Spatiotemporal variations of the d 18 O–salinity relation in the northern Indian Ocean Arvind Singh n , R.A. Jani, R. Ramesh Physical Research Laboratory, Navrangpura, Ahmedabad 380 009, India article info Article history: Received 5 August 2009 Received in revised form 3 August 2010 Accepted 9 August 2010 Available online 14 August 2010 Keywords: Salinity Stable isotope Monsoon rainfall Northern Indian Ocean abstract A new data set of oxygen isotopic composition (d 18 O) and salinity (S) of surface and sub-surface waters of the northern Indian Ocean, collected during the period 1987–2009, is presented. While the results are consistent with positive P E (excess of precipitation over evaporation) over the Bay of Bengal and negative P E over the eastern Arabian Sea, a significant spatiotemporal variability in the slope (also intercept) of the d 18 O–S relation is observed in the Bay; the temporal variability is difficult to discern in the Arabian Sea. The slope and intercept are positively and negatively correlated, respectively, with the annual rainfall over India, a rough measure of river runoff. Both the slope and intercept appear to be sensitive to rainfall; the slope (intercept) is higher (lower) during years of stronger monsoon. The observed variability in the d 18 O–S relation implies that caution needs to be exercised in paleosalinity estimations, especially from the Bay of Bengal, based on d 18 O of marine organisms. & 2010 Elsevier Ltd. All rights reserved. 1. Introduction Oxygen isotopic composition (d 18 O) and salinity (S) of the surface ocean are useful in the study of mixing of oceanic water with runoff, sea ice melting, brine release caused by sea ice formation, and even large-scale ocean mixing (Ferronsky and Brezgunov, 1989; Bigg and Rohling, 2000). They are known to covary linearly in the surface ocean in the mid- and high- latitudes, and to a lesser degree, in the tropics (Craig and Gordon, 1965). This linear relation is exploited in paleoceanographic reconstructions: d 18 O of foraminifera from sediment cores is often used to infer past changes in d 18 O of ocean surface (e.g., Duplessy et al., 1981), after correcting for temperature effects, either using alkenones (e.g., Rostek et al., 1993) or Mg/Ca ratios (Rashid et al., 2007). The tacit assumption here is that the spatial d 18 O–S relation observed today is equally valid for the past, the validity of which has been questioned by Rohling and Bigg (1998). Evapora- tion (E), precipitation (P), continental runoff (R), upwelling/ advection and diffusion are the major physical processes that determine the d 18 O–S relation in tropical oceanic surface waters; d 18 O and salinity are known to (i) increase with evaporation, (ii) decrease with higher precipitation and continental runoff and (iii) vary by mixing due to advection and diffusion. Thus the slope and intercept of the d 18 O–S relation might vary seasonally and geographically, as the controlling processes are season-, climate- and location-dependent (Benway and Mix, 2004). The northern Indian Ocean is an ideal region for investigating such variations as it receives monsoon runoff through rivers of Himalayan and peninsular origin. Published d 18 O data pertaining to the northern Indian Ocean are too limited to enable detection of seasonal and interannual variations in the d 18 O–S relation, if there are any. Paul et al. (1999) inferred from 15 GEOSECS data pairs that the d 18 O–S relation for the tropical Indian Ocean as a whole was d 18 O ¼ 0.18S–5.74. LeGrande and Schmidt (2006) estimated a slope of 0.16 for the Indian Ocean. A compilation of the available data in 2001 (e.g., Table 1 of Delaygue et al., 2001) showed 112 pairs of d 18 O–S data from the Arabian Sea (including 7 from GEOSECS), some of which suffered from systematic shifts and were not used by them for analysis. Delaygue et al. (2001) added a set of 78 additional data pairs, mostly from the open Bay of Bengal, with only  6 samples representing the coastal Bay. Most of their Arabian Sea samples derived from the Gulf of Aden, which prevented them from deciding whether the observed d 18 O–S relation (d 18 O ¼ 0.27S–9.2) applied to the Arabian Sea as a whole, on the basis of latitudinal similarity, or if the Gulf of Aden acted only as an end member, with the Arabian Sea having quite a different d 18 O–S relation. They also claimed that a simple box model that ignored mixing between boxes could successfully explain their observations. Somayajulu et al. (2002) reported 5 data points from the northern Bay of Bengal, which showed a relation: d 18 O ¼ 0.23S–7.6. Here we report 152 new data pairs for surface water (0–2 m depth), 43 for sub-surface (50 m) and 16 for deep (200–700 m) samples collected during eight different cruises conducted in the northern Indian Ocean during 1987–2009 and discuss the seasonal and spatial variations of the d 18 O–S relation and its implication for Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/dsri Deep-Sea Research I 0967-0637/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.dsr.2010.08.002 n Corresponding author. Tel.: + 91 79 2631 4263; fax: + 91 79 2631 4900. E-mail address: arvinds@prl.res.in (A. Singh). Deep-Sea Research I 57 (2010) 1422–1431