 (6–7 July; Fig. ). Subsequent deepening of the mixed layer (∼70 m; Fig. d) suggests the role of mixing and entrainment in triggering the intensification of surface chlorophyll. Enhanced vertical processes favour intensification of surface chlorophyll by transporting nutrients to the euphotic zone and by redistributing the subsurface chlorophyll to the surface layers.

The decay period of the observed surface chlorophyll blooms (Fig. 5) coincided with the development of intermittent freshening events at the surface: the first on 4–5 July and the second on 7–10 July. The initial drop in surface salinity during the freshening events was ∼0.4 psu on 4 and 7 July (Fig. ). The surface chlorophyll decreased by about 0.3 and 0.27 mg m-3 during the first and second freshening events respectively (Fig. ). There was an overall reduction in total chlorophyll integrated over the mixed layer by about 20 mg m-2 during both the freshening events (Fig. ).

The freshening events were characterized by the formation of barrier layers (). Vertical profiles of temperature, salinity and chlorophyll from SG620 during different stages of the surface bloom evolution are shown in Fig. a–e. During the peak of the surface chlorophyll blooms (3 and 6 July), the mixed layer was deep (∼50–55 m), with an almost uniform distribution of biophysical properties (Fig. a and c), and the isothermal layer was close to the mixed layer. The days following the peak in surface chlorophyll (4–5 and 8–10 July) were characterized by strong salinity stratification with the arrival of freshwater in the surface layers. Surface salinity decreased by about 0.25 and 0.75 psu for the first and second events respectively (Fig. b and e). The mixed layer shoaled to ∼30 m, whereas the isothermal layer remained around the same depth (Fig. b, d and e). The associated development of barrier layers is noticeable, with a thickness of ∼25–30 m. Following the salinity stratification and barrier-layer formation, surface chlorophyll decreased by about 0.1 and 0.15 mg m-3 during the first and second events respectively. Vertical profiles obtained from the CTD at TSE for the same period are given in Fig. f–j. With the arrival of freshwater, surface salinity from the CTD decreased by about 0.25 and 0.5 psu during the first and second events respectively, and the corresponding decrease in surface chlorophyll was 0.1 and 0.15 mg m-3 respectively. The mixed layers shoaled by about 25–30 m, creating strong barrier layers (Fig. g, i and j). Even though high wind speed (∼10–12 m s-1) conditions prevailed during the decay period of the bloom, freshwater induced stratification was strong enough to overcome the wind effect (Fig. ). The observed biological response to freshwater is similar to that in the northern bay, where stratification inhibits the development of phytoplankton blooms in the surface layers by restricting the vertical transport of subsurface nutrients and chlorophyll ().

The formation of DCM is determined by a variety of mechanisms, including an enhanced growth rate of phytoplankton co-limited by light and nutrients at optimum depths, photoacclimation of pigment content, and physiologically controlled swimming behaviours and buoyancy regulation (). The BoB is reported to have prominent DCM (; ), which contribute to the column-integrated productivity (; ; ), with magnitudes often comparable to the highly productive Arabian Sea (). However, little is known about the distribution of subsurface chlorophyll in the BoB and the associated processes, due to the lack of observations.

During the BoBBLE field programme, both the glider and CTD observations revealed the presence of prominent DCM in the southern bay (Figs.  and a). The chlorophyll maxima were centred at a depth of about 20–50 m, mostly below the mixed layer and above the thermocline (). Similar depth ranges of DCM were reported previously by and in the BoB. Subsurface chlorophyll concentrations ranged from 0.3 to 1.2 mg m-3 (Fig. a), which were 2–3 times higher than the surface values (Fig. ). DCM were prominent in the region of the SLD and along the path of the SMC (Fig. a–c), whereas outside the dome, the subsurface concentrations were weaker (Fig. d).

Vertical profiles of chlorophyll from the gliders during events of enhanced surface chlorophyll are shown in Fig. . The mean DCM were intense, located at a depth of about 20–30 m, in the region of the SLD and the SMC (Fig. a–c). The DCM became weaker, diffused and slightly deeper (30–40 m) at TSE (Fig. d and e). Intensification of DCM in the region of SLD can be related to the doming of thermocline. The vertical transport of nutrients is affected by the changes in thermocline depth, and hence, the variability of nutricline is found to be largely correlated with the variability of thermocline in the tropical oceans (; ; ). The shoaling of thermocline in the region of the SLD indicates an upward sloping of nutricline, indicating nutrient enrichment in the euphotic zone and enhanced accumulation of phytoplankton. At TSE (SG620), where the thermocline was deeper, mixing often penetrated to deeper layers, pushing the mixed layer towards the DCM (Fig. d and e). This favours the dilution of DCM and a decrease in phytoplankton concentration at the subsurface through mixing with the weakly productive surface layers, leaving a near-homogeneous distribution of chlorophyll within the water column (Fig. d and e).

Subsurface chlorophyll concentrations were noticeably higher in the region of the SMC (Figs. c and c). Maximum intensities were recorded by SG532, with magnitudes ranging from 0.7 to 1.2 mg m-3 on 2–7 July (Fig. a). Column-integrated chlorophyll was also observed to be the highest at SG532 (4 July), with total chlorophyll in the top 100 m reaching as high as 35 mg m-2 (Fig. b), which is comparable to the previously observed values in the BoB (; ; ; ). The region of the SMC is characterized by the advection of upwelled chlorophyll-rich water from the western coast of India and the southern coast of Sri Lanka. An isolated maximum (1.2 mg m-3) in the DCM was recorded by SG579 in the region of SLD in the latter half of the observational period (15 July). However, in the absence of surface blooms, the corresponding column-integrated chlorophyll was lower (28 mg m-2) compared to the region of the SMC.

The core subsurface intrusion of the SMC, below the low salinity surface waters of the southern bay, was located around SG532 during the observational period (; ). The vertical salinity structure reveals a high salinity core at 88∘ E, extending up to a depth of about 180 m, with salinity values as high as 35.8 psu (Fig. b). Arabian Sea water, which is rich in nutrients and chlorophyll sliding through the subsurface layers of the BoB, is presumed to contribute to the intensification of the DCM at SG532, suggesting a key role of SMC intrusion in the biological budget of the southern bay. However, it may be noted that the location of the subsurface high salinity core was much deeper relative to the depth of DCM. Most of the high salinity intrusions at 88∘ E occurred below 80 m, in the deeper layers of the euphotic zone. Dynamics behind the distribution of the DCM in the region of the high salinity core are intricate. Though the effect of lateral advection by the SMC on DCM cannot be ignored, the possible contribution of vertical processes in supplying the subsurface nutrients or chlorophyll needs to be examined in detail.

Subsurface chlorophyll concentrations were observed to intensify for shorter durations following the weakening of surface blooms (Fig. ). Increases in DCM concentrations after the decay of surface blooms were about 0.13, 0.37 and 0.25 mg m-3 at SG534, SG532 and SG620 respectively (Fig. a). In the region of the SLD (SG579), the subsurface chlorophyll concentrations increased to ∼0.7 mg m-3 during the peak phase of the surface bloom (1 July). During the decaying phase of the surface bloom (2–5 July), these high chlorophyll levels (0.7 mg m-3) were maintained at the subsurface and weakened afterwards (Fig. a). This indicates enhanced biological productivity at the subsurface, after the triggering mechanisms inducing the surface blooms have weakened. During the decaying phase of surface blooms, the upper layers of the water column became less turbulent or more stably stratified (Fig. ), inhibiting the vertical transport of nutrients and chlorophyll. For example, the surface bloom event at SG620 weakened in response to the freshening event on 8 July (Fig. b). Consequently, there was an increase in DCM, which lasted for a period of about 2–3 days, from 10 to 12 July (Fig. a). The observed intensification of DCM in the absence of surface chlorophyll can be explained in terms of changes in subsurface irradiance levels. During the decaying phase of the surface bloom, the self-shading effect of surface phytoplankton weakens, enhancing the light availability at the subsurface, which is examined in the following section.

Chlorophyll interactive penetrative radiation was calculated at TSE for the period 4–14 July, following the and scheme as given below:

I(z)=IIR⋅e-kIRz+IRED(z-1)⋅e-k(RED)Δz+IBLUE(z-1)1⋅e-k(BLUE)Δz, where I(z) is the penetrative radiation at each depth level, IIR=I0⋅(0.58) represents the infrared band, IVIS=I0⋅(0.42) represents the visible band and kIR=2.86 m-1 is the light attenuation coefficient for the infrared band. The self-shading effect of phytoplankton is taken into account so that at every vertical level (z), the available visible light is computed as a function of irradiance at the level just above (z-1). Δz is the thickness of each layer between two vertical levels, which is 1 m in the present glider data. Visible light is split into two averaged wavelength bands as given below,

2IRED=IBLUE=IVIS2, where IRED and IBLUE are the irradiances in red and blue–green bands respectively.

The light attenuation coefficients for the two visible bands is calculated as a function of chlorophyll concentration ([Chl]) as follows:

3k(RED)=0.225+0.037⋅[Chl]0.629,4k(BLUE)=0.0232+0.074⋅[Chl]0.674. Surface irradiance (I0) for the above calculations was obtained from shipboard AWS (Fig. a) and chlorophyll from SG620 (Fig. d). In order to exclude the effect of daily variation in surface irradiance, a diurnal composite of radiation (Fig. a) for the period 4–14 July is also used for the calculations. Photosynthetically active radiation (PAR) at each vertical level (z) was estimated using the following expression: 5PAR(z)=Ivis(z)×2.5×10186.023×1023, where Ivis(z) is the penetrative radiation (W m-2) in the visible range calculated using the light model and 2.5×1018 quanta s-1 W-1 is the conversion factor obtained from . The depth of euphotic zone (Zeu) was calculated as the depth at which light reduces to 1 % of the surface PAR value. Considering the fact that phytoplankton sees the absolute light level and not the percentage (), the depth of threshold isolume (Z0.415) is taken as the depth where PAR is 0.415 E m-2 day-1 below which light is insufficient to support photosynthesis (; ). An einstein (E) is a mole of photons, i.e., 6.023×1023 photons.

Estimated penetrative radiation (W m-2) using the observed surface irradiance and the diurnal composite are shown in Fig. b and c respectively. The corresponding depths of euphotic zone and the threshold isolume obtained from the calculated PAR values are overlayed. Nearly 40 %–60 % of the radiation was absorbed in the top 1 m of the water column and 80 %–90 % in the top 30 m. Below the DCM, irradiance levels were substantially weaker (<10 W m-2). During the daylight hours of peak insolation, Z0.415 extended to 70–110 m, with a well-defined diurnal cycle. During days of enhanced surface chlorophyll, Z0.415 and Zeu were shallow. Z0.415 shoaled to a depth of about 70–80 m, and Zeu was about ∼50 m during the chlorophyll bloom event at the surface on 6–7 July. The shoaling of the Z0.415 and Zeu indicates the self-shading effect of surface phytoplankton. Elevated levels of chlorophyll enhance the absorption of radiation in the surface layers (Fig. b). Calculations using the diurnal composite of irradiance also give similar results (Fig. c). Enhanced attenuation of radiation by near-surface phytoplankton reduces the irradiance levels in the deeper layers and strengthens the light limitation on phytoplankton growth in the subsurface. As a result, bloom activity weakens in the subsurface layers, despite the availability of nutrients.

Following the decay of surface blooms owing to nutrient limitation, Z0.415 and Zeu increased due to the penetration of radiation to deeper layers (). Z0.415 and Zeu deepened by about 25 and 10 m respectively on 8 July (Fig. b). Enhanced light availability in the subsurface layers favours the intensification of DCM (Figs. d, e and Fig. ). It should be noted that the DCM may not represent a deep biomass maximum, as photoacclimation (; ; ) leads to changes in carbon-to-chlorophyll ratios. At the base of the euphotic layer, the cellular concentration of chlorophyll will increase as an adaptation to the lower irradiance levels ().

The observed spatial distribution of surface chlorophyll in the BoB is well represented by the model (Fig. b), with prominent chlorophyll blooms in the coastal ocean, northwestern bay and the southern bay (). Chlorophyll concentrations are the highest along the coastal regions, with magnitudes exceeding 1 mg m-3. The northwestern bay is characterized by the seasonal occurrence of upwelling blooms triggered by coastal Ekman suction and advection towards the offshore regions (). The southern bay exhibits isolated patches of chlorophyll in the region of the SLD and along the path of the SMC. Surface chlorophyll concentrations are about 0.6–0.7 and 0.3–0.4 mg m-3 in the region of the SLD and the SMC respectively. The model chlorophyll is generally weaker compared to satellite observations. The bias can be attributed either to the deficiencies in external nutrient inputs in the model or to the overestimation of coastal blooms by satellites in the presence of optically active constituents other than chlorophyll (; ). The presence of DCM is well represented by the model, consistent with the glider and CTD observations. Realistic representation of the chlorophyll distribution indicates that the model is suitable for explaining the underlying mechanisms. It may also be noted that the model parameterizations on different biological controls can lead to biases in the simulated fields and processes with respect to the actual observations. For example, the model includes implicit representation of grazing, and hence loss of phytoplankton though grazing is independent of the zooplankton biomass.

While the major seasonal features of the southern BoB are reproduced by the model, they are often not exactly at the observed locations. For example, the SLD is slightly shifted westward and the meandering of the SMC around Sri Lanka is weaker (Fig. c and d), probably due to the discrepancies in the model wind forcing or the simulated remote forcings. The eastward (northward) extension of surface chlorophyll associated with the SMC is overestimated (underestimated). These inaccuracies can be ignored while examining the large-scale seasonal features but may be significant at mesoscales or smaller scales. Hence, the ecosystem model results are used to explain the biological response to seasonal features including the Sri Lanka Dome and the monsoon current, in comparison with the concurrent observations from gliders (SG579, SG534 and SG532) and the shipboard CTD.

The model SLD develops around 85∘ E, 8∘ N, close to the sampling location of SG579. A longitudinal transect extending from 82 to 92∘ E along 8∘ N is selected to examine the vertical distribution of temperature, salinity, nitrate and chlorophyll on 1 July, during the peak phase of the surface chlorophyll bloom in the region of the SLD (Fig. ). The region is characterized by an intense chlorophyll bloom (∼0.5–0.8 mg m-3) at the surface and a prominent DCM (∼0.5–1.2 mg m-3). The DCM lies below the mixed layer, centred at a depth of about 20–30 m, and is about 10–30 m shallower than the nearby regions (Fig. a). Temperature profiles show upsloping isotherms, providing cooler (27 ∘C) waters to the surface layers (Fig. b). Similarly, the salinity distribution shows increased surface salinity (33.5 psu), with isohalines shoaling to the surface (Fig. c). Doming of the thermocline (D20) is evident between 83 and 87∘ E along the transect (Fig. b). The thermocline rises to a depth of ∼60 m, which is about 80 m shallower than the nearby regions outside the dome. The surface layers were enriched with high nitrate concentrations in excess of 10 µmol kg-1 (Fig. a). By the second week of July, cyclonic circulation in the region of dome weakened and shifted towards the northwest, followed by the weakening of chlorophyll.

The dynamics of the SLD favour biological productivity through the vertical transport of nutrients induced by open-ocean upwelling (; ). The time series of minimum SLAs in the region of the SLD shows that the modelled dome peaked on 28 June (Fig. ), 2 days prior to the observed peak. The simulated chlorophyll bloom intensifies during the peak phase of the dome and decays with the weakening of the dome, consistent with the BoBBLE observations. The developing phase of simulated SLD (14–28 July) was characterized by the shoaling of nitracline (Fig. ). We prefer using the 2 µmol kg-1 nitrate isoline as the nitracline rather than the vertical gradient criterion, since the absolute concentration of nutrients available for phytoplankton uptake is more important for bloom generation than the gradients (). The shoaling rate of the nitracline increased to about 1.0 m day-1 by mid-June and closely followed the Ekman vertical velocities. This shows that the vertical supply of nutrients to the surface layers during the developing phase of the SLD can be largely attributed to Ekman suction. During the peak phase of the SLD, both Ekman suction and nitracline shoaling rates weakened. However, the larger shoaling rates during the preceding week indicate a favourable preconditioning for the generation of chlorophyll blooms during the peak phase of the SLD. Ekman suction gradually increased during the decaying phase of the SLD. The corresponding deepening tendency of the nitracline was not consistent with the positive Ekman vertical velocities, indicating the influence of remote forcings.

Chlorophyll distribution in the region of the SMC is influenced by the horizontal advection of both nutrients and chlorophyll. Simulated surface nitrate shows enhanced concentrations along the path of the SMC, indicating the lateral advection of nutrient-rich waters from the Arabian Sea (Fig. d). Advection of phytoplankton from the upwelling regions off the coasts of India and Sri Lanka could further intensify the chlorophyll concentration (; ). The relative role of mixed layer processes in maintaining the chlorophyll concentrations along the path of the SMC is presented in Sect. 3.3.2.

The model DCM shows large spatial variability in terms of intensity and depth. The DCM is strong in the region of the SLD and along the path of the SMC (Fig. a), consistent with the glider observations (Fig. ). Subsurface chlorophyll concentrations increase to about 1.2 µmol kg-1 within the dome, which is more than twice the concentrations outside the dome. At the same time, the depth of the DCM is minimum in the region of the SLD (Fig. c). The DCM shoals to ∼20 m within the dome and deepens to ∼70 m outside the dome. Productivity is closely correlated with SLAs and the depth of the nitracline and thermocline (). The strongest DCM (Fig. a) coincides with the shallowest nitracline (Fig. d). Ekman suction leads to the upsloping of nitracline, which increases the concentration of limiting nutrients in the euphotic zone. The column-integrated chlorophyll is found to be maximum along the path of the SMC (Fig. b), with magnitudes ranging from 50 to 70 µmol kg-1.

The nutrient budget from the ecosystem model is examined to identify the relative roles of mixed layer processes in controlling the summer chlorophyll distribution in the southern BoB. In the TOPAZ ecosystem model, the growth of phytoplankton is determined by a limiting nutrient in a multinutrient environment. Here, inorganic nitrate (NO3) concentration is used to represent the nutrient budget (Fig. ), since the dominant role of nitrate in controlling the biological productivity of the BoB is well known (). The observed nitrate distribution has been used in previous studies to explain phytoplankton distribution in the BoB (). The present simulation also shows that during the pre-monsoon period, productivity in the southern bay is largely limited by nitrate when mixed layer dynamics were less favourable for the vertical supply of nutrients to the surface sunlit layers. Hence NO3 was preferred over PO4 and Fe for explaining the nutrient distribution. In addition, the chlorophyll concentration in TOPAZ is proportional to the nitrogen in phytoplankton (). Total chlorophyll is calculated as 6Chl=C:N⋅12×106⋅(θSm⋅NSm+θLg⋅NLg+θDi⋅NDi), where C:N is the carbon-to-nitrogen ratio, 12×106 is the molecular mass of carbon in µg mol-1, θ is the chlorophyll-to-carbon ratio (Chl:C) and N is the phytoplankton nitrogen concentration in mol kg-1.

Physical processes controlling the model nutrient distribution include horizontal advection and vertical processes (including vertical advection and mixing). The biological processes include a source term represented by nitrification and sink terms comprising denitrification and uptake by the phytoplankton.

The time rate of change of nitrate is given by 7∂NO3∂t=-∇⋅uNO3+∇K∇NO3+SNO3, where u is the velocity vector from the ocean general circulation model (OGCM), K is the vertical diffusivity and SNO3 represents the biological processes.

Weekly averages of the model nitrate budget terms averaged over the mixed layer from 24 June to 21 July, comprising the BoBBLE observational period, are shown in Fig. . The model MLD is defined as the depth at which the buoyancy difference with respect to the surface is equal to 0.0003 m s-2. Before the onset of the summer monsoon, the upper ocean of the southern BoB maintained oligotrophic conditions, where nutrient levels were weak, inhibiting the growth of phytoplankton (not shown). Mixed layer dynamics associated with the monsoonal forcings play a dominant role in controlling the nutrient distribution of the southern BoB. As the monsoon intensifies, the monsoon current becomes stronger and the cyclonic circulation off the eastern coast of Sri Lanka leads to the development of the Sri Lanka Dome (Fig. ).

The last week of June, coinciding with the beginning phase of the BoBBLE observational period, was characterized by a developing phase of the SLD, with strong open-ocean upwelling. Nitrate concentrations in the mixed layer increased (Fig. a) as a result of enhanced vertical transport (Fig. b). At the same time, these nutrients were transported away from the region of upwelling and redistributed to the nearby regions through horizontal advection (Fig. c). Along the southern tip of India and Sri Lanka, coastal upwelling driven by alongshore winds leads to the intensification of nitrate levels, as evident from the vertical processes (Fig. b). Offshore transport of upwelled nutrients occurs at significant rates, enhancing the nitrate concentrations in regions away from the coast (Fig. c). Within the mixed layer, uptake by the phytoplankton is higher than nitrification so that the sink term exceeds the source term. Hence, biological processes contribute to a reduction in total nitrate, mainly in the coastal ocean and the region of SLD, where phytoplankton concentrations are high (Fig. d).

During the first week of July, nitrate levels in the mixed layer reduced slightly compared to the previous week; this period was characterized by the gradual weakening of the SLD and a reduction in the vertical supply of nutrients (Fig. f), leading to a decline in nitrate levels. Consequently, the associated horizontal transport (Fig. g) to the nearby regions also reduced. The nitrate uptake reduced due to the reduction in phytoplankton concentration, which explains the weaker negative tendencies due to biological processes in the region of the dome (Fig. h). Upwelling along the coasts of India and Sri Lanka (Fig. f) and the offshore advection effects (Fig. g) were still prominent during this period.

During the second week of July, nitrate levels in the mixed layer were generally higher compared to the previous week, especially in the region of the SMC (Fig. i). The SLD slightly regained its strength until 10 July and weakened immediately. The related vertical transport of nutrients intensified (Fig. j), and the upwelled nutrients were distributed to the nearby regions (Fig. k). Though the upwelling was not as strong as that in the preceding peak phase (during the last week of June), vertical supply of nitrate occurred at higher rates (Fig. b and j). As a result of strong upwelling in the preceding peak phase of the SLD, the nitrate isolines became shallower (not shown). This preconditioning probably favoured enhanced vertical supply of nitrate to the surface layers during the second peak phase, though the strength of upwelling was weaker.

The simulated eastward velocities associated with the summer monsoon current off the southern coast of India and Sri Lanka strengthened during the second week of July in relation to increasing wind speeds. Along the path of the SMC, a clear patch of increased nitrate levels was evident (Fig. i), which extended from the southern tip of India up to about 85∘ E. This indicates horizontal advection of coastally upwelled nutrients from the southern coasts of India and Sri Lanka (Fig. k) into the southern BoB by the SMC. Lateral supply of nutrients by the SMC supports phytoplankton accumulation along its path. Increased uptake of nitrate by the phytoplankton further enhanced the negative contribution of biological processes (Fig. l).

During the third week of July, nitrate levels along the path of the SMC decreased (Fig. m). Following a reduction in wind speed, the monsoon current off the southern coast of India weakened, and so did the horizontal transport (Fig. o). Vertical supply of nutrients was maintained in the region of dome (Fig. n). Contribution by biological processes decreased as the nitrate uptake weakened following a reduction in phytoplankton concentration (Fig. p). In summary, the above analyses show that the distribution of nutrients and the biological productivity in the southern BoB is largely dependent on the mixed layer dynamics associated with the summer monsoon, and the relative roles of vertical and horizontal processes vary spatially following the circulation features.