An inverse modeling study in Fram Strait. Part I: dynamics and circulation

https://doi.org/10.1016/S0967-0645(99)00018-1Get rights and content

Abstract

In order to reconstruct the circulation in the northern Greenland Sea, between 77°N and 81°N, and the exchanges with the Arctic Ocean through Fram Strait, a variational inverse model is applied to the density field observed in summer 1984 during the MIZEX 84 experiment. An estimate of the three-dimensional large-scale pressure field is obtained in which the solution is decomposed into a limited number of vertical modes and the mode amplitudes are described by piece-wise polynomials on a finite-element grid. The solution should be consistent with a frictional depth-integrated vorticity balance and with the density data. The global model parameters are tuned to ensure agreement between the retrieved geostrophic velocity and independent currentmeter data. In a companion paper (Schlichtholz and Houssais, 1999b), the same method, but without dynamical constraint, is applied to the same hydrographic dataset to perform a detailed water mass analysis and to estimate individual water mass transports.

A comprehensive picture of the summer geostrophic circulation in Fram Strait is obtained in which northward recirculations in the East Greenland Current (EGC) and various recirculations from the West Spitsbergen Current (WSC) to the EGC are identified. It is suggested that the branch of the WSC following the upper western slope of the Yermak Plateau turns westward beyond 81°N and recirculates southward along the lower slope, then merging with a westward recirculating branch south of 79°N. At 79°N, a southward net transport of 6.5 Sv is found in the EGC which, combined with a northward net transport of only 1.5 Sv in the WSC, results in a fairly large outflow of 5 Sv from the Arctic Ocean to the Greenland Sea.

The inverse solutions show that, in summer, the local induction of vorticity by the wind stress curl or by meridional advection of planetary vorticity should be small, so that, in the EGC and in the WSC, the vorticity balance is mainly achieved between the bottom pressure torque and dissipation of vorticity through bottom friction. A substantial barotropic flow associated with along-slope potential energy gradients is indeed identified on both sides of the strait.

Introduction

The seas north of the Greenland–Scotland Ridge constitute a major heat sink in the global thermohaline circulation of the world ocean and therefore a crucial component of the earth climate (Aagaard et al., 1985). A large heat loss to the atmosphere, combined with sea ice production and melting, is responsible for the formation of deep and intermediate waters through winter convection, which, in some basins like the Greenland Sea, may reach down to the bottom. The one-dimensional convection process is strongly affected by the horizontal dynamics of the convective basins including lateral exchanges of heat, salt and ice at their boundaries. Concerning the Greenland Sea, important exchanges occur through its northern boundary, the Fram Strait, which is the only deep connection between the Arctic Ocean and the rest of the world ocean. A better knowledge of the dynamics of the strait is therefore crucial in the context of investigating the processes involved in the deep convection and their variability.

Due to the severe conditions, field experiments in Fram Strait are limited, and the MIZEX 84 hydrographic data (Johannessen, 1987), collected between 77°N and 81°N from the edge of the Greenland Shelf to the Spitsbergen coast (Fig. 1), constitute the largest data set ever collected in the area. The data set offers a unique possibility to obtain a quasi-synoptic description of the dynamics of the strait. The present study, which is presented in two companion papers, is an attempt to estimate the large-scale time-mean circulation and the water mass distribution of the area. The former, which is the focus of the present paper, is obtained from an inverse model which allows one to deduce the geostrophic flow from the observed density field and to discuss some dynamical aspects of this flow. The latter is presented in a second paper (Schlichtholz and Houssais, hereafter SH2, 1999b) and is obtained through interpolation of the observed temperature and salinity fields. These are further combined with the recontructed circulation discussed below to provide a description of the individual transports associated with each of the water masses present in the strait.

To the first order, the circulation in Fram Strait is made of an inflow and an outflow. On the eastern side of the strait, the northward West Spitsbergen Current (WSC) carries relatively warm and salty waters of Atlantic origin above relatively cold and fresh deep waters formed in the Greenland and the Norwegian Seas. On the western side, sea ice and a cold and fresh surface water are exported from the Arctic Ocean in the East Greenland Current (EGC) above relatively warm and salty deep waters. The detailed flow pattern in the strait, however, is more complex and numerous recirculations with large spatial and temporal variability (Hopkins, 1991) make reliable estimates of the transports through the strait more difficult.

Previous estimates of the transports through Fram Strait based on hydrographic measurements mostly rely on the baroclinic component of the geostrophic currents (e.g., Timofeyev, 1962). Direct current measurements of the total (baroclinic+barotropic) flow are extremely sparse and contaminated by the mesoscale activity (Foldvik et al., 1988). Lagrangian observations can give insight into the circulation (Gascard et al., 1995) but are unable to provide transport estimates. Numerical models, which include the northern Greenland Sea, either have too a coarse resolution in view of the complex bottom topography of the strait (e.g., Gerdes and Schauer, 1997) or, considering a restricted domain, use Fram Strait as an open boundary so that currents or transports cannot be reasonably predicted in the strait (e.g., Legutke, 1991). Inverse methods combining hydrographic information with relevant constraints offer an alternative approach to estimate the total large-scale time-mean flow. Moreover, when dynamical constraints are used, the assumed dynamics can be validated through analysis of the estimated flow. In Fram Strait, however, previous inverse studies have been based on global conservation constraints rather than on local dynamical constraints so that the three-dimensional (3D) distribution of the velocity field could not be reconstructed. Instead, the velocity through a scant number of hydrographic sections was used to provide estimates of the transports through the strait (Rudels, 1987; Houssais et al., 1995).

Since the mechanisms driving the circulation in Fram Strait are largely unknown, it is interesting to validate the dynamical assumptions constraining the flow. For instance, while the WSC can be considered as the northernmost extension of the North Atlantic Current system (e.g., Hopkins, 1991), the East Greenland Current (EGC) may be more influenced by local conditions in the strait, being either driven by density differences between the Arctic Ocean and the Greenland Sea (e.g., Hunkins and Whitehead, 1992) or forming the western branch of the wind-driven Greenland Sea gyre (Aagaard, 1970). According to direct current measurements, both the EGC and the WSC are partially barotropic (e.g., Aagaard et al., 1973; Foldvik et al., 1988). Earlier hydrographic analysis (e.g., Quadfasel et al., 1987) have suggested the prominent role of the bottom topography while the local wind in the strait is likely to exert little influence on the large-scale circulation in summer (e.g., Hunkins and Whitehead, 1992). More specifically, Schlichtholz and Houssais (1999a) have recently proposed that a substantial part of the EGC transport in Fram Strait in summer may be due to the JEBAR (joint effect of baroclinicity and relief).

Although the present inverse model is of the same type as the Provost and Salmon (1986) model, it contains several methodological improvements to take into account the specificity of Fram Strait. First, instead of using polynomials to describe the vertical structure of the solution, we use cosine-logarithmic functions that better fit the density distribution in Fram Strait. The function parameters are adjusted by fitting the density anomalies relative to a reference profile rather than the density itself. Secondly, local (instead of uniform) weights are assigned to the original density data, which are deduced from an objective analysis (OA) of these data and are intended to penalize the less statistically reliable data. Thirdly, some OA data are used as an additional data set in order to handle the problem of uneven data distribution. Fourthly, the smoothing is applied to the density, and not only to the pressure, which ensures the smoothness of the a posteriori density distribution and of the vertical velocity shear. In addition, the pressure is smoothed at the bottom rather than in the entire 3D space to prevent spurious countercurrents from emerging in the water column. Finally, the bottom topography is interpolated on the finite-element grid using a 2D analog of the inverse method which allows us to analyse the impact of different degrees of topography smoothness on the 3D solution.

Section snippets

The hydrographic data distribution and the model grid

From a set of more than 1400 CTD casts taken in Fram Strait during the MIZEX 84 experiment, 342 casts have been selected for the present analysis, most of them from the end of May to mid-July and some of them during a post MIZEX cruise in August. Different platforms were operating quasi simultaneously so that the dataset is considered synoptic for scales ranging between 50 and 500 km, i.e. larger that the typical scale of the transients. The stations are distributed over a domain extending from

The dynamics

Given discrete hydrographic measurements and wind stress and bottom topography fields in a limited ocean domain, we seek a large-scale time-mean 3D horizontal velocity field u in that domain. We postulate that outside the boundary layers the Rossby and Ekman numbers are small and the large-scale time-mean circulation is in geostrophic balance. The flow also satisfies the hydrostatic relation and mass conservation. Therefore, the equations describing the interior flow can be written asẑ×fu=−1ρ0

Determination of model parameters and misfits analysis

The choice of the dimensions of the inverse problem (22), N and M, and of the local weights wi's and wj's appearing in Eq. (15) has been made prior to the 3D experiments (Section 4.1). Our model also contains six global parameters that are the interpolation parameters OA,δb), the global weighting factors (α,η) and the parameters relevant for the vorticity constraint (r,αtopo). These parameters are adjustable, therefore guaranteeing the possibility for the model to be adapted to different

Sensitivity of the geostrophic flow pattern to global model parameters

In this section, the a posteriori decomposition of the depth-averaged geostrophic velocity into the relative and bottom components, ūr and ub, is used to compare the sensitivity of these two components to changes in the parameters η,α and αtopo. Five sensitivity experiments (p1p5), in addition to the reference experiment p0, are particularly interesting.

The relative velocity is determined by the density field. The particular parameter choice, δOA=1 and δb=1, for the reference experiment

The geostrophic flow: horizontal and vertical velocity distributions

In this section, the main current patterns based on the horizontal distribution of the depth-averaged flow are briefly described. Then, the currents are analysed in more detail using vertical velocity sections at selected latitudes across the strait (Fig. 11) and horizontal velocity maps in three different layers (0–200, 200–700, 700 m–bottom) (Fig. 12). The boundaries of the layers approximately coincide with the boundaries between the surface, intermediate and deep water masses as discussed in

The volume transports

The different components of the horizontal transport are plotted in Fig. 14 for solution p5. Clearly, the total transport is dominated by the bottom (Fig. 14a) and the relative (Fig. 14b) components of the geostrophic transport. The bottom Ekman transport (Fig. 14c) represents, at the most, a few percents of the geostrophic transport and the surface Ekman transport (Fig. 14d) is even smaller.

The meridional geostrophic transports for solution p5, cumulated eastward from the Greenland Slope along

Discussion

The six selected solutions (p0p5) discussed in the preceding sections are consistent with the data and with the vorticity dynamics to within reasonable misfits. A common feature to most of these solutions is the presence of a relatively strong barotropic component. In the remaining of this section, we discuss the possible origin of this component and then analyse how realistic the flow is in view of results reported in the literature.

Summary

An inverse model has been used to reconstruct the large-scale geostrophic velocity field in Fram Strait in summer 1984 from the MIZEX 84 hydrographic data. The reconstructed flow provides us with the first comprehensive 3D circulation scheme in Fram Strait. In addition to being consistent with the density data and with validated dynamical assumptions, the flow also agrees within a reasonable misfit with some independent current meter measurements.

Our results reproduce the known features of the

Acknowledgements

This work was part of the ESOP-1 project supported by the MAST II programme of the Commission of the European Communities (contract no. MAS2-CT93-0057). The authors would like to thank Dr. Slim Gana for stimulating discussions on the variational inverse method.

References (35)

  • K. Aagaard et al.

    Thermohaline circulation in the Arctic Mediterranean Seas

    Journal of Geophysical Research

    (1985)
  • R.H. Bourke et al.

    Circulation and water masses of the East Greenland Shelf

    Journal of Geophysical Research

    (1987)
  • R.H. Bourke et al.

    The westward turning branch of the West Spitsbergen Current

    Journal of Geophysical Research

    (1988)
  • K.L. Denman et al.

    Correlation scales, objective mapping and statistical test of geostrophy over the continental shelf

    Journal of Marine Research

    (1985)
  • J.-C. Gascard et al.

    Diagnostic study of the Fram Strait Marginal Ice Zone during summer from 1983 and 1984 Marginal Ice Zone Experiment lagrangian observations

    Journal of Geophysical Research

    (1988)
  • J.-C. Gascard et al.

    New insights on large-scale oceanography in Fram Straitthe West Spitsbergen Current

  • R. Gerdes et al.

    Large-scale circulation and water mass distribution in the Arctic Ocean from model results and observations

    Journal of Geophysical Research

    (1997)
  • Cited by (40)

    • The Physical Oceanography of the Arctic Mediterranean Sea: Explorations, Observations, Interpretations

      2021, The Physical Oceanography of the Arctic Mediterranean Sea: Explorations, Observations, Interpretations
    • Pleistocene iceberg dynamics on the west Svalbard margin: Evidence from bathymetric and sub-bottom profiler data

      2017, Quaternary Science Reviews
      Citation Excerpt :

      The West Spitsbergen Current (WSC) flows northwards along the western margin of Spitsbergen, transporting relatively warm Atlantic Water into the Arctic Ocean (Fig. 2a). To the northwest of Svalbard (c. 80°N), the WSC splits into three branches: the North Spitsbergen Current (NSC) is present where the upper 500 m of surface waters are deflected east by the Coriolis Force to flow to the north of Svalbard; the Yermak Slope Current (YSC) occurs when the remaining deeper waters of the WSC continue to flow north and then east around the northwestern corner of the Yermak Plateau; and the Return Atlantic Current (RAC) is located when the western branch of the WSC turns counterclockwise, eventually returning southward along the eastern edge of the East Greenland Current (Figs. 1–2) (Manley et al., 1992; Schlichtolz and Houssais, 1999). The WSC is strongly steered by bathymetry, with current velocities along the western Svalbard margin measured at 9–16 cm/s between 500 and 1500 m depth, whereas the YSC is slower at 1–3 cm/s (Schlichtolz and Houssais, 1999; Fahrbach et al., 2001).

    • Changes in current patterns in the Fram Strait at the Pliocene/Pleistocene boundary

      2014, Quaternary Science Reviews
      Citation Excerpt :

      At around 82°N, the plateau exhibits a roughly SE–NW trending bedrock sill that functions as a large obstacle and likely reduces the current speed. Current velocities decrease significantly along the pathway of the northbound currents in the Fram Strait: Fahrbach et al. (2001) report velocities of up to 24 cm s−1 in the near bottom layer of the core of the West Spitsbergen Current at 79°N, and (Schlichtholz and Houssais, 1999) modeled velocities as low as 1–3 cm s−1 for the Yermak Branch (named Yermak Slope Current in their study). A gradient within the current speeds with a decreasing trend towards north could also be responsible for a discrepancy in the sediment patterns encountered between the southern and northern part of the Fram Strait.

    • Changes in the properties and distribution of the intermediate and deep waters in the Fram Strait

      2012, Progress in Oceanography
      Citation Excerpt :

      Some of the NSDW participates in the cyclonic circulation in the Boreas Basin south of the sill (Schlichtholz and Houssais, 1999), while the rest of the NSDW enters the Fram Strait. A considerable fraction of this recirculates north of the Molloy Deep (Fig. 2), while only a small portion continues its northward course into the Arctic Ocean along the western slope of the Yermak Plateau (Schlichtholz and Houssais, 1999). The NSDW has been followed to the northern side of the Yermak Plateau in 1991 (Jones et al., 1995) and into the Sofia Deep in 1997 (Rudels et al., 2000).

    View all citing articles on Scopus
    1

    Present address: Laboratoire d'Océanographie Dynamique et de Climatologie, Paris, France.

    View full text