Introduction

The Arabian Sea is one of the most biologically productive regions in the world ocean1. The annual cycle of surface chlorophyll distribution exhibits high values during winter and summer and relatively lower values during the inter-monsoon period1,2,3. During the summer monsoon, coastal upwelling associated with low-level alongshore south-westerly winds enhances near-surface chlorophyll concentration by enriching nutrients at the near-surface layers in the western Arabian Sea. On the other hand, convective mixing due to net surface heat loss from the ocean, due to the intrusion of dry continental air into the northern Arabian Sea, is primarily responsible for the enhancement of chlorophyll during winter1,2,3. In addition, the occasional enhancement of chlorophyll on an intra-seasonal time scale, primarily due to vertical entrainment of nutrient-rich subsurface water to the near-surface layer through strong wind-induced vertical mixing, has been reported by earlier studies4,5.

During the photosynthesis process, the phytoplankton converts dissolved inorganic carbon into organic carbon, and it is transported into the deep sea as the organic matter sinks; hence, it plays a significant role in the global carbon cycle6. In addition, one of the most pronounced oxygen minimum zones in the world ocean exists in the southwestern Arabian Sea at the depth range of 100–900 m, where the dissolved oxygen concentration is less than 22 μM,7,8,9,10.

Both phytoplankton and dissolved oxygen (DO) concentrations are directly linked to the marine habitat and play a vital role in determining it. The phytoplankton is the base of the lowest levels of the ocean food web, and it has a strong direct influence on lower trophic level species such as sardines and a delayed response to the higher trophic level carnivore species such as tuna11,12. In addition, DO concentration is another critical parameter that determines the presence of marine species13,14,15,16,17,18. For example, Vaquer-Sunyer and Duarte 200814 showed that the persistence of DO concentration around 100 µM for five days is close to the upper limit of tolerance level of different marine species. Hence, the shoaling and deepening of the Oxygen Minimum Zone (OMZ) significantly influence the vertical distribution of marine habitats and tropical pelagic fishes16,17,18,19.

Several earlier studies, primarily based on satellite ocean color data, have reported the chlorophyll bloom after the passage of tropical cyclone20,21,22,23,24,25,26,27. Naik et al.200828 reported an enhancement of chlorophyll by 4 mg m−3 and 5 µM nitrate concentration at the near-surface layer from ship-based measurements in the Arabian Sea, before and after the passage of Tropical Cyclone (TC), which resulted in a new production of ~ 1.5 g C m−2 day−1. However, it is quite difficult to obtain continuous measurements with ships, particularly during extreme weather events. Thus, the advent of autonomous profilers such as Argo floats equipped with biogeochemical sensors (BGC-Argo) provides the depth-time response of physical and biogeochemical parameters during such extreme weather events. For instance, based on the measurements from the BGC-Argo float in the Bay of Bengal (BoB), Girishkumar et al. 201929 demonstrated how pre-existing hydrographic structure associated with warm and cold-core eddies impacts the upper ocean biogeochemical parameters responses during the TC Hudhud and TC Vardah in the BoB. Several earlier studies also showed that TC can significantly modify fisheries by modulating physical and biogeochemical parameters30,31,32. Giri et al. 201630 have shown a significant increase in Hilsa fish caught in the northern BoB after TC Phailin.

The Indian National Centre for Ocean Information Services (INCOIS) is mandated to provide Potential Fishing Zone (PFZ) advisories to fishermen on an operational basis. The development of marine ecosystem models would greatly help in extending these species-specific PFZ advisories to the fisherman community. Hence, to address these operational and scientific needs, the Regional Ocean Modeling System (ROMS) version 3.733 coupled entirely online with an ecosystem module, was adopted with a horizontal resolution of 1/4° (~ 25 km, hereafter ROMS). Kunal et al. 201934 documented these model evaluations on seasonal timescales. However, the performance of these models under extreme weather conditions, especially during TC, has not been explored previously. Hence, it is imperative to document, validate, and evaluate the model’s ability to simulate the physical and biogeochemical state in response to TCs with respect to in-situ observation. Such studies help us to understand the Arabian Sea’s marine ecology due to the impact of the TC, which has scientific and societal implications.

The BGC-Argo float (WMO ID 2902202; hereafter referred to as A-Float) is equipped with Conductivity-Temperature-Depth (CTD), chlorophyll-a, optical backscatter, and dissolved oxygen sensors that took measurements very close (~ 22 km) to the TC Ockhi (29 November to 06 December 2017; the life cycle of TC Ockhi is mentioned in the data and method section) track in the eastern Arabian Sea (16.36°N and 69.66°E) (Fig. 1; A-Float location marked as green and red color triangles within white rectangular box). Initially, the A-float was configured to obtain profiles every ten days at predefined depths. The A-float was then triggered to obtain two profiles in a day from 09 December 2017 onwards. Note that TC Ockhi made landfall on 06 December 2017. Though A-Float was close to the TC track, we do not have the measurements from the float during the life cycle of TC Ockhi. However, measurements obtained from A-float before and after the TC, provide a unique opportunity to examine the physical and biogeochemical variability due to the impact of the TC. Moreover, the peak enhancement in chlorophyll and dissolved oxygen that was observed 10–15 days after the passage of TC from in-situ observations is crucial for understanding the marine ecology29,35. Hence, monitoring such evolution and evaluating the model’s ability to capture these responses are essential. Before performing the analysis, the chlorophyll measurements from A-float was validated and corrected with in-situ water sample measurements using the linear regression method (Fig. 2; data and method section). All our analysis in this study was restricted to the A-Float location when A-Float was near the TC Ockhi track (within the box of 68.5°E to 70.4°E longitude and latitude 15.9°N to 16.9°N).

Fig. 1
figure 1

The six-hourly best tracks (black line overlaid on green color) and intensity (open colored circles corresponding to the dates and intensity) of TC Ockhi are marked. The classification of tropical systems by IMD with respect to wind speed is classified as Depression (D) (17–27 knot; kt), Deep Depression (DD) (28–33 kt), Cyclonic Storm (CS) (34–47 kt), Severe Cyclonic Storm (SCS) (48–63 kt), and Very Severe Cyclonic Storm (VSCS) (64–89 kt) and are marked as cyan, blue, green, orange, and purple circles respectively along the track of TCs. The number adjacent to each circle indicates dates of cyclone reached at that location. Red colored triangles inside a white rectangular box indicate the A-Float (WMO ID 2902202) profiling locations before and after TC Ockhi (29 November 2017 to 29 December 2017). The dark green triangle indicates A-Float profiling locations in close proximity during TC Ockhi: 29 November to 06 December 2017. The star symbol represents the location of A-float measurements and the water sample collected from the ship (~ 2 km distance) on 05–06 February 2018 for validation and correction.

Full size image
Fig. 2
figure 2

Validation of chlorophyll profiles measured from water sample collection (thin black line with a circle on top) with (a) portable ship CTD (blue) on 06 February 2018 at 00:20 UTC, (b) A-float (thick black lines) on 05 February 2018 at 18:25 UTC, and (c) both. The dashed blue and dashed black lines indicate a corrected chlorophyll profile, and the continuous thick blue and black line indicates an uncorrected, direct measured profile.

Full size image

The primary objectives of this study are as follows:

  1. (1)

    To document the impact of TC Ockhi on physical and biogeochemical variability in the eastern Arabian Sea using the A-float observations.

  2. (2)

    To evaluate the model’s capability of simulating the physical and biogeochemical state in response to TC Ockhi with respect to A-float observations.

Atmospheric responses to TC Ockhi

When TC Ockhi approached the A-float location, the wind stress from the satellite (CCMP) and model showed sudden enhancement and reached a maximum magnitude of 0.4 N m−2 and 0.3 N m−2, respectively, on 04 December 2017 (Fig. 3). In response to this strong wind forcing, the Ekman pumping velocity also showed upwelling of 9.5 m day−1 and 12.7 m day−1 (04 December 2017) from the satellite and the model, respectively. During this period, the net surface heat flux (NHF) also showed a maximum cooling with magnitude of −195 Wm−2 ( −345 Wm−2) from satellite (model) on 04 December 2017 as a response to the TC Ockhi at the A-Float location (Fig. 3). The magnitude of wind stress and Ekman pumping velocity from both satellite and model showed similar temporal evolution with comparable magnitude during the study period, except on 04 December 2017 (Fig. 3a). However, the magnitude of NHF loss from the ocean is approximately two times higher in the model forcing field than in the Tropflux on 04 December 2017 (Fig. 3c). It is worth pointing out that the model NHF showed a consistent warming tendency after the passage of TC, which was in contrast to the cooling tendency apparent in Tropflux data.

Fig. 3
figure 3

Temporal evolution of (a) wind stress (m s−1), (b) Ekman pumping velocity (m day−1), from satellite-CCMP data (black line), and (c) Net surface heat flux (W m−2) from Tropflux data (black line) and model forcings (red line) during 29 November 2017 to 29 December 2017. All the middle and right panel parameters are averaged over the box 15.9°N–16.9°N and 68. 5°E–70.4°E (marked as the green box in Fig. 8b). The brown rectangle-shaded box at the bottom of the figure indicates the TC Ockhi period (29 November to 06 December 2017).

Full size image

Evaluation of model temperature and salinity with in-situ measurements in response to TC Ockhi

Before the passage of the TC, both the model and observation showed similar temperature structure, and the mixed layer depth (MLD) was also found to be the same (~ 40 m depth) (Figs. 4a, 5a, and 6a). The A-float measurements showed relatively warm and low salinity waters in November 2017 and relatively cold and high salinity waters in December 2017, which is a typical hydrographic characteristics of the Arabian Sea at the near-surface and are rightly reproduced by the model (Figs. 4a and b).

Fig. 4
figure 4

Depth-time section of (a) Temperature (°C), (b) Salinity (psu), (c) Chlorophyll (mg m−3), (d) Dissolved Oxygen (µM) from A-Float (WMO ID 2902202; left panel), simulated outputs from ROMS model (right panel) from 29 November 2017 to 29 December 2017. Except for A-Float measurements (left panel), all the parameters in the right panels are averaged over the box 15.9°N–16.9°N and 68. 5°E–70.4°E (marked as the green box in Fig. 8b). In panel (a), a thick black dashed line represents the depth of the 23 °C isotherm (D23; m), and a thin line is the depth of the 27 °C isotherm (D27; m). In panel (b), the thick black line represents MLD (m) and the thin dashed line is ILD (m). In panel (c), a thick pink line represents euphotic depth, a thin black dashed line represents ILD (m), and a thin red line represents MLD (m). In panel (d), a thin pink (black) line: depth of 175 (210) µM dissolved oxygen (m). The brown rectangle-shaded box at the bottom of the figure indicates the TC Ockhi period (29 November to 06 December 2017; all panels). The blue inverted triangle on top of the panel (a) indicates the surface date of A-Float profiling measurements.

Full size image
Fig. 5
figure 5

Temporal evolution of Sea Surface (a) Temperature (°C) (b) Salinity (psu), (c) Chlorophyll (mg m−3) and (d) Dissolved Oxygen (µM) from satellite (blue line), A-Float (5 m; black line with symbol), ROMS model simulations (1 m; green line). All the parameters from satellite and ROMS, except A-Float, are averaged over the box 15.9°N–16.9°N and 68. 5°E–70.4°E (marked as green box in Fig. 8b). The brown rectangle-shaded box at the bottom of the figure indicates the TC Ockhi period (29 November to 06 December 2017).

Full size image
Fig. 6
figure 6

Profiles of (a) Temperature (°C) (b) salinity (psu), and Sigma-T (kg m−3) from A-Float (black line), and model simulations from ROMS (green line; all panels). The solid and dashed lines indicates before (29 November 2017) and after (09 December 2017) the passage of TC Ockhi. All the parameters for ROMS (except A-Float), are averaged over the box 15.9°N–16.9°N and 68. 5°E–70.4°E (marked as the green box in Fig. 8b).

Full size image

Due to the impact of TC Ockhi, strong cooling in the near-surface temperature was recorded by A-Float, decreased by approximately 3.6 °C, i.e., 28.7 °C on 29 November 2017 to a minimum of 25.1 °C on 12 December 2017. Though the temporal evolution of near-surface temperature from ROMS simulations was similar to A-float measurements before the passage of TC, the cooling tendency was slightly higher in A-float measurements than in ROMS after the passage of TC. A similar reduction in the SST (~ 1.8 °C) was also noticed in the satellite measurement (Fig. 5a). It is worth pointing out that the maximum cooling apparent from the satellite and ROMS model was slightly underestimated (~ 1.8 °C) compared to the A-float observation (~ 3.6 °C) (Fig. 5a).

In response to TC, the MLD was deepened to 49 m (10 December 2017). In ideal conditions, when surface wind forcing and NHF loss are maximum, the depth at which pre- and post TC Sigma-T profiles intersect can be considered as the mixing depth (deepest depth unstable to shear instability36)(Sengupta et al. 200837, Girishkumar et al. 201929). Based on this, it is worth speculating that TC Ockhi-induced influences could reach as deep as 95 m in the water column since the sigma-t profiles before and after the TC Ockhi intersected at 95 m depth (Fig. 6c). The MLD simulated by the ROMS was found to be shallower (30–35 m) than observations after the passage of TC (Fig. 4a). It is interesting to note that the model estimated physical variables reached pre-cyclonic conditions immediately after the passage of the TC when compared with observations (Figs. 4a, 5a, and 6a).

In response to TC Ockhi, the near-surface salinity from A-float showed a mild increasing tendency (~ 0.3 psu) from 36 psu (29 November 2017) to 36.2 psu (11 December 2017), and this characteristic is also apparent in satellite measurements, such as salinity increased by 0.3 psu between 29 November (35.7 psu) and 09 December 2017 (36 psu) (Fig. 5). In contrast, SSS from ROMS simulations showed a mild reduction (~ 0.07 psu) between 29 November 2017 and 09 November 2017 (Figs. 4b, 5b, and 6b). Besides, the subsurface salinity maxima, which occurred between 30 and 60 m before TC, were nearly absent after the passage of TC (Fig. 6). This characteristic suggests that the vertical mixing of sub-surface high salinity water with relatively low saline near-surface water might be the plausible explanation for the absence of subsurface salinity maxima due to TC, as clearly apparent in the observation (Figs. 4 and 6). Before the passage of TC (on 29 November 2017), the near-surface salinity profile from the ROMS model was fresher than observations. As noted in A-Float, the ROMS also showed slight sub-surface salinity maxima around 25–50 m depth, but comparatively thinner than A-Float. Thus, in response to TC, the ROMS reasonably captured the physical phenomenon close to observations.

Chlorophyll response to TC Ockhi

Biogeochemical measurements from A-Float also showed a significant response to TC. Due to the impact of the TC Ockhi, the A-Float showed a substantial enhancement in the chlorophyll concentration at the near-surface, increased approximately by a factor of 35 (3.6 mg m−3; 11 December 2017) compared to the pre-TC state (0.1 mg m−3) (Figs. 4c, 5c, and 7c). It is also apparent from the A-float observation that the enhancement in the chlorophyll concentration (on 09 December 2017), started five days after the passage of the TC at the study location (04 December 2017), and reached back to that magnitude (0.4 mg m−3) on 19 December 2017 (Fig. 5c). This indicates that chlorophyll concentration enhancement persisted upto ten days after the passage of the TC (Fig. 5c). A similar evolution in the enhancement of the chlorophyll concentrations was also noticed from the satellite (Figs. 5 and 8). The peak enhancement in the near-surface chlorophyll was witnessed from both A-Float and satellite after the passage of TC Ockhi, with nearly comparable magnitude during 10–18 December 2017 (Figs. 5 and 8). It is worth pointing out that the peak enhancement observed in the chlorophyll concentration after the passage of the TC Ockhi was much smaller than other TC cases reported earlier in the Arabian Sea. For instance, Naik et al. 200828 and Subrahmanyam et al. 200238 reported an increase in the chlorophyll concentration of magnitude 4 mg m−3 and 5 mg m−3, respectively, in the north-eastern Arabian Sea (NEAS; ~ 15°N–17°N, 66°E–71°E) after the passage of a TC in December 1998 and May 2001.

Fig. 7
figure 7

Profiles of Chlorophyll (mg m−3) from A-float (black) and ROMS (green) on (a) 29 November 2017, (b) 09 December 2017, (c) 11 December 2017, (d) 20 December 2017 (before and after TC Ockhi).

Full size image
Fig. 8
figure 8

Spatio-temporal evolution of Sea Surface Chlorophyll (mg m−3) from (a–d) Satellite (i–l) ROMS model simulations. The green box represents the study location (15.9°N–16.9°N and 68.5°E–70.4°E).

Full size image

The ROMS simulation shows a substantial enhancement in chlorophyll concentration at the A-Float location in response to TC Ockhi (Figs. 4c, 5c, and 7). The surface chlorophyll concentration simulated by ROMS shows a reasonably good agreement with observation, though it slightly underestimates its values (~ 2 mg m-3) during the peak phase of the enhancement due to post-TC impact (Fig. 5c). The simulated chlorophyll from ROMS shows the immediate response to the TC and attains its maximum magnitude at the near-surface on 08 December 2017, whereas the observations reach their maximum magnitude on 11 December 2017 (Figs. 5c and 7c). The euphotic depth from ROMS also showed a shallow tendency from 72 m (29 November 2017) to 32 m (08 December 2017), but comparatively less than the observation (Fig. 4c). The chlorophyll distribution from the A-Float and ROMS along the depth-time section shows that the duration of the near-surface chlorophyll bloom event after the passage of TC was captured reasonably in the ROMS, as evident in the observation. However, the ROMS slightly underestimates the chlorophyll concentration at the near-surface with respect to observation (Figs. 4c, 5c, and 7).

The above analysis suggests that the ROMS model reasonably simulates the chlorophyll concentration in response to TC, and agrees with the observation, but with slight underestimation during the peak magnitude phase. The subsequent section will explain the plausible causative mechanism responsible for this chlorophyll bloom in response to TC Ockhi.

Plausible causative mechanism

The processes, such as new production due to vertical transport of subsurface nutrients into the near-surface layer, lateral redistribution through horizontal advection, and vertical transport of subsurface chlorophyll pigments into the near-surface layer , can work in tandem for the enhancement of chlorophyll concentration in response to TC 10,29,39,40,41.

Horizontal advection

The magnitude of horizontal advection processes on surface-chlorophyll showed a mean positive value in the model and satellite data from 29 November to 09 December 2017, suggesting that lateral transport of chlorophyll pigment to the A-float location in response to the TC Ockhi (Table 1). It is worth pointing out that though the mean contribution of horizontal advection of chlorophyll from both ROMS and satellite showed nearly comparable magnitude (0.01 mg m−2 day−1) at A-Float location,  − but shows the difference in the magnitude between satellite (0.02 mg m−2 day−1) and ROMS (0.4 mg m−2 day−1) after the passage of the TC (on 09 December 2017; Table 1). The difference in the magnitude of horizontal advection noticed between the ROMS and the satellite might be due to the alleged role of background chlorophyll conditions.

Table 1 Horizontal advection of chlorophyll from satellite and ROMS model.
Full size table

Thus the spatial distribution of chlorophyll in the NEAS before and during the period of TC Ockhi shows that both the ROMS and satellite showed a comparable magnitude from a qualitative perspective (Fig. 8). However, we can negate the role of horizontal advection on the enhancement of chlorophyll during the TC Ockhi since the chlorophyll concentration outside the track of TC was much smaller than the values along the track (Fig. 8).

Subsurface chlorophyll

This section evaluated the relative role of vertical transport of subsurface chlorophyll concentration on chlorophyll bloom after TC Ockhi passage. For this purpose, the vertical chlorophyll profiles from A-float and ROMS were examined before and after the passage of TC Ockhi at the A-float location (Fig. 7). As depicted in the previous section, the peak enhancement of near-surface chlorophyll noticed from ROMS simulation slightly underestimates the A-Float measurements (Figs. 5 and 7c).

The existence of subsurface chlorophyll maxima (SCM) due to the optimal availability of nutrients and light is an omnipresent feature in the tropical Indian Ocean10,42,43. The ROMS has nicely captured the depth regime of SCM (~ 25–66 m depth) as apparent in the observation (~ 30–64 m). However, its magnitude was much higher (2.6 mg m−3) in ROMS than the observation (0.5 mg m−3) (Fig. 7a). In response to TC, the SCM was completely absent during the peak phase of chlorophyll concentration at the near-surface (Fig. 7b). This characteristic suggests that there is the possibility of vertical transport of chlorophyll pigment from SCM to the near-surface layer.

However, to quantify the contribution of SCM to chlorophyll enhancement at near-surface in response to TC, we have considered depth-integrated chlorophyll upto 100 m depth. The choice of depth-integrated chlorophyll upto 100 m depth is due to the non-availability of light, leading to the lack of or negligible chlorophyll variability after 100 m depths.

Before the passage of the TC, the depth-integrated chlorophyll upto 100 m depth from A-float was just 18 mg m−2 (29 November 2017), and it increased initially to 32 mg m−2 (09 December 2017) after the passage of the TC (Table 2). It gradually increased further to 76 mg m−2 (11 December 2017) and reached its maximum magnitude upto 95 mg m-2 on 15 December 2017, which is a factor of three higher than the immediate post-TC depth-integrated chlorophyll and a factor of five higher than the pre-TC depth-integrated chlorophyll. The initial increase of 14 mg m−2 (from 18 to 32 mg m−2) in the depth-integrated chlorophyll immediately after the passage of TC might be due to the contribution from the upward movement of the chlorophyll from SCM to the near-surface i.e., almost 78% of increase in chlorophyll is due to the contribution from subsurface. Please note that after this period, there is no evidence for SCM (Fig. 7). Thus, the role of SCM is very limited for further enhancement. Moreover, depth-integrated chlorophyll upto 200 m also showed a similar factor of enhancement compared to the estimation of chlorophyll based on depth-integration upto 100 m. This suggests that the integration depth does not significantly impact inference. This analysis shows that SCM provides the biomass background for rapid phytoplankton growth near the surface immediately after the passage of the TC. However, their contribution is relatively small compared to the peak magnitude of the chlorophyll bloom period at the near-surface.

Table 2 The depth-integrated chlorophyll up to 100 m depth from A-float and ROMS before and after the passage of TC Ockhi.
Full size table

Similarly, depth-integrated chlorophyll simulated from ROMS showed an increase in the chlorophyll of approximately 47% immediate after the passage of TC (45 to 66 mg m−2; Table 2). The increase in the depth-integrated chlorophyll immediate after the passage of TC is quite high in ROMS (21 mg m−2) than the observation (14 mg m−2). This might be due to the enhanced SCM in the ROMS than observations before the passage of the TC.  However, the percentage contribution from the SCM to the increase in depth-integrated chlorophyll is relatively higher in the observations (78%) as compared to ROMS (47%). This shows that ROMS reasonably mimics the percentage contribution of SCM on the enhancement of depth-integrated chlorophyll immediate after the passage of TC, as apparent in the observation. It is worth pointing out that the percentage contribution of SCM on depth-integrated chlorophyll enhancement reported here is consistent with previous studies29.

Nutrient input

As stated above, depth-integrated chlorophyll derived from A-Float and ROMS showed an enhancement of 78% and 47% of chlorophyll, immediately after the passage of the TC Ockhi, primarily due to the vertical entrainment of SCM.   However, there was no substantial evidence of SCM during the peak magnitude phase of chlorophyll at the near-surface during the post-TC period. Hence, it is worth speculating that the enhancement in the chlorophyll (3.6 mg m−3 on 11 December 2017) at the near surface might be due to the new production with optimum availability of nutrients and light. . Please note that the study region is far away from the river mouth; hence, the measurements were free from suspended inorganic sediments at the near-surface layer. The nitrate profile from the ROMS simulation showed an upward flux of nutrients from the sub-surface to the near-surface layer (Fig. 9b). Before the passage of TC Ockhi (on 03 December 2017), ROMS simulated nutricline depth (defined as 2.5 μM) was around 48 m, and near-surface oligotrophic conditions (~ 0.2 μM) extended up to the depth of 30 m. After the passage of TC, the nitrate concentration from ROMS increased only up to 0.5 μM on 06 December 2017 at the near-surface, and the nutricline was found around 27–33 m depth (Fig. 9b). Please note that in-situ nitrate measurements were not available during the TC Ockhi period. Earlier studies have demonstrated good agreement between Isothermal Layer Depth (ILD) and nutricline depth29, suggesting that the ILD can be considered as a proxy for nutricline depth in the observation. The ILD estimated from A-Float showed reasonable agreement with ILD and nutricline depth simulated by the ROMS, except during shallowing tendency (Fig. 9a). The ROMS simulated ILD, as well as nutricline depth, showed shallowing tendency immediately after the passage of the TC (04 December 2017) at the A-Float location, whereas, in A-Float tends to shallow from 09 December 2017 onwards, five days after the passage of the TC and reaches shallowest depth on 10 and 12 December 2017 which was coherent with the peak of the chlorophyll bloom at the near-surface (Fig. 9a). These characteristics indicate that the enhancement of chlorophyll is not due to the entrainment of nutrients into the near-surface during the post-TC period. Hence, an increase in nutrient supply to the near-surface layer due to vertical mixing during the TC forcing period might be a plausible explanation for the enhancement of chlorophyll during the post-TC period. Moreover, an increase in nitrate concentration does not lead to an immediate, rapid increase in chlorophyll concentration at the near-surface layer during the TC29.

Fig. 9
figure 9

(a) ILD from A-Float (black) and ROMS (green) along with nutricline depth (blue); (b) Depth-time section of Nitrate (µM) from ROMS model during 29 November 2017 to 29 December 2017.

Full size image

Dissolved oxygen during the TC Ockhi

Similar to chlorophyll concentration, DO concentration also showed a strong response due to the impact of the TC Ockhi. Before the passage of TC, the DO concentration of 22 μM (which is assumed as the OMZ)8 was found to be shallow, which deepened after the passage of the TC, and it was observed in the range between 80 to 100 m depth from A-float measurements (Fig. 4d). The maximum vertical gradient (considered as oxycline) was observed around 30–50 m depth. In response to the TC Ockhi, the near-surface DO concentration from A-float showed a reduction initially of about 11 μM (from 184 μM on 29 November 2017 to 173 μM on 09 December 2017) (Figs. 4d and 5d). The plausible cause for the reduction of DO concentration at the near-surface layer might be due to the strong wind-induced vertical mixing that causes subsurface low oxygen concentration water intrusion into the near-surface layer. This might be evident from the shallow tendency of the 175 μM contour of DO from 34 m (before TC; on 29 November 2017) to the near-surface after the passage of TC. Later, dissolved oxygen concentration at the near-surface layer showed enhancement in oxygen concentration (> 200 μM) from 10 to 14 December 2017, with a maximum observed at near-surface on 11 December 2017 (209 μM), which was approximately 25 μM higher than the pre-TC period (Figs. 4d and 5d). This might be due to two possible factors that might lead to the enhancement of oxygen concentration at the near-surface layer: (i) the entrainment of overlying air due to strong winds29,44 (ii) the enhancement of photosynthetically produced oxygen due to the enrichment of chlorophyll concentration at the near-surface layer in response to TC Ockhi29. However, satellite estimation showed a significant reduction in wind stress after the passage of TC, which negates the role of strong winds-induced enhancement of oxygen concentration at the near-surface layer (Fig. 3). Hence, we can assume that the plausible factor for the enhancement of oxygen concentration must be associated with the enhancement of chlorophyll concentration at the near-surface29.

A mild response was noted from the ROMS simulated DO concentration. In response to TC Ockhi, the near-surface DO concentration from the ROMS model showed a mild reduction of 2 μM, i.e., from 200 μM (03 December 2017) to 198 μM (05 December 2017) and returned to pre-TC state on 06 December 2017 (200 μM) onwards (Figs. 4d and 5d). This can be evident from a mild shallowing tendency of the 175 μM contour (Fig. 4d). However, the 175 μM contour does not reach to near-surface due to the impact of the TC, as evident in the observation (Fig. 4d). The ROMS estimated DO at the near-surface completely overestimated the observation, except during the peak period of enhanced oxygen in A-Float due to the impact of TC, where the model slightly underestimated the observation (Fig. 5d). It is worth pointing out that the near-surface DO concentration from the ROMS took only 2 days to reach the pre-TC condition, whereas, from observations, it took 5 days to reach the pre-TC state after the passage of TC. The strong vertical gradient of DO concentration from the ROMS estimated contour of 22 μM (considered OMZ) was much deeper than 100 m as compared to observations.

Evaluation of model-simulated upper ocean primary productivity with estimation based on in situ measurements in response to TC Ockhi

Figure 10a shows the comparison of estimated with surface chlorophyll, temperature (solid lines; 5 m for A-Float and 1 m for others; ) and integrated over Zeu (dashed lines; ) from A-Float (black), Satellite (blue; only with surface merged chlorophyll; mg m-3) and MW-OI SST (°C), and ROMS model (green). Figure 10b and c show the depth-time evolution of PP estimated from vertical emperature and chlorophyll profiles of the A-Float and ROMS. In general, the temporal evolution of chlorophyll was very similar to the temporal evolution of estimated PP. In response to TC Ockhi, the estimated from the satellite and A-float were increased maximum of 1.5 g C m−2 day−1 (15 December 2017) and 2.0 g C m−2 day−1 (12 December 2017), respectively. A similar increase in the estimation of (1.7 g C m−2 day−1; 08 December 2017) was also noted in the ROMS after the passage of the TC. Theestimation from A-float and ROMS also showed similar but with less enhancement compared with (1.2 g C m−2 day−1). This indicates that average production within euphotic depth was approximately 1 and 0.9 g C m−2 day−1, respectively, from the A-Float and ROMS model estimated PP. It is worth pointing out that after the passage of TC, the estimated from A-Float and ROMS were slightly higher than . It might be associated with using realistic subsurface chlorophyll information rather than the values derived from the surface values45. The maximum value of from the ROMS model was in good agreement with observations as well as with satellite-based estimation. Due to the impact of TC, the satellite and model estimated PP showed an immediate response during and after TC, whereas that information was unavailable from A-Float due to lower temporal resolution. However, earlier studies suggest that there was an immediate response of estimated PP from Argo float observations during and after the passage of TC Hudhud and TC Vardah29.

Fig. 10
figure 10

(a) Temporal evolution of primary productivity (g C m−2 day−1) estimated using surface (solid line; 5 m from A-Float and 1 m depth for others; ) and integrated upto Zeu (dashed lines; ) from A-Float (black), satellite (blue), ROMS (green) model and time-depth section of estimated primary productivity from (b) A-Float (c) ROMS model. Green line in b, c resembles euphotic depth. All the parameters from satellite and ROMS, except A-Float, are averaged over the box 15.9°N–16.9°N and 68.5°E–70.4°E (marked as a green box in Fig. 8b). The brown rectangle-shaded box at the bottom of the figure indicates the TC Ockhi period (29 November to 06 December 2017).

Full size image

Summary and conclusion

Though significant advancements in coupled bio-physical ocean modeling have been achieved during the last two decades, the performance of these models during extreme weather events, such as TCs, is not well documented across the global ocean, primarily due to a lack of in-situ observation. The passage of the Very Severe Cyclonic Storm Ockhi (29 November–06 December 2017) over the Arabian Sea coincided with the proximity of a A-Float (WMO ID - 2902202) equipped with temperature, salinity, chlorophyll-a, and DO sensors, provided a unique opportunity to evaluate ecosystem model skills in simulating the upper-ocean biogeochemical responses to such extreme weather events. Following the TC Ockhi passage, the A-Float was triggered into high-frequency sampling mode, increasing its profiling rate from once every 10 days to twice daily. Before initiating the evaluation of the model simulation with observations from A-Float, chlorophyll-a measurements from the A-Float were validated using chlorophyll estimates derived from ship-based water samples.  Before correction, the chlorophyll measurement from A-float showed very high concentration at near-surface in response to the TC (~ 13 mg m−3 on 11 December 2017; maximum magnitude reached after TC; figure not shown). After correction, the surface chlorophyll concentration from A-float was reduced significantly to 3.6 mg m−3, which is almost close to the satellite estimation. This highlights the importance of validating the biogeochemical variables from BGC floats with water sample measurements.

Due to the impact of the TC Ockhi, the near-surface temperature from A-float showed a strong reduction by 3.6 °C (cooling) and a mild increase in the near-surface salinity due to strong wind-induced vertical mixing. A similar evolution with a mild reduction in SST (~ 1.8 °C cooling) was noticed from the satellite and the ROMS. It is worth pointing out that the model estimated physical variables reached pre-cyclonic conditions immediately after the passage of the TC as compared with observations.

As evident in physical variables, BGC parameters from A-float also showed a significant response due to the impact of the TC Ockhi. In response to the TC Ockhi, the surface chlorophyll from A-Float increased approximately by a factor of 35 (3.6 mg m−3) compared to the pre-TC state (0.1 mg m−3). A similar evolution in the chlorophyll concentration at the near-surface has also been noticed from the satellite and ROMS, where ROMS simulated surface chlorophyll slightly underestimated its values (~ 2 mg m−3) during the peak phase of the enhancement due to the impact of TC. It is worth pointing out that the ROMS simulated chlorophyll concentration showed an immediate response to the TC with the enhancement of the chlorophyll at the near-surface, whereas, in the A-float, the enhancement in the surface chlorophyll concentration was apparent only 5 days after the passage of the TC. Moreover, the ROMS reasonably mimics the percentage contribution of SCM to the enhancement of depth-integrated chlorophyll immediate after the passage of TC, as apparent in the observation. Such that, there is approximately 78% and 47% enhancement in depth-integrated chlorophyll from A-Float and ROMS, respectively, due to contribution from the SCM. However, there was no substantial evidence of SCM during the peak phase of chlorophyll enhancement at the near-surface due to post-TC impact. Hence, the later peak enhancement in the chlorophyll (3.6 mg m−3 on 11 December 2017) at the near surface might be due to the substantial amount of new production with optimum availability of nutrients and light at the near-surface. The ILD (which is considered a proxy for nutricline depth) estimated from A-Float showed a shallowing tendency during the post-TC enhancement phase of chlorophyll. These characteristics indicates that this enhancement of the chlorophyll at the near-surface during the post-TC period is not due to the entrainment of nutrients into the near-surface. Hence, an increase in nutrient supply to the near-surface layer due to vertical mixing during the TC forcing period might be a plausible explanation for the enhancement of chlorophyll during the post-TC period. 

Similarly, the DO concentration from the A-float showed reduction (11 μM) initially in response to the TC and later, enhancement (25 μM) at the near-surface layer. The reduction of the oxygen concentration at the near-surface layer might be due to strong wind-induced vertical mixing that causes intrusion of subsurface low oxygen concentration water to the near-surface layer, and the enhancement of near-surface oxygen might be due to the enhancement of photosynthetically produced oxygen due to enrichment of chlorophyll concentration at the near-surface layer. A similar tendency with mild response was noticed from the ROMS simulated DO concentration. Further, our analysis of PP shows that the average new production within euphotic depth was approximately 1 and 0.9 g C m−2 day−1 from the A-Float and ROMS, respectively.

This analysis suggests that the model-simulated biogeochemical variables were overestimated with in-situ observations at the near-surface, except during the peak magnitude phase of chlorophyll and DO in response to TC impact, where the model showed slight underestimation than observations. Moreover, the temporal response captured by the model was also so immediate in contrast with observation, where substantial time was observed for the chlorophyll bloom event at the near-surface in response to TC. Thus, measurements from A-Float before and after the passage of TC provide a unique and unprecedented opportunity to understand the physical and biogeochemical state in response to TC and are also used to validate the performance of the model’s ability to simulate this physical and biogeochemical state in the study region. Considering the enhancement of tropical cyclone activity, particularly in the global warming scenario, the regional model simulation must be improved to capture this variability, especially during extreme events.

Data and methods

The life cycle of TC Ockhi

The TC Ockhi originated as a depression on 29 November 2017 at 0300 UTC in the western Bay of Bengal, close to the east of Sri Lanka, around 6.5°N, 81.8°E, and on 29 November 2017 at 1200 UTC, it crossed Sri Lanka and entered the Arabian Sea. It developed as a well-organized cyclonic storm (CS; 34–47 knots) on 30 November 2017, 0300 UTC at 7.5°N, 77.5°E. TC Ockhi intensified into a severe cyclonic storm (SCS; 48–63 knots) on 01 December 2017 0000 UTC and further intensified into a very severe cyclonic storm (VSCS; 64–89 knots) on 01 December 2017 at 9.1°N, 73°E, and was continued to be as VSCS until 04 December 2017 1500 UTC. On 04 December 2017 at 1800 UTC, the system weakened to SCS at 16.5°N and 69.8°E. During this period, TC Ockhi crossed nearest to the A-Float profiling location (16.36°N and 69.66°E) with a maximum sustained wind speed of 55 knots (marked as the green triangle in Fig. 1). TC Ockhi made landfall on 06 December 2017, adjoining south coastal Gujarat and north coastal Maharashtra near 19.2°N and 71.9°E (Fig. 1).

Measurements from A-Float

A-Float is equipped with a Seabird (SBE-41CP) CTD sensor, a Wetlabs AP2 sensor to measure optical backscatter (at a centroid angle of 142° at 700 nm) and chlorophyll fluorescence (at an excitation wavelength of 470 nm and emission wavelength of 695 nm) and Aanderaa Optode 4330 sensors (to measure dissolved oxygen) and obtains data at predetermined depths (an interval of 5 dbars between 200 and10 dbars; and an interval of 1 dbars from 10 dbars to surface). In this study, we utilized data in the upper 120 m of the water column. Since the profiles from BGC-Argo floats are not measured uniformly vertically, they were linearly interpolated from the surface to 120 m depth with 1 m intervals. Since there are no measurements in the upper 5 m from the A-float, the measurements at 5 m depth are considered as surface in this analysis. The comparison of chlorophyll data at 5 m from A-float with satellite shows reasonable agreement, which provides us with the confidence to consider 5 m depth measurements as the surface. The accuracies of pressure, temperature, salinity, chlorophyll-a, optical backscattering, and dissolved oxygen sensors from A-Float are 2.4 dbars, 0.002 °C, 0.005 psu, 0.015 mg m−3, 0.0015 m−1, and < 8 µM, respectively. The temperature and salinity data are quality controlled by following Wong et al.46 and Uday Bhaskar et al.47. In this study, we have analyzed data from 29 November 2017 to 29 December 2017.

Validation of A-float measured chlorophyll and dissolved oxygen measurements with water sample estimation

Figure 2 shows the validation of the chlorophyll measurements obtained from A-Float, profiled on 05 February 2018 at 18:25 UTC (upper middle panel; continuous black thick line is uncorrected; dashed black thick line is corrected), ship portable CTD (upper left panel; continuous blue line is uncorrected; dashed blue line is corrected) and in-situ measurements obtained through Ship rosette where seawater samples were collected when the ship was near to A-Float location (16.73°N, 67.95°E; ~ 2 km distance on 06 February 2018 at 00:20 UTC; star symbol in Fig. 1 and thin line with circle in Fig. 2a, b, and c). The CTD Rosette was used to collect seawater samples in the upper 120 m of the water column (at depths 1, 10, 20, 40, 60, 80, 100, and 120 m). The chlorophyll-a concentration from the water samples was estimated using a spectrophotometric method where a seawater sample was filtered onto 47 mm diameter Glass Fiber Filters (GF/F) at low suction pressure. The filters were extracted with 90% acetone for 24 h at low temperatures. An aliquot of the supernatant solution was transferred to a 1 cm path length cuvette, and the extinction coefficient was measured at different wavelengths (630, 645, 665, and 750 nm) using a UV–visible spectrophotometer. Then, Chlorophyll-a concentration was calculated using the equation of Strickland and Parsons48 and expressed as mg m−3. The accuracy of the analysis was ensured with duplicate measurements.

Following Takeshita et al.49, the ship-based chlorophyll measurements (from water samples) were used to correct chlorophyll measurements (Chl-a) from the A-Float29. The expression for corrected BGC-Argo float chlorophyll is:

$$\left[ {{\text{Chl}} - {\text{a}}} \right]_{{{\text{float}}\_{\text{Corr}}}} = {\text{offset }} + {\text{ gain x }}\left[ {{\text{Chl}} - {\text{a}}} \right]_{{{\text{float}}}}$$
(1)

The values of offset and gain in Eq. 1 were determined by performing a model II linear regression on [Chl-a] float versus [Chl-a] ship. Using Eq. 1, the uncorrected chlorophyll measurements from portable ship CTD (on 06 February 2018 at 00:20 UTC) and A-float (on 05 February 2018 at 18:25 UTC) were corrected with water samples based on chlorophyll estimation. The comparison of uncorrected chlorophyll profiles from the A-Float and ship CTD profiler with water sample estimation showed large deviations up to 80 m depth, where the deviation was maximum at subsurface chlorophyll maxima (SCM) depth (Fig. 2a and b). After correction, the chlorophyll profiles from A-Float showed reasonably good agreement with chlorophyll estimation based on water samples, where the Root Mean Square Deviation (RMSD) was found to be 0.11 mg m−3 (Table 3). Further, to avoid fluorescence quenching-related issues, the chlorophyll measurements from A-Float profiled during the nighttime were considered for our analysis.

Table 3 Comparison between chlorophyll measurements obtained from water sample measurements and A-Float during February 2018. RMSD between (a) and (b) is 0.11 mg m−3.
Full size table

Similarly, the comparison of corrected and uncorrected dissolved oxygen measurements from A-Float with water samples based estimation does not show any significant deviation and are reasonably good agreement with each other, and RMSD was found to be 12.7 μM and 9.6 μM, respectively, for uncorrected and corrected dissolved oxygen measurements (Figure not shown; Table 4). The difference in the RMSD of corrected and uncorrected DO is 3.1 μM, which is far less than the accuracies of the DO optical sensors (< 8 μM). Hence, dissolved oxygen measurements from A-Float were directly used for our analysis without any correction.

Table 4 Comparison between dissolved oxygen measurements obtained from water sample measurements and A-Float with and without correction during February 2018. RMSD between (a) and (c) is 12.7 μM, and between (b) and (c) is 9.6 μM.
Full size table

Satellite and reanalysis data

The Remote Sensing Systems Cross Calibrated Multi Platform (CCMP) is a gridded level-4 product that provides a combination of ocean surface (10 m) wind retrievals from multiple types of satellite microwave sensors and a background field from reanalysis50 with a spatial resolution of 0.25° × 0.25° and temporal resolution of 6 h. The TropFlux radiative and turbulent heat flux data51 are utilized to assess the impact of net surface heat flux on the observed variability of mixed layer depth. Advanced Microwave Scanning Radiometer for EOS (AMSR-E), AMSR-2, Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI), and WindSAT Microwave Optimum Interpolation Sea Surface Temperature product (MW-OISST) with 0.25° (~ 25 km) grid spacing52 are used to understand the spatio-temporal evolution of SST. Daily merged satellite product of chlorophyll-a concentration data based on Moderate Resolution Imaging Spectroradiometer (MODIS)-AQUA and Visible and Infrared Imager / Radiometer Suite (VIIRS) with the spatial resolution of 4 km obtained from GlobColour was used to characterize the surface bloom in this study53,54. Three hourly best TC tracks for TC Ockhi are obtained from the Regional Specialised Meteorological Centre (RSMC), Indian Meteorological Department (IMD), India (http://www.imd.gov.in/section/nhac/dynamic/bestrack.htm) to examine the life cycle of TC.

The horizontal advection of the surface chlorophyll is examined using the following expression55.

$$\text{Horizontal Advection }=-\left[u\frac{\partial (ch{l}_{a})}{\partial x}+v\frac{\partial (ch{l}_{a})}{\partial y}\right]$$
(2)

where u and v are the zonal and meridional current velocity \(\frac{\partial (ch{l}_{a})}{\partial x}\) and \(\frac{\partial (ch{l}_{a})}{\partial y}\) are the zonal and meridional gradients of chlorophyll at a near-surface.

Model description

Following Kunal et al. 201934, the hydrodynamic model for the Indian Ocean basin (30° S to 30° N; 30° E to 120° E) has been configured using a Regional Ocean Modeling System (ROMS) version 3.733 coupled entirely online with an ecosystem module that consists of the nitrogen cycle model containing parameterized sediment denitrification as described by Fennel et al. 200656. Following Zeebe and Wolf-Gladrow 200157, the model of carbonate chemistry with horizontal resolution of 1/4° (~ 25 km, hereafter ROMS-25) was adopted. This study used a ROMS model with a resolution of ~ 25 km (ROMS-25; 1/4°) carrying 40 vertical sigma levels to validate with In-situ observations. Detailed model setup, physical and biological variable initialization, boundary conditions, and atmospheric forcing are described in Kunal et al. 201934.

Estimation of primary productivity

Following Girishkumar et al. 201929, the primary productivity (PP) was estimated by modifying the standard Vertically Generalised Production Model (STD-VGPM), a well-known PP model of Behrenfeld and Falkowski 199758. Instead of surface temperature and surface chlorophyll, vertical profiles of chlorophyll (including subsurface chlorophyll information and temperature from A-float measurements were utilized along with information on available light and photosynthetic efficiency to derive primary productivity profiles over the water column and defined as 29. The modified expression for the vertically integrated PP over the Zeu based on the STD-VGPM PP profile () is:

(3)

where, is the daily vertically integrated primary productivity (g C m−2 day−1) in the euphotic layer, z0 is the surface depth (1 m), zeu is the euphotic depth (m), is the surface (at depth z) chlorophyll (mg Chl m−3), and Dirr is the daily photoperiod (decimal hour) representing the number of the daylight hours available at that particular location and calculated based on latitude and day of the year. is the photosynthetically available radiation (PAR) (mol m−2 d−1) at the surface (at a depth z). The penetrative component of shortwave radiation (Qpen) at the surface (at a depth z) was estimated using the algorithm proposed by Paulson and Simpson 197759, the empirical form of a single exponential profile expressed as follows,

$${Q}_{pen} = {Q}_{shortwave}*(1-\gamma )*exp(-Z/\zeta )$$
(4)

Here, Qshortwave is Net shortwave radiation obtained from TropFlux53, Z is a depth in m, and γ is considered as 0.5860,61, which is the fraction of incident sunlight that accounts mainly for the red and infrared part of the spectrum that is absorbed in the upper ~ 2 m in the equation 4. ζ is the diffuse attenuation coefficient of shortwave radiation in the wave band 380–700 nm, and it is estimated using the expression ζ = 1/(0.027 + 0.0518 * CHL0.428), where CHL represents chlorophyll data from A-Float60,61,62. The adoption of this single exponential formulation is subject to the assumption of simulating the penetrative component of shortwave radiation with sufficient accuracy, provided the relatively small spectral variations over the relatively clear-water open ocean regions of interest. Light penetration is inversely related to chlorophyll as the phytoplankton increases, zeu shallows, and vice-versa. The zeu is calculated from surface chlorophyll concentration using the Morel and Berthon63 case-I model. Here, surface chlorophyll is taken from the A-Float. The index of optimum photosynthetic flux in the euphotic layer, is considered to be equivalent to the daily maximum carbon fixing intensity in the euphotic layer (g C (mg Chl)-1 h-1). This index is estimated based on sea surface temperature (water temperature at a particular depth z) by a seventh-order analytic function prescribed by Behrenfeld and Falkowski 199764.

Note that our analysis is primarily restricted to the A-Float location. The availability of sub-surface data from A-Float locations provides an unprecedented opportunity to evaluate the ecosystem model performance comprehensively. The discussion based on data other than the A-Float location will be stated otherwise. Further, Sea Surface Salinity (SSS) data were used from SMAP65. SMAP sea surface salinity (SSS) V6.0 validated data are available from Remote Sensing Systems.