SEARCH

SEARCH BY CITATION

Keywords:

  • SST and salinity of Bay of Bengal;
  • monsoon ISO

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References

[1] High resolution observations of atmospheric and oceanic variables are carried out at northern Bay of Bengal from 22nd July to 6th August 2009 on-board ORV Sagar kanya under the Continental Tropical Convergence Zone research/observational programme. Freshening of surface layer by more than 4 psu within 24 hours is observed, which is followed by warming in the surface layer temperature. The heat and salt budget analysis primarily indicates dominant role of advection processes on the evolution of temperature and salinity. The amount of rainfall received at observation site could not explain the observed freshening, thus an extensive analysis using wavelet coherence is done to find out the source of advected fresh water to the observed location. It is found that surface salinity in the northern Bay of Bengal (at 15° N) varies coherently with the rainfall over Ganga-Brahmaputra catchment area on intraseasonal time scale and with lag of about 60 days. Based on above observations, this study hypothesize that the intraseasonal rainfall variation modulates the amount of river discharge, which in turn modulates the salinity over northern Bay of Bengal on intraseasonal time scale. Since surface warming always follows the surface freshening, the time delay between the rainfall over catchment area and freshening at northern Bay of Bengal may explain the post monsoon warming. Importance of atmosphere-ocean coupling in driving the dynamics of the northern bay of Bengal has been clearly brought out in this study.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References

[2] The northern Bay of Bengal (BoB) during summer monsoon (June-September) is generally characterized by high sea surface temperature (SST), shallow surface layer of low salinity and weak wind compared to the Arabian Sea [Shenoi et al., 2002; Sprintall and Tomczak, 1992; McCreary et al., 1993; Schott and McCreary, 2001]. The weak winds cannot overturn the stratified low-salinity surface layer and hence results into a shallow surface mixed layer. Once the strong stratification in the surface layer takes place, the SST is mainly driven by surface net heat flux. As a consequence, though the Arabian Sea and BoB gets almost same amount of net radiative heating, the SST over BoB is much warmer than that over Arabian Sea [Shenoi et al., 2002]. In general, the SST of northern BoB remains higher than 28° C throughout the year, a condition favorable for generation of active convection [Gadgil et al., 1984; Graham and Barnett, 1987]. SST values higher than 28° C supports large-scale deep convection in the atmosphere during the Indian summer monsoon season. Release of latent heat due to condensation maintains the atmospheric heating and sustains the monsoon circulation and associated rainfall over Indian subcontinent [e.g., Webster et al., 1998]. Most of the studies on monsoon breaks have identified mechanisms involving various atmospheric processes as the possible reason for break monsoon conditions [Krishnan et al., 2000]. But, oceanic parameters (like SST and salinity) also play crucial role in the evolution of break conditions [e.g., Vecchi and Harrison, 2002]. The intraseasonal SST changes can be associated with changes in the surface winds and atmospheric convection over BoB. On the other hand, these atmospheric changes may be responsible for the observed SST variability over the Bay. Therefore, the intraseasonal SST variability and monsoon rainfall are dynamically linked with each other [Parampil et al., 2010; Vecchi and Harrison, 2002].

[3] Fresh water from the local rainfall and river discharges maintains the low salinity surface layer over the BoB [Parampil et al., 2010; Shetye et al., 1996]. During the Bay of Bengal Monsoon Experiment (BOBMEX) in July-August 1999 several active and weak spells of convection occurred [Bhat et al., 2001] and it was found that the SST decreases during rain events and increases in cloud free conditions. During phase-I (27th July to 6th August 1999) of BOBMEX, the northern bay was convectively very active and three monsoon systems developed, SST remained around 28.5° C and arrival of a low-saline water plume was reported [Bhat, 2002; Vinayachandran et al., 2002]. This plume caused the surface salinity to decrease from 33 practical salinity unit (psu) to less than 29 psu in a span of just 4 days. However, the dynamics behind the abrupt decrease in salinity is not understood completely. Phase-II of BOBMEX (13–24th August 1999) experienced a weak phase of convection with low winds and clear sky conditions. An increase of 1.5° C SST was observed (from 28° C to 29.5° C) within an interval of 5 days. Bhat [2002] made an one-dimensional heat budget analysis (so called calorimetric experiment) and used observed surface fluxes to predict SST. This model had limited success and could not predict SST for all conditions prevailing over northern BoB. They argued that, the horizontal advection is important for determining SST, but it was not possible to quantify the relative role of advection on the SST evolution. Vinayachandran et al. [2002] suggested the source of fresh water plume mainly to river discharge, which got advected from east coast due to Ekman flow and consequently accelerated due to presence of two low pressure systems over BoB. This study gave a broad overview of the source of freshwater plume, but could not localize it and cannot identify how intraseasonal oscillation of rainfall modulates this source.

[4] Using in-situ observation, we have for the first time quantified the relative role of advection and surface net heat flux on the evolution of SST and salinity in northern BoB. Longer period of salinity and SST data from Research Moored Array for African-Asian-Australian Monsoon Analysis and Prediction (RAMA) buoy have been utilized to show that the changes in salinity are linked with the rainfall over Ganga and Brahmaputra river catchment area. Section 2 describes the data used for this study and applied methodology. Results are discussed in section 3. The outcomes of this study are summarized in section 4.

2. Data and Methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References

[5] The Indian Climate Research Programme (ICRP) has formulated the Continental Tropical Convergence Zone (CTCZ) programme to understand the mechanisms leading to space-time variation of the CTCZ and the embedded monsoon disturbances during the summer monsoon. The complete overview and detailed objectives of CTCZ programme is available at http://www.imd.gov.in/SciencePlanofFDPs/CTCZ. Under this national program (CTCZ), oceanic and atmospheric observations are taken on-board ocean research vessel (ORV) Sagar Kanya (SK-261). The cruise track is shown in Figure 1. The research vessel started on 15th July 2009 from the port of Chennai and reached the time series location (TSL) at 89° E, 19° N on 22nd July and stationed there for 15 days (up to 6th August). Two hourly profiling up to the depth of 760 m was done for the measurement of salinity and temperature. Total 218 profiles were measured using Conductivity-Temperature-Depth (CTD) instrument, manufactured by Idronaut. To estimate the horizontal advection of salinity and temperature, CTD measurement at 4 other locations situated at North, South, East and West of the TSL and at a distance of 3 nautical miles from the TSL were carried out. In order to avoid the influence of solar insolation, the CTD measurement at North, South, East and West locations are done once at night and up to the depth of 150 m. The temperature and conductivity sensor of Idronaut CTD has an accuracy of 0.001° C and 0.0001 Sms−1 respectively and scan frequency is of 40 Hz. The speed of probe during downcast and up-cast was between 0.5–1 m s−1. The downcast and up-cast data were averaged into 1m depth bins. The tendency of temperature and salinity can be calculated by using the following equations [Nisha et al., 2009]:

  • equation image
  • equation image

where, T is the ocean water temperature, u, v, w are zonal, meridional and vertical component of the currents, Qnet is the net surface heat flux, ρ is the density of water, h is the mixed layer depth, CP = 3993.0J kg−1K−1 is the heat capacity of water, D represents tendency due to diffusion, S is the salinity of water, E, P are evaporation and precipitation rate respectively. wh is the vertical advection below mixed layer (m month−1), H is the Heaviside step function [=0 if (wh + equation image) > 0, = 1 if (wh + equation image) < 0] and Sh is the salinity just below the mixed layer base.

image

Figure 1. Cruise track (yellow lines). Shaded color shows the JJAS climatology mean of surface salinity (in psu).

Download figure to PowerPoint

[6] The net surface heat flux is used from NCEP reanalysis [Kalnay et al., 1996]. The oceanic u, v currents are from NCEP Global Ocean Data Assimilation System (GODAS, data provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their Web site at http://www.esrl.noaa.gov/psd/). High resolution oceanic currents (Ekman and geostrophy) are derived using quick scatterometer (QuikSCAT) [Ebuchi et al., 2002] wind and AVISO (Analysis, Validation and Investigation of Satellite Oceanography) sea surface height data [Ducet et al., 2000] respectively. The Ekman current is calculated using the following equation:

  • equation image

where Coriolis force f = 2Ω sin θ, ρo = mixed layer ocean water density. wind stress τx = ρaCDu2 and τy = ρaCDv2, with CD = 1.5 × 10−3, ρa = 1.252 kg m−3. Using sea surface height (H) data, the geostrophic current is calculated using the following equation:

  • equation image

In addition to the cruise data, we have used daily surface layer salinity, temperature and rainfall measured at 15° N, 90° E from RAMA Buoy for the period October 2008 to December 2010 [McPhaden et al., 2009]. Daily rainfall data from GPCP are used in order to identify the large scale monsoon events (http://precip.gsfc.nasa.gov/index.html). In addition to that, high resolution daily rainfall data from 3B42 [Huffman et al., 1995, 2007] for the period 1st June to 31st December 2009 are used.

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References

3.1. Observed Features

[7] Time series location witnessed three rainfall events; first one on 24–25th July, second on 27–28th July and third on 2–5th August (Figure 2a). The cruise left the TSL location on 6th August at noon, therefore rainfall data of that day is not complete. However, rainfall from 3B42 shows that the last event continued till 9th August (figure not shown). Since there are no clear gap between 1st and 2nd event, together they can be considered as an active phase of monsoon intraseasonal oscillation (ISO) with peak around 27–28th July and the second active phase with peak around 5–6th August. Temperature and salinity profile up to 100 m depth and their evolution during 15 days at TSL location are shown in Figures 2b and 2c respectively. Prior to the beginning of rainfall activity, the SST increased to 28.5° C for a brief period of about 2 days (23–24th July, Figure 2a). A rapid decrease in surface layer salinity during 29–30th July is observed. After the initiation of rapid freshening, water temperature at the surface layer (10–15 m) remained always above 28.5° C. Furthermore, from 30th July onward the salinity of top 10–15 m layer maintained at minimum value with a small variability. Variability of salinity at the top of fresh surface layer may be linked with the local rainfall as evident from the rainfall events (red bar in Figure 2a). Fresh surface water also coincides with the warm (≥28.5° C) layer at the top. The average mixed layer salinity (depth) between 22–28th July is 32.4 psu (30.72 m) and the average between 31st July to 6th August is 28.07 psu (6.297 m). Therefore, salinity dropped by more than 4 psu within the 24 hours (Figure 2f). Changes in temperature of mixed layer also took place almost at the same time. The average mixed layer temperature during 22–28th July (31st July–6th August) is 28.45° C (28.81° C). The mixed layer depth (MLD) started shoaling immediately after the first active phase and continued to shoal until end of the cruise (Figure 2d). Similarly the barrier layer thickness (BLT) started to deepen immediately after the first active phase of rainfall. MLD and BLT are calculated following Lukas and Lindstrom [1991]. Similar decrease in sea surface salinity (SSS) was also observed by Bhat [2002] in BOBMEX experiment [Bhat et al., 2001] in its phase 1, where SSS decreased from 33 psu to 29 psu during 27th July to 3rd August 1999 and remained low for the rest observation period. During BOBMEX salinity dropped gradually (4 psu in 7 days), while during CTCZ cruise similar amount of salinity dropped within 24 hours. Therefore, it is very interesting to understand the mechanism behind rapid decrease of salinity during CTCZ cruise.

image

Figure 2. Rainfall, temperature and salinity profile measured on-board ORV Sagar Kanya at TSL. (a) Rainfall in mm day−1 (black line). Rainfall events in every 2 hour interval are observed (rain or no-rain) and marked by redbars. (b) Temperature of top 100 m layer. Isotherm of 28.5° C is marked by black contour. (c) Salinity of top 100 m layer in psu. (d) Mixed layer depth (in meter) and barrier layer thickness (in meter). Time series of MLD averaged (e) temperature and (f) salinity.

Download figure to PowerPoint

3.2. Heat and Salt Budget

[8] The estimated temperature tendency at TSL using GODAS currents are able to capture the observed tendency reasonably well (Figure 3a). The mixed layer temperature is controlled equally by both, advection and net surface heat flux. As from 30th July onward the mixed layer salinity started to fall, the advection of fresh and cold water decreases the temperature. On the other hand, the net surface heat flux (mainly the radiative flux) causes extreme warming, probably due to presence of shallow barrier layer (Figure 2d). Since the mixed layer depth became shallow, the net surface heat flux became very effective in increasing temperature. The salinity budget analysis reveals that advection is the major source of fresh water at TSL and the sudden fall in salinity on 29–30th July is mainly due to advected fresh water (figure 3b). The change in salinity due local fresh water flux (i.e., E-P) is very small. Similar analysis using estimated Ekman and geostrophic current closes the heat and salinity budgets quite well till 29th July and afterward the advection of salinity and temperature is too strong (not shown). The reasons for overestimation of advection due to Ekman and geostrophic currents are not clear. The estimated geostrophic currents derived from sea surface height (from satellites and in-situ data) over estimate the advection component of budget. Using data collected during BOBMEX expedition, Vinayachandran et al. [2002] concluded that wind driven circulation, i.e., Ekman current determines the path of fresh water in the BoB. They further concluded that fresh water advected eastward from the coast of India to the study location at 17.5° N, 89° E. However, during our observational period the average mixed layer current in GODAS reanalysis was always south and south-westward (Figure 3c). On the other hand, the estimated total current were south-westward when the budgets are almost closed (i.e., till 29th July) and flowed south-eastward when the budget is not closed (i.e., after 30th July; figure not shown). The river discharge from the north coast is much higher (Ganga and Brahmaputra) compared to the east coast (Mahanadi, Godavari, Krishna, and Cauvery) [Rao and Sivakumar, 2003]. It is plausible that if advection of fresh water takes place from north, the freshening will be larger compared to cases when advection is from the east coast of India. Therefore, it is evident from the above analysis that, southward current tends to close the heat and salinity budget reasonably well and hence, freshwater may have advected mainly from the north (i.e., discharged from Ganga and Brahmaputra rivers). This can explain the rapid (slow) decrease in salinity during CTCZ (BOBMEX) cruise.

image

Figure 3. Salinity and temperature budget at TSL using heat flux from NCEP, surface currents from GODAS and in-situ measurement of rainfall, salinity and temperature. Observed and estimated (a) temperature tendency, (b) salinity tendency, and (c) top 20 m and time averaged (22nd July–6th August) current from GODAS (in ms−1). Red star mark represent the TSL location.

Download figure to PowerPoint

3.3. Intraseasonal Variability

[9] Using in-situ data at the TSL, it is not possible to know whether the freshening continued throughout the monsoon season with the same level. If the freshening is dominated by river discharge, the monsoon ISO signal may be present in the surface layer salinity with certain time lag and this can modulate the surface layer temperature. Since long term data are not available to address the above, we have selected a close by buoy data for this purpose. RAMA buoy data at 90° E,15° N (south-east of TSL; Figure 1) have longer observation period and indeed it shows oscillation in the surface layer salinity during monsoon period and beyond (Figure 4b). It may be noted that a sudden freshening started at the beginning of October 2009 and continued for the next few months with some variability. However, the local rainfall is quite negligible and hence change in salinity cannot be explained through local fresh water flux (i.e., E-P). Such variability in salinity/SSTare not unique, but also observed in other years (i.e., 2008 and 2010) as evident from available continuous observation for the period October 2008 to December 2010 (Figures 4b and 4c). Simultaneous response in surface layer warming is also evident (Figure 4c). Power spectrum analysis of June-December 2009 salinity and temperature averaged over upper 35 m show significant variance near 13 days and 15 days respectively. Other significant peaks in salinity and temperature are found between 30–60 days period (Figure 5). SST also shows significant peak quite similar to that of 35 m average water temperature (figure not shown). These two preferred bands are similar to the summer monsoon rainfall variability on the intraseasonal time scale (i.e., 10–20 and 30–60 days mode). Similarly two epochs of rainfall during CTCZ cruise are also separated by about 10 days. Power spectrum analysis using GPCP rainfall, averaged around TSL (87°–92° E, 17°–21° N) for 2009 summer monsoon season (JJAS) shows significant power at 12.5 days period (figure not shown). Therefore, the observed rainfall activity belongs to quasi bi-weekly (10–20 days) mode of intraseasonal oscillation, which in general propagates westward [Chatterjee and Goswami, 2004].

image

Figure 4. Rainfall, temperature and salinity profile measured at RAMA buoy location (15° N, 90° E). (a) Rainfall in mm day−1. (b) Salinity of top 100 m layer in psu. (c) Temperature of top 100 m layer (° C). Isotherm of 28.5° C is marked by black contour.

Download figure to PowerPoint

image

Figure 5. Spectra of surface layer (top 35 m average) (a) temperature and (b) salinity using RAMA buoy data at 15° N, 90° E for the time period 1st June to 31st December 2009. Blue and red curve represent 5% and 95% significance level respectively and green curve represents “red noise” level.

Download figure to PowerPoint

[10] In order to find out the maximum rainfall variability region and to relate it with the observed salinity, variance of 10–100 days filtered rainfall anomaly (June-December 2009) are calculated. The heavy rainfall regions due to monsoon ISOs are depicted by large variance (Figure 6). Therefore, there is possibility that fresh water discharge from those regions may modify the salinity either locally or through advection processes. Thus, four regions (D1, D2, D3, D4) are identified based on the larger rainfall variance which may modify the salinity at the location of RAMA buoy on monsoon rainfall ISO time scale (Figure 6). Domain D1 (86°–93° E, 22°–27° N) and D2 (79°–87° E, 18°–23° N) represent the river catchment area of Ganga, Brahmaputra, and Mahanadi. D3 (88°–95° E, 12°–19° N) represent the area over BoB with maximum rainfall ISO variance and the RAMA buoy is located within this box. D4 (88°–92° E, 13°–17° N) represent the smaller domain surrounding the RAMA buoy and it also covers part of maximum rainfall ISO variance region.

image

Figure 6. Variance of 10–100 days filtered rainfall anomaly from 3B42 data (in mm2day−2) of the period 1st June to 31st December 2009.

Download figure to PowerPoint

[11] Power spectrum analysis of rainfall averaged over individual domain shows significant variances at 3–7 days time scale except in D1, where the significant variances are at around 10 days (Figure 7). In addition to that, rainfall at D1 shows significant variability at about 50 days period. Similar analysis of surface layer salinity and temperature (top 35 m average) at RAMA buoy also reveals the prominence of power at the ISO bands (Figure 5). Thus both rainfall over domain D1 and salinity at RAMA buoy has the dominance on the ISO time scale, which imply that rainfall over D1 may affect salinity at RAMA buoy in a delayed mode and at the same time, the local rainfall as well as river discharged water from Indian east coast cannot explain the observed salinity variability. Cross spectrum analysis of rainfall at D1 and salinity for the period June to December 2009 further reveals large coherence at 10 and 50 days time scale and rainfall leads the salinity by about 60 days (Figure 8). Similar analysis using rainfall of other domain shows maximum coherence at synoptic time scale (3–7 days) and hence can't explain the variation of salinity on ISO time scale. To advect fresh water from the TSL to RAMA buoy, the average speed of southward current should be about 8.75 cm s−1, which is within the range of observed southward current. Therefore, it is very likely that the first pulse of observed freshening at TSL may have advected towards south and reached RAMA buoy location on the beginning of October 2009.

image

Figure 7. Spectra of rainfall time series (1st June to 31st December 2009) using daily 3B42 data and averaged over (a) D1 (86°–93° E, 22°–27° N), (b) D2 (79°–87° E, 18°–23° N), (c) D3 (88°–95° E, 12°–19° N), and (d) D4 (88°–92° E, 13°–17° N). Blue and red curve represent 5% and 95% significance level respectively and green curve represents “red noise” level.

Download figure to PowerPoint

image

Figure 8. (a) Coherence and (b) phase angle between rainfall averaged over domain D1 (86°–93° E, 22°–27° N) and salinity at RAMA buoy (15° N, 90° E) for the time period 1st June to 31st December 2009.

Download figure to PowerPoint

[12] To further confirm this possibility in detail we have used Wavelet analysis [Grinsted et al., 2004]. It is a powerful tool giving both the information of the dominant mode of variability in spectral domain and their exact occurrences in time domain. Cross wavelet transform (CWT) has been performed between salinity at RAMA buoy and rainfall averaged over D1 for the time period 1st June 2009 to 28th February 2010. Significant common power on 10–20 days mode are observed twice viz. 15th July to 15th August and 10th September to 20th October (Figure 9a). Phase angle during July-August period suggests that the salinity change was almost simultaneous with the rainfall over D1. This further indicates that rainfall over both the location is from the active phase of ISO, which occurred almost simultaneously over northern BoB as well as over D1. However during September-October, the rainfall and salinity are in opposite phase. Hence, sudden freshening at the beginning of month October is directly related with the rainfall over D1 which took place earlier. The squared wavelet coherence (WC) also shows significant coherence on 10–20 days time scale during September-October (Figure 9b). This indeed shows that sudden freshening at RAMA buoy on the beginning of October is due to rainfall at D1 and the same fresh water is observed first at TSL and subsequently at the buoy location. Significant coherence is also evident on 30–60 days band during month of August to September. Therefore, salinity at RAMA buoy is affected by rainfall at D1 from both mode of ISOs.

image

Figure 9. (a) Wavelet co-spectrum and (b) squared wavelet coherence between rainfall (averaged over D1) and salinity at RAMA buoy (top 35m averaged) for the period June-December 2009. Thick contour shows the area significant at 95% level and arrow indicates the phase angle between these two time series.

Download figure to PowerPoint

[13] In order to bring out a clear picture of lead/lag relation between salinity and rainfall, filtered salinity anomaly at RAMA buoy are linearly regressed with filtered rainfall and current anomalies (estimated) of all other points. Here, 10–100 days and 30–60 days Lanczos [Duchon, 1979] band pass filter are used. Regressed anomalies averaged over 88°–92° E (Figure 10) show a clear southward propagation of fresh water and it takes about 50–60 days to reach RAMA buoy from D1. Rainfall regressed with SST at RAMA also shows similar lead/lag relation (figure not shown). Furthermore, regressed current anomaly shows strong southward component (Figure 11), which implies southward advection of river discharged fresh water through oceanic currents. Lead/lag regressed current also shows periodicity of about 50–60 days (figure not shown). As evident from spectrum analysis, the 30–60 days mode has large variance and as a consequence all the regressed anomalies are stronger on the same ISO band.

image

Figure 10. Lag-latitude section (88°–91° E average) of regressed rainfall with top 35m average salinity at RAMA buoy (1st June–31st December 2009). The (a) 10–100 days filtered anomalies and (b) 30–60 days filtered anomalies.

Download figure to PowerPoint

image

Figure 11. Regressed current anomaly (estimated) with top 35 m average salinity at RAMA buoy (1st June–31st October 2009). The (a) 10–100 days filtered Ekman current, (b) 30–60 days filtered Ekman current, (c) 10–100 days filtered total (Ekman + Geostrophy) current, and (d) 30–60 days filtered total current.

Download figure to PowerPoint

4. Summary and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References

[14] Under the CTCZ program, oceanic and atmospheric observations are taken on-board ORV Sagar Kanya. She started on 15th July 2009 from the port of Chennai and reached the TSL on 22nd July and stayed there up to 6th August. In every 2 hours of interval, salinity and temperature profile up to the depth of 760 m was measured. Total 218 profiles were measured using Idronaut CTD. To estimate the horizontal advection of salinity and temperature, CTD measurement at 4 other locations situated at North, South, East and West of the TSL and at a distance of 3 nautical miles from the TSL are carried out.

[15] A sudden freshening is observed during 29–30th July, where the mixed layer averaged salinity dropped by more than 4 psu within 24 hours time. A simultaneous increase in mixed layer averaged temperature by 0.36° C is also observed. The budget analysis of salinity and heat primarily indicates dominant role of advection processes on temperature and salinity at the northern BoB. As the local rainfall contribution to the salinity change was meager, the freshening can only be explained through the southward Ekman flow which advected river discharged water from the north. Even though previous studies expected that due to strong stratification, heat budget into northern BoB can be considered as one dimensional problem, we have presented an evidence that advection also plays a major role.

[16] Daily salinity and temperature data (averaged over top 35m) at south-east of TSL from RAMA (90° E, 15° N) reveals significant variance at time scale which falls within the 10–20 days bands of monsoon ISO. Again, the local rainfall is not large enough to explain the salinity change. Among the different catchment area, only rainfall at Ganga-Brahmaputra catchment areas shows maximum coherence at 10 and 50 days time scale with lag of about 60 days. Wavelet co-spectrum and coherence analysis of above two time series further reveals significant power and maximum resonance respectively at 10–20, and 30–60 days time scale during September-October. Anti phase relation between time series of salinity at RAMA location and rainfall over Ganga-Brahmaputra river catchment area further suggests that intraseasonal variation of salinity is mainly due to river runoff and the same varies with the monsoon rainfall ISO over river catchment areas. Regressed rainfall, oceanic current with salinity and temperature at RAMA buoy illustrate that, the southward advection of fresh water from the Ganga-Brahmaputra river catchment area influences salinity and temperature at RAMA on ISO time scale.

[17] This study aids in further our understanding of ocean advection, playing a crucial role in forming barrier layer and thus modifying SST significantly. Thus providing a different perspective in terms of lagged response of salinity which dictates the convective activity over BoB during the withdrawal and post monsoon phases. Most of the climate models prescribe fixed amount of river discharge. As a result models may not be able to capture the post-monsoon SST/SSS variability merely due to the lagged response of rainfall over Ganga-Brahmaputra catchment area and hence may affect the genesis of post-monsoon cyclone over BoB. As of now, daily fresh water discharge data from Ganga and Brahmaputra are not available. However, such data will be very important for further understanding and eventually may improve the post-monsoon rainfall predictability in coupled models.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References

[18] The authors acknowledge the financial support from the Department of Science and Technology (DST) under the research project “Oceanographic Observations in the northern Bay of Bengal deep convection during CTCZ.” Freeware Ferret and Grads is used extensively in this study. The GPCP 1° daily precipitation data set were provided by NASA/Goddard Space Flight Center's Laboratory for Atmospheres, which develops and computes the 1° data as a contribution to the GEWEX Global Precipitation Climatology Project. We thank three anonymous reviewers for their constructive comments which helped to improve this manuscript. We wish to express our sincere thanks to D. Shankar, NIO, Goa, for providing support during the cruise SK-261.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Results
  6. 4. Summary and Conclusions
  7. Acknowledgments
  8. References