Enhanced semidiurnal-band velocity shear across the shelf halocline layer (SHL) was found during land-fast ice edge mooring-based acoustic Doppler current profiler (ADCP) and conductivity-temperature-depth (CTD) observations over the eastern Laptev Sea shelf (∼74°N, 128°E) in April–May 2008 and April 2009. In 2008, the major axis amplitude for the lunar semidiurnal M2tidal ellipses demonstrated intermediate maximum in the SHL at 11–13 m (15 ± 3 cm/s), gradually decreasing to subice and near-bottom layers to ∼9 ± 3 cm/s (at 7 m) and 7 ± 2 cm/s (at 19 m), respectively. In 2009, the semidiurnal tidal flow exhibited similar patterns, but velocities were reduced by about factor of 2. Our estimates of gradient Richardson numbers suggest that the velocity shear associated with semidiurnal baroclinic tidal flow may be strong enough to play a role in water mass modification, promoting shear instabilities, turbulence, and vertical mixing of seawater properties across the SHL. This suggestion is consistent with near-homogeneous water layers episodically occurring in the SHL. Differences in the background stratification and local tidal dynamics between 2008 and 2009, together with rapid responses of the semidiurnal motion to polynya openings, suggest that the baroclinic tide is locally generated.
 The Laptev Sea shelf halocline layer (SHL) represents a transition and critical buffer between the fresher surface layer and the saltier bottom layer beneath. The SHL is vertically stratified in salinity, and the associated density gradient suppresses vertical fluxes across the SHL. In fact, the SHL represents a barrier between the fresh surface layer and the saltier waters below that suppresses vertical mixing throughout the year. Because of the strong salinity dependence of density at low temperatures, the halocline is also a marked pycnocline [Aagaard et al., 1981]. Generally, the SHL occupies the depths beneath the surface mixed layer from ∼10 m down to the depth of the bottom mixed layer, which usually rises ∼5–7 m above the seafloor. In this paper, the SHL is defined as the stratified midwater layer with vertical salinity gradient exceeding 0.2 psu/m interlaid between the subsurface and near-bottom layers.
 The SHL is controlled by wind forcing, Siberian river discharge, thermodynamic ice formation and melting, brine rejection in coastal polynyas, and sea ice and water exchange with the Arctic Ocean and adjoining seas. During summer, the SHL over the eastern Laptev Sea shelf is fed from above by a combination of (i) summer riverine water spreading over the sea surface and (ii) surface meltwater formed due to seasonal sea ice melting. It is also impacted by wind-forced vertical mixing and solar radiation heating. The SHL is maintained from below by the on-shelf inflow of the saline water from the adjacent Arctic Ocean. The strength of the summer SHL depends on the river runoff diversion rather than on the bottom water dynamics. The salinity standard deviations for the surface layer exceed those for the bottom layer at least by a factor of 2 [Dmitrenko et al., 2008a]. The pathways of summer riverine water are primarily wind driven [e.g., Steele and Ermold, 2004; Dmitrenko et al., 2005a] with approximately 500 km3 of freshwater migrating from the eastern Siberian shelf to the Arctic Ocean through the northeastern Laptev Sea during summer when the vorticity of the atmosphere on the eastern Siberian shelves is anticyclonic [Dmitrenko et al., 2008a]. When the vorticity of the atmosphere is cyclonic, wind-forced circulation diverts more riverine waters onto the shelf of the East Siberian Sea, resulting in a weakening of the vertical salinity (density) stratification over the central and northeastern Laptev Sea shelf. Moreover,Dmitrenko et al. [2010c] recently demonstrated that summer patterns in surface hydrography can be maintained through the entire winter season from September to March, while the winter salinity field represents the remnants of the summer field modified by seasonal sea ice formation.
 During winter the Laptev Sea continental shelf is also known for active water mass transformation due to seasonal thermodynamic sea ice formation that provides a strong salt input through brine release [Dmitrenko et al., 2008b, 2009]. In addition, the offshore components of winter surface wind forcing over the Eurasian coast create persistent areas of open water and young ice downwind of the land-fast ice [Martin and Cavalieri, 1989; Bareiss and Görgen, 2005]. The extensive stretches of open water (up to 200 km wide), known as the Great Siberian Polynya, combined with extremely low air temperatures, significantly enhance thermodynamic ice formation and salt flux to the underlying shelf water [Dmitrenko et al., 2010a]. All these processes give rise to seasonal salinity changes [Dmitrenko et al., 2008b, 2009], and substantially weaken vertical salinity (density) stratification. However, the ice formation is not believed to be critical in eroding preexisting SHL stratification on a regular basis, and the summer SHL over the eastern Laptev Sea is usually strong enough to persist through winter [Dmitrenko et al., 2005b; Krumpen et al., 2011]. For the eastern Laptev Sea shelf, the long-term mean probability for winter convective mixing down to the seafloor is ∼20% [Dmitrenko et al., 2005b]. In general, over the shelf area deeper than 15 m, the SHL suppresses vertical mixing year-round, consequently damping property exchange between the surface and bottom water layer as well as top-to-bottom ventilation.
 Previous reports on currents in the Laptev Sea in winter are restricted to a number of summary articles based more on general geographical knowledge than on systematic observations [e.g., Dobrovolskiy and Zalogin, 1982]. The alongshore southward current from the northwest has been described as turning farther to the northeast, contributing to the cyclonic surface circulation over the Laptev Sea. Hydrodynamic models [e.g., Pavlov and Pavlov, 1999] predict that circulation is mostly controlled by the large-scale surface salinity distribution, with winter current velocities generally below 5–8 cm/s. The modeled geostrophic currents in the area north of the Lena River Delta are directed to the north, dominating the offshore winter circulation regime in this area.
 While significant progress has been achieved in understanding the formation of the surface and bottom water layers [e.g., Steele and Ermold, 2004; Dmitrenko et al., 2005a; Bauch et al., 2009; Dmitrenko et al. 2010b, 2011], the vertical exchange across the SHL remains poorly understood. Numerical models suggest that lunar semidiurnal tides (M2) with a period of half a lunar day (12 h and 25.2 min) are likely to be among the most energetic high-frequency motions on the Siberian shelf [e.g.,Kowalik and Proshutinsky, 1994; Kagan et al., 2008]; however, little is known about Siberian shelf tidally driven water dynamics and mixing. When the barotropic tide interacts with rough bottom topography, internal waves may be generated with similar tidal frequencies. These waves, known as internal or baroclinic tides, may play an important role in promoting shear instabilities, turbulence, and the mixing of seawater properties [e.g., Padman and Dillon, 1991; D'Asaro and Morison, 1992; Muench et al., 2002]. Furthermore, the ice cover modifies shallow water tidal dynamics due to the additional friction in the boundary layer of the ice-water interface [e.g.,Prinsenberg and Bennett, 1989; Prinsenberg and Ingram, 1991; St-Laurent et al., 2008]. The under-ice damping of the tidal flow contributes to an enhancement in the shear and hence baroclinicity of the tide. In the following, under the baroclinic tidal currents we mean the depth-dependent tidal currents generated by the interaction of barotropic tidal currents with bottom topography, stratification and ice cover.
 As a part of the Russian-German project framework “Laptev Sea System,” a fieldwork program was conducted in the Laptev Sea West New Siberian (WNS) coastal polynya in April–May 2008 and 2009 to study the SHL response to polynya activity. The domain of interest is in the vicinity of the Lena Delta in the southeastern Laptev Sea (Figure 1).
 This paper is motivated by observations of middepth nearly homogeneous water layers recorded across the SHL during April 2008 and 2009 beneath the land-fast ice cover, but close to the fast ice edge contouring the onshore perimeter of the coastal polynya (Figures 2 and 3). These nearly homogeneous layers were observed during both the 2008 and 2009 field campaigns characterized by relatively weak (∼2.5 psu, April 2008) and extraordinarily strong (∼10 psu, April 2009) vertical salinity difference across the primary SHL from 6–7 m to 18–19 m (Figures 1c, 2b, and 3b). The main goal of this paper is to examine the Laptev Sea SHL susceptibility to the local vertical mixing apparently induced by shear instability of tidal baroclinic flow. In this study, however, we focus on the consequences instead of the origins of the tidal flow baroclinicity. We demonstrate that the nearly homogeneous layers across the SHL coincide with enhanced shear-induced vertical mixing that promotes vertical exchange and ventilation of the bottom water layer. In this study, our SHL observations complemented by tidal flow harmonic and wavelet analysis allow speculation on the role of baroclinic tide and associated shear instability for the SHL vertical mixing.
 The paper is structured as follows. Section 2 presents a brief description of the observational data obtained in April–May 2008 and April 2009. Section 3 documents evidence of nearly homogeneous layers in the SHL, along with patterns of water tidal dynamics. In section 4.1we consider advection-forced changes in stratification not directly associated with baroclinic tides.Section 4.2discusses shear instability of tidal flow to emphasize the role of the tidal-driven dynamics in modifying the SHL properties. Insection 4.3 we discuss the influence of observed salinity (density) stratification and sea ice conditions on our results, placing them within the context of previous research in this area. Section 5 summarizes our conclusions.
2. Data and Methods
 Between April and May 2008 and in April 2009 the moorings were deployed from the stationary land-fast ice in the vicinity of the fast ice edge, contouring the onshore perimeter of the WNS coastal polynya (Figure 1). The moorings were anchored at the seafloor and on land-fast ice at 21 and 23 m water depth in 2008 and 2009, respectively. In 2008 the mooring (for position, seeFigure 1a) was deployed for 24 days beneath 57 cm thick land-fast ice 10 m from the fast-ice edge. At the time of recovery this mooring was situated beneath 87 cm thick fast ice, ∼1 km from the fast-ice edge (seeDmitrenko et al. [2010a]for more details). In 2009 the mooring was deployed for 15 days beneath 112 cm thick land-fast ice ∼800 m from the fast-ice edge (Figure 1b). The mooring locations in 2008 and 2009 were separated by ∼25 km (Figure 1).
 Both moorings carried a combination of 300 kHz downward looking Workhorse Sentinel acoustic Doppler current profilers (ADCPs) by Teledyne RD Instruments, and Sea-Bird Electronics Inc. SBE-37s with conductivity-temperature-depth (CTD) sensors. The velocity data from the ADCPs were taken at 1 m depth intervals, with a 5 min ensemble time interval and 60 pings per ensemble. The first valid bin was located at 6 m depth. In the 2009 ADCP record, the velocity data from the bin at 7 m depth demonstrate a permanent reduction in velocity magnitude, which is most likely due to the acoustic shadow from the SBE-37 deployed on the mooring line at the same depth level. These data were omitted from the analysis presented in this study.
 The SBE-37s provided 1 min interval conductivity, temperature, and pressure at fixed depths sampling the surface (∼4.5 m) and bottom (∼19 m) water layers in 2008, and 2, 7, 15 and 22 m in 2009. The 2008 CTD records from unpumped SBE-37s were free of rapid spikes in salinity/conductivity raw data. In contrast, the 2009 records from pumped SBE-37s contained numerous rapid spikes in salinity/conductivity raw data, that can be attributed to the frazil ice crystals inside the conductivity cell [Skogseth et al., 2009]. These frazil-ice-contaminated records were eliminated from further data processing.
 We use velocity observations from the yearlong nearby moorings Anabar and Khatanga (for mooring positions, see Figure 1a) equipped with 300 kHz upward looking ADCPs to assess spatial patterns of the baroclinic tide observed near the fast ice edge in April–May 2008. The velocity records from the 2008 and 2009 land-fast ice moored ADCPs were insufficiently long for reliably resolving the spring-neap tidal cycle conditioned by interaction between M2 and S2tidal constituents. Instead, velocity data from Anabar are used to estimate the spring-neap tidal cycle in April–May 2008. There is no available velocity record from Anabar and Khatanga for April 2009.
 The moored observations were complemented by independent CTD profiles taken at the mooring positions using a SBE19+ CTD during mooring deployment and recovery. In addition, three CTD casts were taken in both 2008 and 2009 between mooring deployment and recovery (Figures 2 and 3).
 According to manufacturers' estimates, individual temperature and conductivity measurements are accurate to ±0.005°C and ±0.0005 S/m, respectively, for the SBE-19+, and to ±0.002°C and ±0.0003 S/m, respectively, for the SBE-37s. The ADCP velocity precision and resolution are ±0.5% and ±0.1 cm/s, respectively. The ADCP velocity estimated error was 1.8 cm/s. Compass accuracy is ±2.7 deg. The current direction was corrected by adding local mean magnetic deviation (−16 deg). All CTDs were calibrated before the expedition.
 The domain of interest in the vicinity of the Lena Delta receives on average 435 km3 of freshwater from the Lena River runoff each summer (June–September). Between November and April a steady but low discharge of ∼38 km3is recorded (Arctic-RIMS data,http://rims.unh.edu). Marked differences observed in the winter surface salinities over two consecutive winter seasons of 2007–2008 and 2008–2009 in the area of the winter coastal polynya can be primarily attributed to preconditioning from differences in summer wind-driven diversion of river runoff [Dmitrenko et al., 2010c]. In the summer of 2007, dominant along-shore westerly winds of a cyclonic atmospheric regime force the Lena River runoff to flow eastward. In contrast, in the summer of 2008, dominant along-shore easterly winds over the East Siberian Sea and onshore northerly winds over the Laptev Sea of an anticyclonic regime retain the riverine water in the vicinity of the Lena Delta. Within the coastal polynya, in the southeastern Laptev Sea these differing patterns explain the under-ice surface salinity difference of 8–16 psu between the winters of 2008 and 2009. Consequently, the winter 2009 SHL is significantly sharper than in winter 2008 (Figures 2 and 3).
 Over the mooring deployment period in 2008 and 2009, the primary SHL was located between ∼6–7 m and ∼18–19 m (Figures 2b and 3b); however, in 2009, it was overlaid by a significantly fresher under-ice water layer with salinity ranging from ∼12 to 20 psu (Figure 3b). Across the primary halocline, the salinity increases from ∼28 psu to 30.5 psu in 2008 and from 20 to 22 psu to 31–32 psu in 2009. The vertical salinity gradient varies in time with a maximum value of ∼0.86 psu/m observed in 23 April 2009 (Figure 3b). Among the five CTD profiles spread throughout the each of the 2008 and 2009 mooring deployment periods, two consecutive profiles from each year exhibit nearly homogeneous water layer across the SHL with almost uniform vertical salinity distribution at 12–17 m (2008, Figure 2b) and at 11–15 m (2009, Figure 3b). In contrast, the CTD profiles taken at deployment and recovery demonstrate continuous salinity stratification with no homogeneous water layers across the SHL.
 The velocity record appears to be strongly influenced by semidiurnal tides (Figures 4 and 5). However, over the Laptev Sea shelf the inertial frequency ƒ, defined as twice the rotation rate of the Earth multiplied by the sine of the latitude, is close to the frequency of the dominant semidiurnal M2 tidal constituent. Thus, our data record lengths are too short to distinguish between inertial motion with a period ranging over the Laptev Sea shelf (between 73° N and 76° N) from 12.51 h to 12.33 h and semidiurnal tides (12.42 h and 12.00 h for M2 and S2, respectively). Hence, inertial oscillations may be included in the tidal signal inferred from our observations.
 A substantive fraction of the energy of semidiurnal tidal currents over the shelf areas is associated with a baroclinic component [e.g., Marsden et al., 1994; Furevik and Foldvik, 1996; Kulikov et al., 2004, 2010]. In the following, we analyze the tidal flow assuming it to consist of barotropic and irregular baroclinic components. However, based on the records we have, these two different components cannot be accurately separated from the total current. We considered it inappropriate to take depth averages of the resolved measurements as the absolute barotropic component and the remainder as the absolute baroclinic tide because the observations omit 40% of the whole water column (due to the ADCP blanking distances and hard reflections off the bottom). Hereafter, we focus on analyzing the total tidal current, assuming it may consist of both barotropic and baroclinic components.
 Amplitudes of the main tidal harmonics (M2, S2, K1, and O1) were estimated using an algorithm by Foreman , and were computed for each valid bin. Our tidal analysis revealed the dominance of the lunar semidiurnal constituent M2(78 ± 3% and 72 ± 15% of the tidal energy in 2008 and 2009, respectively, depending on depths) with a depth-dependent amplitude ranging from 9 to 16 cm/s in 2008 and from 5 to 9 cm/s in 2009. The concurrent solar semidiurnal constituent S2with a period of half a solar day (12.00 h) contributes 21 ± 3% and 16 ± 7% in 2008 and 2009, respectively, with a depth-dependent amplitude ranging from 5 to 8 cm/s in 2008, and from 2 to 4 cm/s in 2009. The contribution of the K1 and O1 constituents is negligible.
 The resulting tidal ellipses for predominant constituent M2 are plotted in Figure 6. The axis of maximum variance (major axis) is well defined. In 2008, the dominant tidal flow was approximately parallel to the isobaths and the nearby shoreline (Figure 1a). The M2tidal ellipses rotated anticlockwise with the major axis aligned with ∼90 deg in the subsurface layer (6–8 m), and to ∼131 deg in the near-bottom layer (17–19 m) (Figure 6a). In the SHL (9–16 m), the M2tidal ellipses rotated clockwise with the major axis aligned with ∼125 deg. In 2009, we observed almost unidirectional tidal flow aligned with 145 deg in the subsurface layer (6–8 m), to 135–137 deg in the SHL (9–18 m), and to 125–130 deg in the near-bottom layer (19–21 m) (Figure 6b). As in 2008, in 2009 the M2 tidal ellipses are oriented approximately along the axis of the local submerged valley; however, in this case this is roughly perpendicular to the local isobaths (Figure 1). In both years, the tidal flow ellipticity (defined as the ratio of minor to major ellipse axis) tends to decrease approaching the depths of the surface and bottom boundary layers (Figure 6).
 A striking feature of the tidal flow is the increased amplitude of the semidiurnal tidal flow velocity at the depth range of the SHL (Figure 6), indicating significant shear in the tidal currents. For example, in 2008 the major axis amplitude for the M2tidal ellipses achieves a maximum of 15.4 ± 2.9 cm/s in the SHL at 11 m, while for the subsurface layer at 7 m and the near-bottom layer at 19 m it is 9.8 ± 2.9 cm/s and 7.5 ± 2.0 cm/s, respectively (Figure 6a). In 2009, the semidiurnal tidal flow exhibits similar patterns, but velocities are reduced by a factor of 2 (8.5 ± 2.9 cm/s in the SHL at 15 m, 3.5 ± 2.2 cm/s in the subsurface layer at 8 m, and 3.4 ± 2.1 cm/s in the near-bottom layer at 21 m,Figure 6b). In 2008 and 2009, the clockwise-polarized tidal ellipses exhibit considerably more baroclinic character than the nearly depth-invariant counterclockwise polarized ellipses (Figure 7). This result is consistent with numerical simulation of the barotropic M2 tidal currents by Kagan et al. . Over the same region under the ice cover, they reported the major axis amplitude for the predominant barotropic M2 tide to be ∼4–6 cm/s, a value significantly less than the major axis amplitudes revealed across the SHL (Figure 6a).
 Our observations are likely to be influenced by an interaction between the major semidiurnal tidal constituents M2 and S2that produces the spring-neap tidal cycle (∼14 days). As the spring-neap cycle is not sufficiently resolved by our velocity record (24 day in 2008 and 15 day in 2009), we use velocity data from the nearby yearlong mooring Anabar (Figure 1a) to estimate the spring-neap tidal cycle in April–May 2008. The M2 and S2 tidal constituents are subtracted from the yearlong Anabar velocity record from the bin at 14 m in the SHL using an algorithm by Foreman  (Figure 8). Comparing the Anabar spring-neap tidal flow to fast-ice edge tidal velocities, obtained by high-pass filtering with cutoff period 24 h, it is clear that that semidiurnal tidal flow at the nearby fast ice edge is strongly modulated by the spring-neap cycle (Figure 8). At the end of April 2008, a significant reduction in spring-neap tidal flow near the fast ice edge is observed in the meridional component (Figure 8, bottom). In April 2009, the semidiurnal tidal flow is substantially lower, with weaker spring-neap signal. The spring tide occurs between 11 and 14 April, and the neap tide occurs between 18 and 21 April (not shown). This is also consistent with the spring-neap cycle computed based on data from yearlong mooring Anabar (2008–2009).
 The M2 semidiurnal tidal currents are sheared over the depth range of the nearly homogeneous water layer across the SHL (Figures 6 and 7). This can be represented by the layer-mean shear squaredVΔ = [(∂u/∂z)2 + (∂v/∂z)2]/2, computed across the SHL by taking the difference between the top and bottom boundary layer velocities and divided by the SHL thickness. The layer-mean shear squared exhibits predominance of semidiurnal periodicity (Figures 4c and 5c). For 18–28 April 2008, the velocity shear is clearly amplified by spring-neap modulation (Figures 4c and 8), and a nearly homogeneous water layer across the SHL in 18 April and 24 April 2008 coincides with enhanced velocity shear across the SHL during the spring tide (Figures 4c and 8). For April 2009, the SHL mixing events are also tied to enhanced semidiurnal velocity shear (Figure 5c), but this shear is not clearly modulated by the spring-neap cycle, as we observe in April 2008 (not shown). In fact, semidiurnal currents are found to vary significantly with depth, pointing to the possibility that internal baroclinic tides may influence the formation of the middepth mixed layers across the SHL.
 An additional perspective on tidal flow baroclinicity and spring-neap tidal modulation is provided by wavelet analysis of velocity records. The wavelet transform method is used to analyze time series that contain nonstationary power at different frequencies. The results of the wavelet transform allow the spectral composition of nonstationary signals to be measured and compared [Foufoula-Georgiou and Kumar, 1995]. The wavelet spectrum for the zonal and meridional currents was computed for the semidiurnal band as a function of depth (Figures 9 and 10). In April–May 2008, the semidiurnal velocity component experiences spring-neap modulation with significant tidal flow amplification in the SHL, also seen inFigure 4. The most striking example is seen from 18 to 28 April 2008 with the semidiurnal currents exhibiting clear baroclinic structure and reaching a maximum amplitude of 30 cm/s in the SHL (Figure 9). This episode is consistent with vertically mixed layers observed across the SHL where the semidiurnal flow experiences significant vertical velocity shear (Figure 9). In contrast to 18–27 April, from 28 April to 3 May, the semidiurnal currents were relatively weak and were dominated by barotropic flow component (Figure 9). In April 2009, the semidiurnal flow wavelet spectrum contains substantially lower background energy with several episodes of enhanced baroclinicity that were not linked to spring-neap tidal modulation (Figures 10a and 10b). For example, for April 2009, the layer-mean shear squared peaks in 15–17 April (Figure 10c) are a result of a phase shift between the surface and lower layer. This is also evident from Figures 10a and 10b, demonstrating depth-dependent displacement of spectrum maxima in 15–17 April 2009. Subsequently, the upper layer tidal currents were almost out of phase relative to the lower layer in a manner that enhances shear in the semidiurnal velocities. The “mixing event” in 21 April 2009 seems to be not directly associated with baroclinic flow. Both the layer-mean shear squared (Figure 5c) and wavelet spectrum of the layer-mean shear squared calculated for the semidiurnal frequency band (Figure 10c) exhibit enhanced but relatively low values.
 In the following section we speculate on the role of the semidiurnal baroclinic tide and associated shear instability in driving vertical mixing across the SHL. Before commencing on a detailed discussion of the tidal-driven shear instability, insection 4.1 we consider the possibility that the nearly homogeneous layers observed in SHL can be explained by (i) horizontal advection of a middepth mixed layer created remotely from the observing region and (ii) local generation by shear instability of residual baroclinic currents not associated with tidal flow. In section 4.2 we examine SHL susceptibility to local vertical mixing driven by shear instability of semidiurnal tidal flow. Section 4.3provides additional evidence on the origin of the velocity shear observed across the SHL. We speculate on the source region of the baroclinic tide on the eastern Laptev Sea shelf and discuss the role of semidiurnal baroclinic tides in forcing the Laptev Sea shelf hydrography during the ice-covered period.
4.1. Residual Currents
 The residual currents are obtained by low-pass filtering the ADCP velocity records with cutoff period of 24 h. Residual currents occurring simultaneously with the nearly homogeneous middepth SHL layers in 18–24 April 2008 and 14–21 April 2009 are shown inFigures 11a and 12a, respectively. These residual currents are used to investigate the possible advection of the nearly homogenous water layer into the study region.
 Over the depth range of the SHL, in 18–24 April 2008, the mean lateral transport is aligned with 294 deg with a mean velocity of 2.8 cm/s. This observed residual mean flow is capable of advecting water from the remote land-fast ice covered region as far as ∼17 km onshore from the fast ice edge. Similarly to 2008, the mean residual flow in April 2009 is aligned with ∼283 deg with a mean velocity of 4.5 cm/s that is consistent with hypothetical remote origin of source water at most ∼24 km southeast of mooring position.
 The question then becomes: what mixing mechanisms occurring under land-fast ice surrounding potential mixing site can create the middepth mixed layer (Figure 1)? If onshore vertical mixing associated with sea ice and/or atmospheric forcing is responsible, we may expect convection within a coastal polynya to produce vertically mixed waters down to the middepths [e.g., Krumpen et al., 2011]. Recovery of stratification after polynya events can result in a middepth vertically mixed layer similar to those shown in Figures 2 and 3. However, satellite imagery from mid-February 2008 and 2009 shows that the fast ice edge was stationary and did not migrate offshore, such that there were no coastal polynyas coincident with our hypothetical regional mixing sites (not shown).
 The middepth mixed layer across the SHL can be generated remotely under the land-fast ice area southeast of mooring position as indicated by ADCP record. Historical hydrographic data from the 1920s to 2010 for the southeast land-fast ice area allow for estimation of the Baines parameter which determines if interaction between barotropic waves, stratification and bottom topography can result in internal wave generation [Baines, 1986] (see section 4.3 for more details). Our estimates of the Baines parameter indicate that baroclinic waves may not be generated in this area and cannot thus drive mixing at that location.
 The middepth mixed layer across the SHL can be generated locally by shear instability of residual baroclinic currents. For 18–24 April 2008 and 14–21 April 2009, the layer-mean shear squared associated with shear of residual current (low-pass filtered with cutoff period 24 h) across the SHL is shown inFigures 11b and 12b, respectively. For April 2008, significant layer-mean shear squared up to 1.8 · 10−4 s−2is observed in the upper SHL from 5 to 11 m. In contrast, over the depth range of lower SHL, where the near-homogeneous water layer has been recorded, the layer-mean shear squared is substantially lower, not exceeding 0.6 · 10−4 s−2 (Figure 11b). In April 2009, the residual current exhibits vertical patterns more consistent with barotropic-like behavior, with layer-mean shear squared across the SHL not exceeding 1·10−4 s−2 (Figure 12b). In both 2008 and 2009, the maximum layer-mean shear squared does not exceed 2 · 10−4 s−2which constitutes less than 25% of the total layer-mean shear squared across the SHL (compareFigures 4c and 11b and Figures 5c and 12b). This indicates the high-frequency shear mainly associated with baroclinic tide has a leading role in generating vertically sheared currents across the SHL that can lead to shear instability and local vertical mixing.
4.2. Assessment of SHL Susceptibility to the Local Vertical Mixing
 The gradient Richardson number provides a strong constraint for identifying turbulence produced by shear instability. It is defined as:
where Nis Brunt-Vaisala frequency, andS2 = Uz2 + Vz2(squared vertical shear of flow) is twice the layer-mean shear squared (S2 = 2VΔ), but calculated with greater vertical resolution than in Figures 4c and 5c. According to the inviscid theory of Miles [1961, 1963] and Howard  the flow should be stable and laminar if Ri > 0.25. While Ri < 0.25 is the generally accepted critical value below which turbulence can be expected to occur, observational data are rarely sufficiently resolved to determine Ri with such precision, and in practice, turbulent mixing can be assumed to occur at higher observed values of Ri < 1 [Abarbanel et al., 1984; Miles, 1986; Polzin, 1996].
 The ADCP velocity record was used to compute the gradient Richardson numbers to evaluate the flow stability. In this context, we consider all the shear as coming from the baroclinic tide and from other sources of baroclinicity (mean background baroclinic circulation etc.). Given the sparse CTD profiling data, two sets of idealized experiments were conducted for the different background stratification estimated based on two first CTD profiles taken at mooring positions, assuming these profiles properly represent the background stratification and preexisting vertical conditions (Figures 2b and 3b). For the first set of calculations, we use only the CTD profile taken at mooring deployment (Figures 13a and 14a). For the second experiment, the background stratification was estimated based on the second CTD profile (Figures 13b and 14b).
 As seen in Figures 13a and 14a, and based on the usual criterion for shear instability of Ri < 0.25, for the first set of computations, within the SHL there is no evidence of shear of sufficient magnitude to transform the background stratification into the nearly homogeneous layer. However, Richardson numbers less than 1 are evident within the SHL in 22–24 April 2008 (Figure 13a) and in 15–17 April 2009 (Figure 14a), when enhanced cross-SHL shear was observed in the semidiurnal baroclinic tide (Figures 4c, 5c, 9c, and 10c). In 2008, the SHL at ∼15–16 m also episodically exhibits Ri < 1, except during the 29 April to 3 May period (Figure 13a) when the semidiurnal tidal current was relatively weak and the currents were mainly barotropic across the SHL (Figures 4c and 9). In contrast, the nearly homogeneous layer recorded in 21 April 2009 seems to be disassociated with shear instability consistent with relatively low velocity shear in Figures 5c and 10c. This suggests that lateral advection may be responsible for transporting vertically mixed SHL water from nearby locations to the mooring position.
 The second set of computations reveals episodic SHL susceptibility to local vertical mixing (Ri < 0.25) across the SHL occurs only in 2008 (Figures 13b). This is consistent with (1) the nearly homogeneous layer across the SHL present in the CTD profiles taken in 19 and 24 April 2008 (Figure 2) and (2) the enhanced velocity shear across the SHL (Figure 4c). Furthermore, in April 2008, there is a general increased incidence of turbulent dissipation-favorableRi < 1 [Polzin, 1996] across the lower part of the SHL (Figure 13b). Partial recovery of the SHL at the end of recording period in 5 May 2008 (Figure 2) at Ri ∼ 0.25 (Figure 13b) can only be explained by enhanced lateral advection that transports continuously stratified water from other locations.
 A locally pervasive and almost barotropic flow of up to 13–15 cm/s aligned northeast along the offshore edge of the polynya, was recorded between 28 April and 2 May 2008 (Figure 3). Dmitrenko et al. [2010a]speculated that this motion arose from geostrophic adjustment to the sizable temperature and salinity gradients created across the fast ice edge at wind-driven polynya opening observed between 28 April and 3 May 2008 (Figure 15). This locally pervasive flow is consistent with our hypothesis concerning the importance of littoral fluxes in restratification of the SHL. At the same time, the SHL restratification within a time scale of several days suggests spatial differences in the efficiency of tidal-driven vertical mixing. This is in consistent with the velocity records from the nearby Anabar and Khatanga moorings (Figure 1a). In contrast to the fast ice edge mooring, the tidal ellipses computed based on velocity records from Khatanga and Anabar (11 April to 5 May 2008) shows almost no tidal flow baroclinicity (Figure 16a, Khatanga) or a relatively weak baroclinic component (Figure 16b, Anabar) associated with M2 tidal constituent.
 Expanding on section 4.1, we suggest that susceptibility to shear instability can be partly associated with the baroclinic tide. In 2008, Richardson numbers <1 across the SHL exhibit variability coherent with tidal-driven forcing (Figures 9 and 13). The wavelet power spectrum for the time series of currents computed for the semidiurnal frequency band as a function of depth (Figure 9) demonstrates energetic patterns consistent with shear instability events defined by Ri < 0.25 (Figures 13b). In fact, over the entire mooring record in 2008, the SHL Ri numbers are generally tied to the baroclinic tidal forcing. For 2009 with stronger salinity stratification (Figure 1c) and weaker baroclinic tide (Figure 6), only the 15 April event can be associated with tidal-driven shear instability.
 While our tidal observations may include an inertial component, the formation of sea ice cover in winter significantly shields the sea surface from the surface wind forcing that is mainly responsible for the inertial current generation [e.g., Rainville and Woodgate, 2009, and references therein]. Furthermore, our wavelet analysis shows that the phase of the semidiurnal currents is dominated by the M2 tide instead of randomly phased inertial oscillations (Figures 9 and 10). Thus, the short-term amplified semidiurnal currents in 2009 are unlikely to be the inertial currents generated by individual intensive atmospheric events.
 The overall impression from the mooring observations and CTD profiling is that during recorded period in 2008, the nearly homogeneous layer has been maintained across the SHL by shear instability of the baroclinic tidal flow. This conclusion is based on our estimation of the gradient Richardson numbers in Figures 13b which detects occurrence of shear instability (Ri < 0.25) in the SHL prior to the appearance of the nearly homogeneous layer recorded by CTD profiling. In 2009, no convincing evidence for the occurrence of shear instability is provided, while several episodic events with Ri < 1 indicate that the shear instability may play a role in the local modification of SHL.
4.3. Semidiurnal Internal Tides Over the Eastern Laptev Sea Shelf: Source Regions and Local Sea Ice Impact
 There are two issues here that inspire further discussion. First, the differences in local tidal dynamics and background stratification recorded by our moorings between 2008 and 2009 allow some speculation on the “remote” origin of the baroclinic tide on the eastern Laptev Sea shelf. Second, the rapid tidal flow response to polynya openings and to associated changes in vertical stratification as well as strong spatial variability of baroclinic patterns are highly suggestive that the internal tide is generated reasonably close to our moorings. The 2009 record also suggests that enhanced semidiurnal-band velocity shear across the SHL can be associated with depth-dependent phase behavior. In this study, we focus on the consequences instead of the origins of the depth-dependent phase differences.
 A striking difference in the amplitudes of the baroclinic semidiurnal currents was found between 2008 and 2009 (Figure 6). Studies from other regions demonstrated good correlation between changes in the stratification and changes in the internal tide energy. In general, internal tide energy is expected to increase with increasing stratification [e.g., Rosenfeld, 1990; Lerczak et al., 2003; Makinson, 2002; Makinson et al., 2006]. In fact in 2008, we observe the opposite, with stronger baroclinic tidal flow coinciding with weaker stratification (Figures 2, 3, and 6–9). This discrepancy may be associated with changes in the density field at a distant internal tide generation site that does not affect the measurement site in the same way [e.g., Holloway, 1984]. In this context, our results are suggestive that the internal tide observed at the mooring site occurs remotely. It seems that enhanced vertical stratification in 2009 was a local phenomenon that had no impact on the “remote” generating area, and the baroclinic shear in 2009 is due to an internal tide generated at some distant site with weaker stratification. This is in line with results by Dmitrenko et al. [2010c]. They have demonstrated that enhanced vertical stratification over the mooring site in winter 2009 was locally preconditioned over the very limited area northeast off the Lena Delta by wind forcing during the preceding summer of 2008.
 At the same time, the patterns of semidiurnal tidal flow seen in Figures 8 and 9 are tied to the local sea ice conditions. Polynya openings from 28 April to 3 May 2008 (Figure 15) coincide with damping of the baroclinic tidal flow (Figures 8 and 9) that is not fully explained by the spring-neap tidal cycle (Figure 8). This is consistent with earlier observations from a 1998–1999 first ever yearlong ADCP record from mooring Yana at the Laptev Sea shelf in a position located reasonably close to our mooring sites [Dmitrenko et al., 2002] (Figure 1b). At mooring Yana, Dmitrenko et al. reported a relatively weak baroclinic tidal flow component coinciding with ice-free conditions during summer as well as during a spring polynya opening (Figure 17). In contrast, during winter below the ice cover, enhancement of the baroclinic tidal flow has been observed in the SHL. This finding is also supported by similar results by E. Litvinov and I. A. Dmitrenko (personal communication, 2008) obtained based on 2003–2004 yearlong ADCP record at mooring Lena (see Figure 1b for mooring position). However, the nearby moorings Khatanga and Anabar, in April–May 2008, revealed no velocity shear that can be attributed to baroclinic tidal forcing below the ice cover (Figure 16).
 The nature of the baroclinic tide amplification on the shelf below the ice cover is beyond the scope of the present paper. Hereafter we limit our efforts to the discussion of several important issues directly related to present analysis. First, we suggest that enhanced vertical mixing below the ice cover can be related to both the “true” baroclinic tide and increased frictional damping by sea ice cover.
 Turbulence generated by the bottom stress can be expected to play an important role in vertical mixing within the boundary layers bordering the SHL, such that the vertical structure of tidal currents can be maintained by friction, even in the absence of salinity (density) stratification [e.g., Prinsenberg and Bennett, 1989; Prinsenberg and Ingram, 1991; Furevik and Foldvik, 1996]. We note that the shear arising from friction may also have a role in modifying the barotropic tidal currents below the ice cover. The freezeup onset usually results in damped tidal flow underneath the ice [e.g., St-Laurent et al., 2008] (e.g., Figure 17). The under-ice damping of the barotropic tide contributes to an enhancement in the shear and hence baroclinicity of the tide. However, the increased under-ice friction has no effect on enhancement of the tidal flow across the underlying SHL that tends to be ∼10 cm/s higher in winter than in summer (Figure 17). Projecting this result on our records strengthens our suggestion that the “true” baroclinic internal tide has a role in enhancing the semidiurnal tidal flow across the SHL (Figure 6).
 Second, we assume that our records also implicate regional sea ice conditions in generating the baroclinic tide during the ice-covered season. In summer, when stratification is relatively strong, the baroclinic response to barotropic tidal forcing should be more prominent. In winter (ice-covered period), weaker stratification leads to damped internal tides [e.g.,Makinson, 2002; Makinson et al., 2006]. In fact, Dmitrenko et al.  observed the opposite, with stronger baroclinic tidal flow during winter at weaker stratification (Figure 17). In general, the stratification changes gradually as the seasons turn, so that seasonal variations of internal tides should likewise be slow. In contrast, Dmitrenko et al.  and E. Litvinov and I. A. Dmitrenko (personal communication, 2008) observed abrupt amplification of the baroclinic tidal flow related to freezeup onset that implicates the role of local sea ice conditions (Figure 17). Furthermore, all these records, including our data shown in Figures 8 and 9, demonstrate damping of the baroclinic tidal component at the polynya opening.
 Polynya formation results in significant weakening of the local salinity (density) stratification. For example, from 28 April to 3 May 2008 the surface layer salinity from SBE-37 increased from ∼26 psu to ∼30 psu due to strong salt input to the underlying shelf water associated with brine release at polynya ice formation [Dmitrenko et al., 2010a]. Following the polynya opening, the vertical salinity difference across the SHL was reduced by 4.1 psu (from 5.4 psu in 27 April to 1.3 psu in 2 May; Figure 18). We speculate that local damping of the baroclinic tide at polynya opening demonstrates the high sensitivity of the baroclinic tidal flow to rapid changes in salinity stratification. This suggests any internal tides propagating past our moorings seem to be generated within a region reasonably close to our moorings. Note that the polynya opening in 13–15 April 2009 was relatively small (compare the polynya areas in 1 May 2008 and 15 April 2009 in Figure 1) with majority of polynya newly grown ice to the south from 74.5°N (Figure 1b). Based on wavelet spectrum for semidiurnal velocity shown in Figure 10, we reveal changes in semidiurnal flow that can be attributed to the near-inertial waves generated by wind stress over the polynya open water. This interpretation makes sense given the short duration of the high-shear event (<2 days) in 13–15 April 2009 (Figure 10b).
 The latitude where the tidal frequency ω equals the inertial frequency ƒ is referred to as the critical latitude φcrit. It is believed that local generation of internal tides at near-critical latitudes is an essential contribution to baroclinic currents [Furevik and Foldvik, 1996; Robertson, 2001; Kulikov et al., 2004, 2010], and changes in stratification over the critical latitude region are likely to significantly affect the tidal current profile [Makinson, 2002]. The φcrit for M2 is at 74°28.5′N, which is ∼55 km poleward of the fast ice edge mooring positions. Indeed, the polynya opening at the end of April to the beginning of May 2008 extends over the critical latitude (Figure 15), such that we expect the change in salinity stratification from the polynya opening can locally modify the generation of the internal tide. The enhanced CW rotation component (Figure 7) is also typical of the M2 tide near its critical latitude [e.g., Furevik and Foldvik, 1996]. At the same time, mooring Khatanga, located in the vicinity of critical latitude (Figure 1a), exhibits no tidal flow baroclinicity that can be attributed to the enhanced generation of internal tide (Figure 16). One reason the Khatanga tides are barotropic is that the water column may be well mixed due to the merging of thick top and bottom boundary layers that have been known to occur near the critical latitude [e.g., Furevik and Foldvik, 1996]. However, the role of vertical stratification here is unclear—there are no CTD profiles from the Khatanga area taken in April–May 2008. Moreover, it is equally possible that there is no source of baroclinic tides that influences the Khatanga area.
 Generation of baroclinic internal waves strongly depends on steepness of the bottom topography and the background stratification. Baines  introduced a parameter γ for determining the necessary conditions for internal wave generation:
where ∂H/∂y is bottom slope, ω is wave frequency, and Nis the Brunt-Vaisala frequency (s−1). Where γ ∼1, the slope is critical, resulting in the greatest generation of internal waves. Where γ is <1, the slope is subcritical, and fewer internal waves are generated. Where γis >1, the slope is supercritical and internal waves may be generated, but only propagate off-slope [Robertson, 2001].
 We eliminate the Laptev Sea shelf break as a possible source of the M2 internal tide at the mooring site. Based on the theory, the internal waves with M2 frequency are not permitted to propagate freely inshore from the Laptev Sea shelf break located north of the M2 critical latitude and ∼300 km north of our moorings. However, for the region of mooring deployment located at the nearby critical latitude (Figure 1), the ω(M2) → f, α → 0, and γ tends to increase even for weakly stratified conditions (small N) and relatively gentle bottom slope (Figure 1). For winter mean hydrographic conditions over the region of mooring deployment (bottom slope ΔH/Δy = 10 m/15–25 km), the value of N near the seabed (18–22 m) varies from ∼0.04 to ∼0.05 s−1. Based on these estimates, the value of γ(M2) is within the range of supercritical numbers from 1.85 to 3.08 that implies local baroclinic internal wave generation caused by interaction of the M2 tide with stratification and topography. For April 2008 and 2009, γ(M2) is ∼3 and 2.2, respectively. However, excluding interannual changes in bottom topography, the Baines criteria provide no explanation for the 2008–2009 discrepancy between tidal dynamics and stratification.
 In general, our discussion about the source region for the baroclinic tides on the shelf implicates the M2critical latitude effects to be important influence in the immediate response of semidiurnal water dynamics to local sea ice and vertical stratification conditions. Giving limited data coverage in 2009, the 2008–2009 discrepancy between tidal dynamics and stratification can be explained by assuming that the region of critical latitude has experienced no changes in winter 2009 since becoming preconditioned by wind-forced diversion of river runoff during the preceding summer.
5. Summary and Conclusions
 We have used the CTD and ADCP data obtained in April–May 2008 and April 2009 at two mooring sites located along the land-fast ice edge over the southeastern Laptev Sea shelf to investigate the SHL susceptibility to local vertical mixing. These data show the modification of the SHL stratification through the appearance of nearly homogeneous water layers. The tidal harmonic analysis reveals vertical shear of velocity for the lunar semidiurnal constituent M2across the SHL enhanced by spring-neap modulation in 2008. This is also consistent with results of semidiurnal flow wavelet analysis. Estimates of the gradient Richardson numbers from the 2008 observations provide evidence that the baroclinic shear is sufficient to transform the background continuous stratification into the nearly homogeneous layers episodically recorded across the SHL. In contrast, no clear evidence for the occurrence of shear instability is present in the 2009 observations, but several episodic events are suggestive of the role of shear instability in modifying the SHL. In summary, we have demonstrated that shear-driven vertical mixing can locally disrupt the SHL by tending to form a homogeneous mixed water layer within the SHL, favoring ventilation of the bottom water layer. However, the advection of nearly homogeneous patches from surrounding and remote locations may also be important.
 Our analysis was significantly limited by the data available. For example, mixing events are expected to be intermittent and given the sporadic CTDs, our observations at best can give us estimates of the potential for high-shear events to destabilize the SHL. Our records also allow no clear attribution of the velocity shear across the SHL. Baroclinicity of the tidal flow seems to be among the sensible explanations, but the tidal flow modification by friction with sea ice cover at the top and with seafloor near the bottom may also be important. While some of our discussion and conclusions are necessarily speculative, the data show a signature of shear-driven vertical mixing that can erode the SHL and promote vertical exchange between surface and bottom layers during an ice covered period, when the water column is decoupled from wind forcing.
 Acknowledging the deficiency of our analysis, we suggest several hypotheses summarized below. We mean that our estimations of system susceptibility to shear instability are consistent with enhancement of baroclinic tidal flow by the spring-neap tidal cycle. However, we note that this is a region where M2baroclinic tides are prohibited from propagating southward from shelf-break-generating sites located north of the critical latitude. Hence, observed variability in the local tidal dynamics and background stratification between 2008 and 2009 is consistent with a rapid tidal flow response to abrupt buoyancy changes associated with polynya openings. Given the limitations on internal tide propagation, this points to a regional source of internal tide generation close to our moorings. The proximity of the M2 critical latitude ∼55 km north of our moorings sharply increases the likelihood that critical latitude effects will be an important influence on the tidal currents and increase their sensitivity to changes in sea ice and stratification. Baines criteria analysis also shows that M2 interaction with bottom topography is capable of locally generating baroclinic internal waves.
 Further research is needed to quantify the efficiency of the tidal-driven mixing in the Laptev Sea shallow stratified water. This research should be based on improved in situ measurements by automatic profiling equipment able to make high-frequency measurements of temperature, salinity and velocity. Such profiling data are indispensable for assessing the intensity of the vertical turbulent exchange across the SHL associated with shear instability of the semidiurnal tidal flow beneath the ice cover. The amplification of the baroclinic tide during winter below the ice is of special interest and requires further investigations.
 Financial support through the BMBF project “System Laptev Sea” is gratefully acknowledged. Yueng-Djern Lenn was funded by the UK Natural Environment Research Council ASBO Arctic IPY Consortium grant with additional support from NERC grants NE/F002432 and NE/H016007/1. This study also received financial support from the Canada Excellence Research Chair (CERC) program. Torben Klagge (IFM-GEOMAR) has rendered invaluable assistance in performing oceanographic measurements in the Laptev Sea in spring 2008 and 2009. We are grateful to Thomas Krumpen (AWI, Germany) for the ENVISAT ASAR satellite imagery shown inFigures 1 and 15. Jens Hölemann (AWI, Germany) kindly provided velocity data from moorings Anabar and Khatanga. We also appreciate Igor Polyakov (University of Alaska Fairbanks, United States) and Louis Fortier (Laval University, Canada) for lending the oceanographic equipment for the 2008 field campaign. We appreciate the editor and two anonymous reviewers for their significant efforts on improving our manuscript.