It is well known that the seasonally reversing monsoon winds affect the circulation of the Bay of Bengal (BoB) significantly [Cutler and Swallow, 1984; Hastenrath and Greischar, 1991; McCreary et al., 1993; Schott et al., 2009, S2009, hereafter; Durand et al., 2009]. An extensive series of surveys was conducted in the Bay during 1983–1991 covering postmonsoon (October–November 1983), monsoon (June–August 1984), and spring transition (March–April 1987 and 1991). Seasonal and interannual circulation variabilities were studied by many, including Babu et al. , Murty et al. [1992, 1993], Suryanarayana et al. , Shetye et al. , and Somayajulu et al. . The seasonality is well established. An anticyclonic circulation evolves during the spring transition (or intermonsoon) period, which collapses with the intensification of the southwest monsoon in May or early June. Subsequently, a large cyclonic gyre occupies the central bay south of 15°N during the southwest monsoon (June–September). During October through December, the western boundary current (WBC) flows equatorward [Shetye et al., 1996]. A recent overview by S2009 describes many aspects of this unique circulation system. The seasonality of the WBC in the bay is described by Durand et al. [2008, 2009] on the basis of satellite altimetry.
 This study focuses on the structure and variability of the springtime western boundary current, which is set up mainly in response to the basin-wide anticyclonic gyre during January through March. This western boundary current flows northward along the coast during spring in contrast to the southward flowing East India Coastal Current (EICC) during postmonsoon and early winter. In many studies [e.g., Shetye et al., 1993, 1996; Durand et al., 2008, 2009, and others cited above] the EICC has been used as a common term to describe both phases of the boundary current (northward during spring and southward during fall-winter). Here we distinguish between the two phases and refer only to the northward-flowing current as the springtime WBC, which exists and evolves during February–May, preserving the EICC to the southward-flowing phase, consistent with its more “coastal” character [Shetye et al., 1996].
 The motivation for the distinction derives from the following rationale. In a classical Sverdrupian sense, the springtime response might be similar to the formation of the major western boundary currents such as the Gulf Stream, the Kuroshio, the Brazil Current and the East Australian Current. Typically for the BoB, which has an effective width of about 1000 km between Andaman and Nicobar Islands and the east coast of India (10°N–18°N), the phase speed of the first baroclinic Rossby wave is on the order of 8–12 cm/s [see Gill, 1982, Figure 12.3]. Thus, it would take about three months for these waves to cross the Bay. Therefore, it is reasonable to expect the formation of a springtime WBC due to the anticyclonic wind gyre starting in November and continuing through April–May, by Ekman pumping or integrated wind-stress curl forcing [S2009; Durand et al., 2009]. Here we focus on the nature of this current during March and April before the onset of the southwest monsoon. In other words, we are looking for the answer to the following simple question: “Is there a continuous, meandering springtime WBC in the Bay of Bengal?” We show from data analysis that the answer is “yes.” The potential vorticity and dynamic height fields show that the WBC is a continuous flow along the western boundary, eventually separating from the coast at around 18°N and flowing eastward.
Potential Vorticity Analysis of the Babu et al.  Data
 Based on AVHRR satellite observations, Legeckis  indicated the existence of a continuous western boundary current in the western BoB during March–April (see their plate #1). Babu et al.  (B2003, hereafter) analyzed the temperature, salinity and velocity data in the region of 12°N–20.5°N during the spring intermonsoon period of 22 March to 28 April 1987. The upper 1000 m of temperature and salinity data were further geostrophically adjusted to analyze the velocity structure of the WBC at a few sections (see their Figure 1). Based on these derived velocity fields and discrete meridional transport calculations, they concluded that (i) the WBC is a northward flowing current during the spring intermonsoonal period when the ACG on the Central Bay was two anticyclonic cells, and (ii) the anticyclonic cells merge into one in April, when the Curl τ intensifies. B2003 also identified two cyclonic eddies CE1 and CE2 to the inshore of the meandering northward WBC and noted that the WBC turns eastward at around 17.5°N and continues as a straight jet at the edge of the ACG in the Bay.
 In this study, we extend the analysis of the B2003 hydrographic data set in terms of potential vorticity computation. Potential vorticity (PV) was used by Bower et al. , Qiu and Miao , and Jochum and Malanotte-Rizzoli  as a conservative tracer along typical western boundary currents like the Gulf Stream, the Kuroshio, and the Brazil Current. Bower et al.  was one of the early studies to identify the Gulf Stream as a continuous current from hydrographic data. For a continuously stratified fluid, the PV is given by (f + ζ) × E (equation (1)), where f is the Coriolis parameter and ζ is the relative vorticity. E, the static stability, is defined by –ρ–1 δρ/δz. This form is known as the Ertel's PV [Gill, 1982], and is conserved following the motion in a stratified fluid. Thus, the presence of an isoline of the Ertel's PV (EPV) can be interpreted as the signature of continuity of a meandering current system such as the WBC. Furthermore, a cross-stream “jump” in the EPV is a signature of a strong current, called “barrier” by Bower et al. , and the uniform PV region within the current is called the “blending” region.
 As indicated by B2003, the surface signature of the current is masked by isothermal layers; so, we focus our analysis on the 50 m depth. Figures 1a and 1b show the objectively analyzed temperature and EPV fields from the B2003 data set. The multiscale objective analysis (OA) procedure was described by Lozano et al. . The observational stations from B2003 are shown within the OA domain (Figure 1), which extends from 81°E, 12.5°N to 87.5°E, 19.5°N with a 25 km grid. A spatial decorrelation scale of 100 km with a zero-crossing at 150 km was used in both zonal and longitudinal directions. The mean value of all of the observations was used as the background for the multiscale OA.
 The sharpest EPV gradients denote the axis of the boundary current. The PV contours of 0.2–0.3 (m–1s–1) between the two high pools of EPV is identifiable as the core of the WBC in March 1987. The EPV sections across 14°N (Figure 2b), across 16°N (Figure 2a) and across 85.5°E (Figure 2c) are presented. The WBC core can be identified near 84°E at 14°N, near 83°E at 16°N and flowing eastward along 18°N (in the section for 85.5°E; also see Figure 1b). The stream separates from the western boundary between the 2500 and 3000 m isobaths, where the topographic steering vorticity (f/H, where f is the Coriolis parameter and H is the ocean depth) is in the range of 1.6 to 1.8 × 10–8 m–1s–1. Note that the temperature field at 50 m has a clear signature of separation (as seen in Figure 2 of B2003). The similarity of the EPV contours in Figure 1b after separation to the superimposed envelope of f/H (designated by the 1.6 and 1.8 10–8 m–1s–1 contours) indicates the possibility of combined effects of baroclinicity and topography [Holland, 1972] being in play for controlling the separation and path thereafter of this WBC.
Altimetric View of the Springtime WBC during 2000–2010
 To examine the year-to-year changes in the structure of the spring WBC, we next turn our attention to dynamic height from satellite altimetry from 2000 through 2010. Figure 3a shows the mean (2000–2010) dynamic height field and corresponding geostrophic currents from AVISOanalysis (http://www.aviso.oceanobs.com/ducas). Evidently, in the mean, the WBC extends from the southern BoB (about 8°N) to 17°N–18°N, with two anticyclonic eddies offshore of the WBC. This picture is consistent with the quasi-stationary anticyclones (the spring-summer phase of the EICC) obtained from sea level anomaly analysis by Durand et al. . The altimetry derived maps for April during 2000–2010 (Figure 4) show that the WBC consistently separates at around 18°N for all the years during 2000–2010, while exhibiting interannual variability in its eddy-meandering pattern between 12°N and 17°N. The two-eddy pattern is dominant in the years 2000, 2002, 2003, 2005, 2007, and 2009. The interannual variations include: (i) absence of one of the anticyclones (2004, 2008), (ii) occurrence of a single large anticyclonic cell (2001, 2006), and (iii) occasional existence of a cyclonic eddy at different latitudes inshore of the WBC (2000, 2005, 2008), similar to those reported in observations by Shetye et al.  (16°N) and by Sanilkumar et al.  (13°/14°N). Because of the dominance of the eddy activity in some of these years in the 12°N–16°N region, there may be an apparent absence of the WBC in the southern BoB.
 The WBC is thought to be forced by negative wind stress curl in the BoB and remote wind stress forcing in the equatorial Indian Ocean. We computed wind stress curl from QuikSCAT scatterometer surface winds in the spring season (February–March–April) of different years from 2000 through 2007. The February–April mean wind stress curl (in units of 10–7 N/m3) integrated over the BoB (80°E–98°E) is presented in Figure 3b for the individual years. This year-by-year variability clearly demonstrates two aspects of the wind-curl related WBC dynamics in the BoB: (i) the separation and eastward flow regime of the WBC is where the integrated curl reaches its maximum negative value (16°N–18°N) and decreases rapidly thereafter; and (ii) there is considerable interannual variability to the south of 16°N in the wind-stress curl forcing of the WBC. Whether the southern curl distribution covaries with the separation and the eastward flow is an interesting question that needs further modeling and observational studies. The integrated wind-stress curl is somewhat weaker in 2000 (−0.72) or in 2001 (−0.96) compared to that in 2004 (−1.27) or in 2005 (−1.3). It is thus possible that the overall strength of the WBC does respond to the integrated wind forcing.
 The eddy structure of the spring WBC, however, also seems to depend on the flow field in the southern BoB in the winter season. For example, in January 2000 and 2001 (Figure 5), a large cyclonic eddy occupies the southwestern Bay, which is associated with equatorward flow along the coast. This cyclonic circulation might be responsible for the weaker WBC in the south in the years 2000 and 2001, which was also observed in the NIO hydrographic data set (B2003). It is clear that the structure of the spring WBC depends on the circulation in the southern BoB in the preceding winter; in other words, it is a function of the history of local and remote wind stress forcing over at least one season.