Dense water formation in the Laptev Sea flaw lead

Authors


Abstract

[1] Two primitive equation-based models are used to estimate the formation of total volumes either of Arctic cold halocline water (CHW), intermediate water (IMW), or deep water (DW) through freeze-related salt rejection in the Siberian Laptev Sea flaw lead system. Model A assumes that the rejected salt remixes with surface mixed water (SMW) beyond the leads until salinities allow for contribution to the midlayers of either the CHW, the IMW, or the DW. Model B simulates direct salt rejection to the upper layer of the cold halocline, and, after remixing here, further contribution to the midlayers of CHW, IMW, or DW. Averaging both model estimates, Laptev leads contribute either 0.161 Sv of CHW, 0.075 Sv of IMW, or 0.065 Sv of DW, which represents as much as ∼23%, ∼16%, or ∼30% of Arctic-wide lead derived dense water contribution to the appropriate layer, respectively. Northwestern Laptev leads produce the greatest amount of dense water. These lead sections show very short buoyancy equilibrium timescales (∼6 to ∼13 days), and local dense water production may potentially be amplified by lateral brine injection into the cold halocline through bottom eddies. Central-southern and southeastern leads generally produce little salt due to low surface water salinities. As definite separation mechanisms and proportion distributions of rejected lead brines into CHW, IWM, and DW are still unidentified in nature, a combination of lead salt rejection and remixing (model A) and direct downward expulsion of brine packages (model B) is assumed to steer Laptev lead dense water production.

1. Introduction

1.1. Background

[2] Various theoretical and numerical models have been applied in the past 2 decades to describe the maintenance and feeding of the Arctic cold halocline and deeper layers by dense, saline water masses produced on the circum-Arctic shelves subsequent to ice formation [e.g., Bjørk, 1989; Jones and Anderson, 1986; Chapman, 1999; Signorini and Cavalieri, 2002; Winsor and Chapman, 2002]. Becker and Bjørk [1996] described the transformation of shelf surface mixed-layer water to different types of higher saline waters and their interleaving in the model water column at their respective level of neutral density. The authors estimated a total rate of 0.8 Sv of cold, high-salinity water produced on the Arctic shelves.

[3] Gawarkiewicz and Chapman [1995], Chapman and Gawarkiewicz [1997], Chapman [1999], and Gawarkiewicz [2000] developed theoretical, several-step models describing the production and offshore flux of freeze-related dense brines in idealized high-latitude coastal polynya areas. The model results suggest an initial stage of geostrophic adjustment associated with dense water accumulation within the polynya water body subsequent to continuous constant negative buoyancy forcing, followed by frontal meandering instabilities equalizing coastal constraints, which then finally lead to the development of cross-shelf eddy fluxes transporting polynya-produced dense water offshore. In contrast to the above theoretical Gwarkiewicz and Chapman model predictions that dense polynya water is transported offshore by eddies, Danielson et al. [2006] found in the shallow, gently sloping St. Lawrence polynya area that this mechanism of cross-shelf brine flux was negligible and that dense water was rather advected offshore before significant eddy fluxes could establish.

[4] Salt release and subsequent dense brine rejection during continuous ice formation in Siberian and western Arctic flaw leads (Figure 1; elongated open water between fast ice and drift ice) may locally and temporarily contribute strongly to the maintenance of the cold halocline and to the renewal of Intermediate and deep water [e.g., Aagaard et al., 1981; Martin and Cavalieri, 1989; Cavalieri and Martin, 1994; Winsor and Björk, 2000]. Aagaard et al. [1981] estimated ice growth rates in divergent ice zones on the circum-Arctic shelves required to raise the local water salinity high enough to feed the cold halocline layer (assumed salinity of 33.5). The authors proposed that the most effective areas to produce cold halocline water are located between Spitsbergen and Franz Josef Land, west of Novaya Zemlya (Barents Sea), and west of Severnaya Zemlya (Kara Sea).

Figure 1.

(a) Modern configuration of circum-Arctic flaw leads between fast ice and offshore drifting ice (stippled line) as derived from unpublished NOAA/advanced very high resolution radiometer and Landsat satellite images, Barnett [1991], Buzov [1991], Groves and Stringer [1991], Dethleff et al. [1993, 1998], Pavlov and Pfirman [1995], and Brigham [1996] and, based on estimates of ice thicknesses from 10 day integrated Russian ice charts for the period from 1972 to 1990 (Arctic and Antarctic Research Institute, St. Petersburg). The Laptev Sea is indicated by the bold dark gray-lined box. (b) Sections of the Laptev Sea flaw lead system used in this study. The early winter lead configuration in the inner Laptev Sea is marked by the stippled bold black line, whereas the mid-to-end winter lead position is indicated by the dark grey bold line. The gray a-to-b line in lead section A2 indicates the shelf cross section shown in Figure 2, whereas the black bold line c–d off lead sections A2 and A3 denotes the profile shown in Figure 6. Stippled bold line e–f in lead section A1 represents the position of a salinity profile from Rudels et al. [2000] discussed in chapter 4.2. Water depths were taken mainly from the Arctic Ocean bathymetry provided by Perry and Fleming [1986], which is one of the basic International Bathymetry Chart of the Arctic Ocean sources.

[5] Midttun [1985] reported 34.95 salinity bottom water west of Novaya Zemlya and attributed this phenomenon to winter freezing. This was corroborated by model results from Harms [1997] showing the connection between local polynyas and dense water formation off the west coast of Novaya Zemlya. Martin and Cavalieri [1989] and Cavalieri and Martin [1994] estimated the salt rejection and dense water production in Russian, Siberian, and western Arctic shelf polynyas (flaw leads) in more detail. According to their calculations, both Siberian and Alaskan Arctic polynyas together generate about 0.7–1.2 Sv of dense water annually. Melling [1993] found that flaw leads on the Mackenzie shelf contribute substantially to the annual shelf flux of roughly 0.04 Sv of dense water to the Arctic cold halocline. The flux is proportional to the fraction of the Arctic shelf area that the considered region represented. Winsor and Björk [2000] modeled circum-Arctic polynya ice formation events and related salt flux for a period of 4 decades. They found that the mean multidecadal polynya salt flux can sustain a flow of 0.2 Sv dense water, representing ∼30% of the flux estimated to maintain the Arctic halocline layer.

1.2. Laptev Sea

[6] The Laptev Sea is one of the broadest and shallowest circum-Arctic Ocean shelf seas and is also among the most controversially discussed Arctic seas in terms of dense water formation related to flaw lead and polynya ice production. Early on, Zubov [1945] proposed that the dense water produced in the Laptev Sea flaw lead may flow from the shelf down to greater water depths. Zakharov [1966] reported that the Laptev Sea flaw lead salinity regionally increases by 2–6 units compared to the surrounding shelf water through intense winter ice extraction. Aagaard et al. [1981] concluded that the inner Laptev Sea summer salinities “are generally too low to allow a winter production of water sufficiently saline to feed the halocline.” However, they also suggested that the outer shelf area might be a dense water source. Martin and Cavalieri [1989] and Cavalieri and Martin [1994] studied dense lead water production in the western and northeastern Laptev Sea but neglected extended lead sections on the central-southern and eastern Laptev shelf, as they were unknown to the authors at that time. Cavalieri and Martin [1994] came up with a total brine flux of 0.7–1.2 Sv from all circum-Arctic shelf polynyas investigated in that study, with ∼0.12 Sv (10%–17%) forming by brine rejection from the flaw leads in the western Laptev Sea along the eastern Severnaya Zemlya and Taymyr Peninsula coasts (Figure 1).

[7] Schauer et al. [1997] concluded that the Barents and Kara seas are the only source areas for shelf waters ventilating the Nansen Basin below the cold halocline and that winter shelf water from the Laptev Sea cannot contribute to layers deeper than the upper halocline. Controversially, Ivanov et al. [2004] and Ivanov and Golovin [2007] identified extended parts of the northwestern Laptev Sea shelf east off Severnay Zemlya as region where freeze-related dense brines (∼0.02 Sv) are cascading down the shelf slope and ventilating salt into deeper layers of the Nansen Basin.

[8] Winsor and Björk [2000] estimated a four decadal average annual salt flux of ∼32.5 × 1011 kg from 28 extended circum-Arctic shelf polynya areas, to which western, central southern, and northeastern Laptev Sea flaw leads contribute ∼28% (∼9 × 1011 kg). However, there is an exceptionally weak salt contribution from central southern Laptev flaw leads situated near the Olenek and Lena river mouths (Figure 1). Conversely, in a model study, Johnson and Polyakov [2001] identified the southern central shelf area as a source for the salinization of the Laptev Sea and for the alteration of the Eurasian Basin cold halocline water formation during the 1990s. The authors related this to enhanced dense brine rejection subsequent to strong ice growth in flaw leads. Dethleff [2010] reached a different conclusion following a detailed investigation of 46 circum-Arctic flaw leads. The author showed that the central southern and southeastern Laptev Sea flaw leads produce relatively little salt due to low initial salinities combined with reduced seasonal ice production rates as winter offshore fast ice development regularly closes the southeastern Laptev leads by November to late December [Dethleff et al., 1998].

1.3. Purpose of the Study

[9] The purpose of this study was to provide a detailed estimate of freeze-related salt production from the surface mixed water (SMW) in 14 individual Laptev Sea flaw lead sections based on historic ice formation rates (volumes) and salinity data. Furthermore, the theoretic contribution of the individual lead sections to the maintenance of the local cold halocline water (CHW; salinity, 34.20) and to the renewal of intermediate water (IMW; salinity, 34.75) and deep water masses (DW; salinity, 34.93) are investigated using two different idealized equation-based models (Figure 2).

Figure 2.

(left) Temperature and salinity profiles of the eastern Nansen Basin (adapted from Schauer et al. [1997] and Ivanov and Golovin [2007]). (right) Idealized models of dense lead water contribution to the cold halocline and to intermediate or deep water masses described in chapter 2. Light arrows indicate pathways of dense water in model A, whereas bold arrows denote the pathways of the rejected brines in model B.

[10] Although Gawarkiewicz and Chapman [1995], Chapman and Gawarkiewicz [1997], and Chapman [1999] and Gawarkiewicz [2000] developed a series of universal, complex, and dynamic numerical models of dense water formation and eddie-driven offshore transport in idealized polynyas on shallow, sloping continental shelves, the target of this study is to present rather simplified, more static model approaches easily allowing the comparison of the outcome with results of various other studies dealing particularly with the Laptev Sea region during the past 2 decades [Becker and Bjørk, 1996; Martin and Cavalieri, 1989; Cavalieri and Martin, 1994; Winsor and Björk, 2000; Johnson and Polyakov, 2001; Dethleff, 2010]. The above Gawarkiewicz and Chapman studies will be discussed in relation to the model results presented and to potential transport processes of dense water away from the Laptev Sea flaw lead system down the continental slope.

[11] Model A considers the rejection of salt (dense brines) through lead-ice extraction, and the remixing of the salt with shallow lead water until the salinity and density increases to the level where the water sinks to the deeper layers (CHW, IMW, DW). In contrast, model B assumes direct downward rejection of dense lead brines as “pockets” or “parcels.” These dense brine parcels sink through the shallow lead water without mixing until reaching the cold halocline boundary. Mixing occurs at this boundary quasi en route, producing dense water that contributes to the deeper water masses. The models are described in detail in section 2. The model results are presented in section 3 and discussed in section 4. The latter section further discusses the possible upwelling of higher saline Atlantic water during storm events particularly into western Laptev lead sections and its significance for the process of dense water production.

2. Material and Methods

2.1. Flaw Lead Ice Production

[12] The rates of winter ice production in 14 individual Laptev Sea flaw lead sections (Table 1 and Figure 1) were taken from Dethleff et al. [1998]. The authors quantified the initial ice formation over open water during the 1991/1992 winter season applying a combination of numerical heat flux modeling and satellite observations. For more details see the study cited.

Table 1. Surface Salinities (Fall and Winter) From Different Sources, as Well as Ice Thicknesses and Volumes From Dethleff et al. [1998] for the Entire Laptev Sea Flaw Lead System
Flaw Lead SectionSurface SalinitiesSeasonal Sea Ice Equivalent Thickness (cm)Ice Volume (km3)a
FallWinterOct. 1, 1991–Dec. 31, 1991Oct. 1, 1991–June 15, 1992Jan. 1, 1992–June 15, 1992
Northwestern Laptev Sea
A129.50b34.00c,d,e1113.90 59.04 
A227.00b34.00c,d1085.00 21.70 
A324.00b33.00c,d1035.50 17.60 
Southwestern Laptev Sea
B119.00b32.00c664.90 8.64 
B216.00b29.00c726.60 11.63 
Southern Central Laptev Sea
C117.00b29.00f1017.20 25.43 
C2g24.45f444.00  12.43
C35.00bh482.1015.91  
C412.00bh481.4015.89  
Southeastern Laptev Sea
D117.00bh154.504.94  
D220.00bh128.103.46  
D3g28.50f790.60  18.97
Northeastern Laptev Sea
E27.00b31.34f1306.90 26.14 
F28.00b32.00c,d,e901.80 16.23 
Total  10,332.40 258.01i 
MC/CMj31–34 3470.00 95.00k 

2.2. Salinities and Oceanography

[13] To estimate the potential salt rejection and dense water production in the individual Laptev Sea flaw lead sections more precisely, both fall and winter surface salinities were used for the calculations (Table 1). Salinities were largely taken from different Russian studies [e.g., Pavlov et al., 1994; Dmitrenko et al., 2005; Ivanov and Golovin, 2007] and from own measurements (Microprocessor Conductivity Meter, WTW LF 196).

[14] The oceanographic characteristics of the Arctic cold halocline layer regionally vary strongly [Becker and Bjørk, 1996; Anderson et al., 1994; Signorini and Cavalieri, 2002], and particularly the evolution of the Eurasian Basin Halocline water masses experienced some major changes through the past decades [e.g., Steele and Boyd, 1998; Johnson and Polyakov, 2001; Boyd et al., 2002]. According to Schauer et al. [1994], Anderson et al. [1994], Schauer et al. [1997], Dmitrenko et al. [2009], and Lenn et al. [2009], the stratification of northern Laptev Sea (Figures 1 and 2) water masses was defined for this study as follows: (1) surface mixed water (∼0–30 m): salinity <33.50, at the freezing point; (2) cold halocline water (∼30–130 m): salinity from 33.50 to 34.40 (±0.2), temperature <0°C; (3) intermediate water (∼130–1000 m): salinity from >34.40 (±0.2) to 34.90, temperature >0°C; and (4) deep water (>1000 m): salinity >34.90, temperature <0°C.

2.3. Idealized Model Prerequisites

[15] The present models (Figure 2) assume that the flaw lead water column is well mixed at any time, not stratified and allowing no lateral water remixing or intrusion/export, and is at freezing throughout. Well-mixed flaw lead water columns down to 30 m have been shown by Dethleff and Kuhlmann [2009] in the Kara Sea, and Dmitrenko et al. [2005] demonstrated high potential of flaw lead convection down to the shallow bottom in parts of the Laptev Sea. The models also presume that the frazil ice volume extracted and the brines rejected from the flaw lead are compensated by surface water of the assumed initial salinity. Thus, the model assumes that the entire ice volume used in the estimates is theoretically formed from seawater of the initial value (Table 1). All descending water masses in both models A and B are also considered to be at freezing point.

2.4. Salt Rejection and Dense Water Models

[16] Salt flux: After Maykut [1985] and Martin and Cavalieri [1989], the total salt flux ST over one season in a lead area can be described as

equation image

where ρi is the density of sea ice (0.92 × 103 kg m−3), VT is the total ice volume formed during the freezing period in the flaw lead, and Sw and Si are sea water and ice salinities, respectively. Sea ice salinity was estimated after Martin and Kauffman [1981] by the following relation

equation image

[17] Dense water: Winsor and Chapman [2002] assumed in a polynya study that all salt rejected by ice extraction is evenly vertically mixed with the water column below the lead but does not significantly mix with surrounding shelf water. Melling [1993] suggested that the entire water column beneath the lead must reach the desired salinity and be close to or at freezing to maintain to the cold halocline or to contribute to deeper water masses. Following Melling's suggestion, this requirement is assumed in model A in order to allow modified lead water to contribute to the cold halocline water and to intermediate water or deep water masses. This is practically achieved in the model by mixing of the dense brines (rejected salt) expelled in the lead subsequent to ice formation with ambient lower saline lead water, until the resulting water masses reach the required salinities allowing for contribution to deeper layers (see Figure 2).

2.4.1. Model A (Lead Mixing)

[18] The dense water mixing equation

equation image

was modified after Alfutis and Martin [1987] and Martin and Cavalieri [1989], where VSW represents the volume of dense water that is produced after instantaneous remixing of the rejected salt with ambient lead water of the initial salinity SL (Table 1 and Figure 1) until the water is salinized sufficiently to contribute either to the midlayers of (1) the cold halocline, (2) the intermediate water, or (3) the deep water (each referred as to SML as appropriate). SML has approximate midlayer salinities of 34.20 (cold halocline), 34.75 (intermediate water), and 34.93 (deep water). PSW is the density of the midlayer water masses (∼1034.5 kg m−3).

[19] In model A, the salt rejection and related dense water formation of Laptev lead sections A1, A2, A3, B1, B2, C1, E, and F were calculated (Table 2), using each the appropriate fall and winter salinity data and the annual ice formation rates (volumes) presented in Table 1. In sections C2 and D3, only winter salinities were used, whereas in sections C3, C4, D1, and D2 only fall salinities were applied, as these lead sections only occur temporarily either during winter or fall, respectively (Table 1 and Figures 1 and 3).

Figure 3.

Seasonal lead ice formation rates, salinities, and production (Sv) of 34.20 salinity cold halocline water, 34.75 salinity intermediate water, and 34.93 salinity deep water in (left) model A (lead mixing) and (right) model B (direct contribution). For comparison of sections A1–A3 between both model cases, the numbers in cubic kilometer represent the amounts of the appropriate dense water produced.

Table 2. Cold Halocline, Intermediate, or Deep Water Formation in the Lead Sections as Calculated From the Fall and Winter Salinities According to the Recipe Given in Model A (Lead Mixing)
Flaw Lead SectionSalt Flux (1011 kg) (Fall Salinity)Salt Flux (1011 kg) (Winter Salinity)Production of
34.20 Salinity Water34.75 Salinity Water34.93 Salinity Water
(103 km3/Sv) (Fall Salinity)(103 km3/Sv) (Winter Salinity)(103km3/Sv) (Fall Salinity)(103 km3/Sv) (Winter Salinity)(103 km3/Sv) (Fall Salinity)(103 km3/Sv) (Winter Salinity)
  • a

    Lead section not occurring during early winter.

  • b

    Lead section not occurring during mid winter.

Northwestern Laptev Sea
A111.012.70.226/0.0076.141/0.1950.203/0.0061.638/0.0520.196/0.0061.321/0.042
A23.74.70.050/0.0022.273/0.0720.046/0.0010.606/0.0190.045/0.0010.489/0.015
A32.73.70.026/0.0010.298/0.0090.024/0.0010.204/0.0060.024/0.0010.185/0.006
Southwestern Laptev Sea
B11.01.80.006/00.079/0.0030.006/00.063/0.0020.006/00.059/0.002
B21.22.10.006/00.039/0.0010.006/00.035/0.0010.006/00.034/0.001
Southern Central Laptev Sea
C12.74.70.015/00.087/0.0030.015/00.079/0.0030.015/00.077/0.002
C2a1.90/00.019/0.0010/00.018/0.0010/00.018/0.001
C30.5b0.002/00/00.002/00/00.002/00/0
C41.2b0.005/00/00.005/00/00.005/00/0
Southeastern Laptev Sea
D10.5b0.003/00/00.003/00/00.003/00/0
D20.4b0.003/00/00.003/00/00.003/00/0
D3a3.40/00.058/0.0020/00.053/0.0020/00.051/0.002
Northeastern Laptev Sea
E4.55.20.060/0.0020.176/0.0060.056/0.0020.147/0.0050.055/0.0020.140/0.004
F2.93.30.045/0.0010.145/0.0050.042/0.0010.116/0.0040.040/0.0010.109/0.003
Total32.343.50.447/0.0139.315/0.2970.411/0.0112.959/0.0950.400/0.0122.483/0.078

2.4.2. Model B (Direct Contribution)

[20] In contrast to model A (lead mixing), in model B (direct contribution), the salt flux ST is used to estimate a more direct contribution of dense lead brines (rejected salt) to cold halocline water, the intermediate water, or deep water without salt mixing into lower salinity ambient lead water. Therefore, Melling's suggestion is neglected, and the model follows the observations made by Muench et al. [1995], Weingartner et al. [1998], and Shcherbina et al. [2003] that brine plumes rejected through flaw lead and polynya ice extraction may go directly downward to the shelf bottom or to deeper water layers without significant remixing in the lead water body. Precisely, Muench et al. [1995] observed “sinking, brine-enriched water parcels” in Conductivity-Temperature-Depth profile time series even penetrating the cold halocline [Weingartner et al., 1998].

[21] All salt expelled in the lead areas in the model B case subsequent to ice extraction, and even if it is very little salt due to low lead salinities or ice formation rates, is assumed to descend directly downward as “brine pockets” to the upper boundary layer of the local cold halocline. Here it mixes with the ambient water until the resulting water masses reach salinities allowing for further sinking and contribution to the midlayers of (1) the cold halocline itself, (2) the intermediate water, or (3) the deep water.

[22] The equation

equation image

is a modified version of equation (3). All terms are defined as above, however, SH denotes the salinity of the upper boundary layer of the cold halocline in the Eurasian Basin [salinity of 33.5, after Aagaard et al., 1981], with which the directly downward rejected salt has to be mixed.

[23] Model B uses mostly fall salinities to estimate the dense water production (Figure 3B). Only lead sections C2 and D3 were computed with winter salinities, as these sections do not occur before mid winter (Figure 2). According to the above theoretical model assumptions, the entire salt volume rejected from lead-ice formation descends downward to the upper layer of the cold halocline, and does not remix and contribute to the increase of the SMW salinity of the lead. Control calculations (not shown) with winter salinity data revealed that the dense water formation estimated was similar in most lead sections as compared to the flux based on fall calculations. Therefore, both data sets were not averaged.

2.5. Closing Model Reflections

[24] Mixing of descending dense, cold lead water with warmer and higher salinity Atlantic water, as proposed by Aagaard et al. [1981] and Martin and Cavalieri [1989] in order to produce intermediate or deep water is not considered in this study. Mixing of descending shelf dense water with inflowing Atlantic water according to the 40:60 ratio proposed by the above authors would undoubtedly increase the production of (cooled and salinized) intermediate water, however, it would not reflect the theoretical (and realistic) contribution of the shelf lead brines to deeper water masses.

[25] Furthermore, partitioning or fractionation of dense lead brines to different deeper layers or the direct contribution of either halocline water to intermediate and/or deep water or vice versa (upward vertical mixing) is also not considered in the calculations. The models assume that the entire salt rejected in the lead sections moves downward either to the cold halocline, to the intermediate water or to the deep water.

[26] Conclusively, the models consider that all of the dense water formed by ice extraction goes into one of the three major water types (CHW, IW, or DW). This is done simply for comparison between total mass formation of the different water types and to show the general magnitude of Laptev lead salt rejection and dense-water contribution to deeper layers. As nobody knows precisely how the dense water is physically formed and partitioned into the different deepwater masses, the study consistently presents theoretical values of dense water production for each individual deeper layer.

3. Dense Water Production

3.1. Lead Mixing (Model A)

[27] The model A results of fall and winter calculations are both shown in Table 2. For illustration of the annual dense water formation (Figure 3, left), the results of fall and winter calculations from Table 2 were averaged (Table 3). As the freeze-related salt rejection and production of dense water substantially depend on the salinity of the initial source water, the averaged results are strongly dominated by the estimates derived from the significantly higher winter salinities (compare Tables 2 and 3) but do approximate the “valid” winter lead dense water rejection better than either the low fall salinities or the high winter salinities alone.

Table 3. Averaged Dense Water Production for Model A According to Table 2
Flaw Lead SectionProduction of
34.20 Salinity Water (103 km3/Sv)34.75 Salinity Water (103 km3/Sv)34.93 Salinity Water (103 km3/Sv)
Northwestern Laptev Sea
A13.184/0.1010.921/0.0290.759/0.024
A21.162/0.0370.326/0.0100.267/0.008
A30.162/0.0050.114/0.0040.105/0.003
Southwestern Laptev Sea
B10.043/0.0020.035/0.0010.033/0.001
B20.023/0.0010.021/0.0010.020/0.001
Southern Central Laptev Sea
C10.051/0.0020.047/0.0020.046/0.001
C20.010/0.0010.009/0.0010.009/0.001
C30.002/0.0000.002/0.0000.002/0.000
C40.005/0.0000.005/ 0.0000.005/0.000
Southeastern Laptev Sea
D10.003/0.0000.003/0.0000.003/0.000
D20.003/0.0000.003/0.0000.003/0.000
D30.058/0.0020.053/0.0020.051/0.002
Northeastern Laptev Sea
E0.118/0.0040.102/0.0040.098/0.003
F0.095/0.0030.079/0.0030.075/0.002
Total4.954/0.1581.755/0.0571.500/0.046

[28] The total annual fluxes of cold halocline water, intermediate water, or deep water from the entire Laptev Sea flaw lead system in model A amount to 0.158, 0.057, and 0.046 Sv, respectively (Table 3). Remarkably, 91% (0.143 Sv) of the calculated cold halocline water, 75% (0.043 Sv) of the intermediate water, or 76% (0.035 Sv) of the deep water are produced in lead sections A1–A3. Lead section A1 alone may annually contribute as much as 0.101 Sv (64%) of the cold halocline water, 0.029 Sv (51%) of the intermediate water, or 0.024 Sv (52%) of the deep water. This finding matches the Chapman [1999] statement that longer polynyas over deeper shelves tend to produce more dense water.

[29] The contributions of lead sections B1, B2, C1, C2, and D3 to deeper water masses are significantly lower than in the A sections, whereas E and F are in the range of section A3. The lead sections C3, C4, D1, and D2 do not contribute to the formation of dense water masses, because the entire salt flux is used simply for increasing the lead water salinity toward the required (but not reached) value for contribution to a deeper water mass. In this context, Chapman [1999] states that shallow polynyas could potentially produce very dense water (if the starting salinity requirements are fulfilled) but do generally not significantly contribute enhances brine volume fluxes.

3.2. Direct Contribution (Model B)

[30] In model B, all lead sections may potentially contribute substantially to the formation of cold halocline, intermediate, or deep water masses (Table 4 and Figure 3, right). This is due to the fact that the rejected salt is not used up for salinity increase of the lead SMW and that all rejected salt is firstly mixed into the upper cold halocline. After mixing with cold halocline upper boundary water (salinity of 33.5) the salt masses then contribute to the appropriate deeper layers.

Table 4. Dense Water Production According to the Recipe Given in Model B (Direct Contribution)
Flaw Lead SectionSalt Flux (kg, × 1011)Production of
34.20 Salinity Water (103 km3/Sv)34.75 Salinity Water (103 km3/Sv)34.93 Salinity Water (103 km3/Sv)
Northwestern Laptev Sea
A111.01.520/0.0480.851/0.0270.744/0.024
A23.70.511/0.0160.286/0.0090.250/0.008
A32.70.373/0.0120.209/0.0070.183/0.006
Southwestern Laptev Sea
B11.00.138/0.0040.077/0.0020.068/0.002
B21.20.166/0.0050.093/0.0030.081/0.003
Southern Central Laptev Sea
C12.70.373/0.0120.209/0.0070.183/0.006
C21.90.263/0.0080.147/0.0050.128/0.004
C30.50.069/0.0020.039/0.0010.034/0.001
C41.20.166/0.0050.093/0.0030.081/0.003
Southeastern Laptev Sea
D10.50.069/0.0020.039/0.0010.034/0.001
D20.40.055/0.0020.031/0.0010.027/0.001
D33.40.470/0.0150.263/0.0080.230/0.007
Northeastern Laptev Sea
E4.50.622/0.0200.348/0.0110.304/0.010
F2.90.401/0.0130.224/0.0070.196/0.006
Total37.65.169/0.1642.909/0.0922.543/0.084

[31] All lead sections in model B produce together either 0.164 Sv of cold halocline water, 0.092 Sv of intermediate water, or 0.084 Sv of deep water. Lead sections A1–A3, C1, C2, D3, and E–F are the biggest contributors to dense water formation (Figure 3, right). Sections A1–A3 together contribute 46% (0.076 Sv) of the cold halocline water, 47% (0.043 Sv) of the intermediate water, or 45% (0.038 Sv) of the deep water flux. Lead section A1 alone contributes annually either 0.048 Sv (29%) of the cold halocline water, 0.027 Sv (29%) of the intermediate water, or 0.024 Sv (29%) of the deep water. Section E is the second biggest contributor with 0.020, 0.011, or 0.010 Sv, respectively. Southwestern and southeastern lead sections such as B1, B2, C3, C4, D1, and D2 contribute significantly less to deeper water mass production.

3.3. Comparison of Model A and B

[32] The production of cold halocline water (0.164 Sv), intermediate water (0.092 Sv), or deep water (0.084 Sv) is significantly higher in model B than in model A (0.158, 0.057, or 0.046 Sv, respectively; compare Tables 3 and 4). This discrepancy is due to the direct salt contribution to the upper layer of the cold halocline and then, after mixing, to deeper water masses in model B and also to the salt consumption in model A through brine remixing for increasing the SMW lead salinity to the required cold halocline, Intermediate or deep water values. Averaging both model results, the entire Laptev Sea flaw lead may annually contribute either as much as 0.161 Sv of cold halocline water, 0.075 Sv of intermediate water, or 0.065 Sv of deep water to the Arctic Basin.

[33] In both model cases, lead sections A1–A3 are the major regional producers of 34.20 (cold halocline), 34.75 (intermediate), or 34.93 (deep) salinity dense water, and sections E and F in the northeastern Laptev Sea are the second biggest contributors (Figure 3). However, the total flux of cold halocline water from lead sections A1 and A2 in model A is about twice the flux of the model B estimates (Figure 3, highlighted by gray bar in the background, and Tables 3 and 4), which is related to the dominance of the average model A brine flux by the significantly higher dense water estimates based on the higher winter salinities (see Table 2). This dominance is caused by the consumption of the salt released in the fall salinity calculations for the increase of the lead water body salinity and by the fact that the winter salinities of 34.00 in lead sections A1 and A2 are already higher than those assumed for the upper layer of the cold halocline and thus are close to the required salinity of 34.20 for contributing to the mid layer of the cold halocline. Thus, very little salt is used up to increase the salinity of the lead water in the winter salinity-based calculations, and the salt expelled through lead ice extraction contribute “straight” to the midlayer of the cold halocline.

[34] Conversely, in model B (calculated only with the lower fall salinities; Tables 1 and 4) less salt is rejected through lead section A1 and A2 ice formation (Table 2), and additional parts of the directly downward rejected salt are used up to increase the salinity of upper cold halocline water masses from 33.50 to the required salinity of 34.20, so that less salt remains to contribute to the midlayer of the cold halocline than in the model A case. In contrast to the cold halocline water, the production rates of intermediate and deep water s in lead sections A1 and A2 are in the same range or nearly identical in both models (Figure 3 and Tables 3 and 4). In lead sections A3 through F, the model B fluxes of cold halocline water, intermediate water, or deep water masses exceed those of model A (Figure 3 and Tables 3 and 4). This is due to the fact that in model B all extracted salt ends up in deeper water masses, whereas in model A, the released salt is partitioned between surface and deeper water masses (compare Table 1).

3.4. Control Calculations of Model B

[35] There is a controversial discussion on the disappearance, subsequent partial recovery, and the maintenance of the Arctic halocline at least through the past decade [e.g., Steele and Boyd, 1998; Johnson and Polyakov, 2001; Boyd et al., 2002; Fer, 2009]. In order to test possible seasonal, interannual, or even decadal changes of dense water rejection from the Laptev leads potentially contributing to the Arctic halocline in adjacent basins, control calculations were carried out for model B related to varying upper-boundary salinity conditions of the regional cold halocline (Figure 4) using equation 4 and higher upper halocline salinities (33.80 and 34.00 [e.g., Anderson et al., 1994]) as proposed above (33.50 [Aagaard et al., 1981]).

Figure 4.

Control calculations of direct lead-brine contribution to the mid-cold halocline (34.20) as well as mid-intermediate (34.75) and mid-deep water (34.93) after brine mixing with different upper halocline salinity conditions.

[36] The calculations (Figures 4a–4c) show enhanced fluxes of cold halocline water (salinity, 34.20), intermediate water (34.75), or deep water (34.93) with increasing salinity of the cold halocline upper boundary layer. Because of remixing of the expelled lead salt with higher salinity water in the “preconditioned” upper cold halocline boundary layer, less salt will be used up for reaching the required salinities (34.20, 34.75, and 34.93), and thus more “excess” salt can go downward to contribute to the midlayers of the cold halocline, the intermediate water, or the deep water.

[37] Compared to the 33.50 salinity case (Figure 4a), a theoretic two-step salinity increase of the upper layer of the cold halocline over 33.80 to 34.00 (Figures 4b and 4c) would each cause a 90%–100% increase in 34.20 salinity water production, whereas the 34.75 or 34.93 salinity dense water formation increases between 15% and 35%, and about 15%–25%, respectively. A potential seasonal or interannual increase of the halocline salinity in the Laptev Sea may for example result from the inflow and upwelling of Atlantic water masses from the Central Arctic Ocean basin (which will be discussed below), or from intense lead-salt rejection subsequent to extreme, temporary freezing events.

[38] Figure 5 shows the 34.20, 34.75, and 34.93 salinity dense water fluxes for a continuously increasing salinity of the halocline upper boundary-layer using salt rejection of lead section A1 (11 × 1011 kg) (see Table 2) as an example. The production of dense waters increases with rising salinity of the upper boundary layer of the local cold halocline. Particularly, the production of 34.20 salinity cold halocline midlayer water reveals the strongest amplification when the salinity of the upper layer of the halocline approaches this value.

Figure 5.

Relationship between increasing salinities of the upper boundary of the local cold halocline and the formation of cold halocline water (salinity, 34.20), intermediate water (34.75), and deep water (34.93; all polynomial-fitted curves) in flaw lead section A1.

[39] The numerical models proposed by Gawarkiewicz and Chapman [1995], Chapman and Gawarkiewicz [1997], Chapman [1999], and Gawarkiewicz [2000] estimate polynya dense water formation based on the theoretical model outcome that eddie convection near the bottom is an efficient process to steer lateral brine transport into deeper water masses. Although Danielson et al. [2006] found significantly decreasing winter eddie transport activity and rather identified downstream advection of dense polynya water before eddies could establish, the above theoretical Chapman and Gawarkiewicz approaches may help to better explain the model assumptions and results provided in the present study.

[40] Model B of the present study (Figure 1) assumes the direct and sustainable descend of brine packages [Muench et al., 1995; Shcherbina et al., 2003]. These brines can then be laterally advected by near bottom eddies after reaching the buoyancy equilibrium if enough dense water has been accumulated in the polynya water body during the geostrophic adjustment phase [Gawarkiewicz and Chapman, 1995]. Gawarkiewicz [2000] proposed that polynya dense water eddies are capable to produce lateral brine injections within the depth range of the Arctic cold halocline. This mechanism could effectively support the undelayed lateral dense brine transport into the halocline as proposed in model B (Figure 1), and thus, drive the salinity increase of the upper boundary layer of the local cold halocline and related dense water fluxes (Figure 5).

[41] Chapman [1999] proposes a typical dense brine volume flux of ∼0.04 Sv per 100 km water body of a coastal polynya, dense water that could possibly maintain the cold upper halocline. Considering the dense water estimates based on local Laptev winter salinities (Table 2), the figure of ∼0.04 Sv is well reproduced particularly for Laptev flaw lead sections A1 and A2, which represent the deepest, and, thus, most productive, polynya sections in that shelf area (A1: length ∼530 km, CHW ∼0.195 Sv, CHW/100 km ∼0.037 Sv; A2: length ∼200 km, CHW ∼0.072 Sv, CHW/100 km ∼0.036 Sv). All other (much shallower) Laptev flaw lead sections produce by far lower dense water volume flux, which is also consistent with the findings by Chapman [1999].

[42] Chapman and Gawarkiewicz [1997] provide test calculations for the time period to reach density equilibrium in a polynya water body (see their equation (11)). The buoyancy equilibrium timescale is te = β(fWb/B0)1/2 for the time within which a polynya water mass reaches density equilibrium between the surface buoyancy flux and the balancing downstream bottom eddy buoyancy flux. Danielson et al. [2006] came up with an equilibrium timescale calculation for the St. Lawrence polynya located in the Bering Sea. According to that applied study the constant β is ∼3; f represents the Coriolis parameter (1.3 × 10−4 s−1), W is the width of the forcing decay region (where B0 decreases from a maximum value over open water to close to zero where accumulated surface ice stops the buoyancy flux) that is equivalent to the polynya width, b is the width of the polynya, and B0 denotes the time-averaged surface buoyancy forcing for temporarily recurring polynyas. Danielson et al. [2006] calculated equilibrium timescale for the 25 km wide St. Lawrence polynya between 15 and 30 days based on different threshold values of surface buoyancy flux (B0 = 4 × 10−7 m2 s−3, and B0 = 1 × 10−7 m2 s−3, respectively). In a study including parts of the Laptev Sea flaw lead system, Cavalieri and Martin [1994] calculated a buoyancy input from frazil ice growth to the underlying polynya water body of B0 = 2.7 × 10−7 m2 s−3. However, the authors did not present precise polynya widths so that no equilibrium timescales could be estimated.

[43] In the present study, the scaling of buoyancy equilibrium timescales is tested for the highly brine-productive Laptev lead sections A1 and A2 using the above empirical buoyancy surface fluxes [Cavalieri and Martin, 1994; Danielson et al., 2006] and considering a flaw lead width of 10 km [Dethleff et al., 1998]. For B0 = 1 × 10−7 m2 s−3, the timescale for reaching the buoyancy equilibrium in lead sections A1 and A2 is ∼13 days; for B0 = 2.7 × 10−7 m2 s−3, the timescale is ∼8 days; and for B0 = 4 × 10−7 m2 s−3, it is ∼6 days. In conclusion, besides parameters such as salinity and ice formation rates, timescales of polynya buoyancy equilibrium response relate strongly to the width of the flaw lead or polynya system, and polynya dense water productivity also relates to water depth.

4. Discussion

4.1. Flaw Lead Dense Water Formation

[44] Aagaard et al. [1981] surveyed data from the literature to assess the potential salt rejection through intensive ice formation from high saline waters in shallow Arctic shelf seas. The authors calculated that an ice growth of 1–4 m in different regions of the Laptev Sea is required to raise the salinity of the summer water column to a value of 33.5. Martin and Cavalieri [1989] and Dethleff et al. [1998] showed that as much as 13–16 m of new ice can be formed in different Laptev Sea flaw lead sections during one winter, which is sufficient to produce dense water even if starting salinities of ∼28–32 are considered [Dethleff, 2010].

[45] According to results of long-term summer field studies (1977–1989 and 1993) reported by Churun and Timokhov [1995], bottom waters with temperatures as low as −1.97°C and salinities of as much as 34.81 were trapped in closed bathymetric depressions below lead section A3, which is classified as one of the biggest dense waters producers in this study. The salt concentration of these water masses by far exceeded the salinization of 33.5 assumed by Aagaard et al. [1981] for this area and was interpreted by Churun and Timokhov as remnants of dense brines formed in the local flaw leads during the previous winter. The authors assumed that the saline waters might have contributed to intermediate and deep water formation using the former, and now submarine, Kathanga River valley as a possible downslope pathway.

[46] Table 5 shows the potential contribution of the Laptev Sea flaw lead to Arctic cold halocline, intermediate, or deep water s in comparison to various computing, modeling, and predictive studies. According to the comparison, all individual Laptev lead sections together may contribute between 6% and 23% to the circum-Arctic lead brine fluxes maintaining the cold halocline, whereas lead sections A1–A3 alone may contribute at least 4%–16%. Here, the present estimate of 0.11 Sv well reproduces the Cavalieri and Martin [1994] calculation of 0.12 Sv (averaged form their Table 7). Alternatively, the entire Laptev lead may contribute a portion of 9%–16% to the Arctic intermediate water annually produced on the entire shelves [Becker and Bjørk, 1996] or rejected from leads between the Barents and Laptev seas [Martin and Cavalieri, 1989], respectively, whereas sections A1–A3 alone could contribute 5%–9% to that type of water mass. Finally, all individual Laptev lead sections together could produce 30% of the deep water formed in the leads between Spitsbergen and the eastern Laptev Sea, and sections A1–A3 could contribute 18% to the deep water in that region. The above comparison shows that besides the Barents Sea [e.g., Cavalieri and Martin, 1994; Schauer et al., 1997], the Chukchi Sea [Winsor and Björk, 2000], and parts of the Beaufort Sea [Dethleff, 2010], particularly the northwestern Laptev lead sections A1–A3, are among the biggest dense water-producing recurrent leads in the entire Arctic Basin.

Table 5. Dense Water Production Rates of the Laptev Sea Flaw Lead Compared to Entire Arctic Shelf Flux Rates
SourceFlux SvFlux Fraction in %
Dense Water SalinityValue or RangeEntire Laptev LeadLead Sections A1–A3
  • a

    Averaged values from model A and model B.

  • b

    Four year average.

Cold Halocline Water
This studya
   Entire Laptev Sea flaw lead34.200.16--
   Lead sections A1–A334.200.11--
Other calculations
   Aagaard et al. [1981] (cold halocline fed from circum-Arctic shelves)33.5–34.52.5∼6∼4
   Cavalieri and Martin [1994] (Barents, Kara, Laptev, East Siberian, Chukchi, Bering, and Beaufort flaw leads)∼33.000.7–1.213–239–16
Predictions
   Bjørk [1989]∼34.001–1.511–167–11
   Bjørk [1990]∼34.001.2139
Intermediate Water
This studya
   Entire Laptev Sea flaw lead34.750.075--
   Lead sections A1–A334.750.043--
Other calculations
   Martin and Cavalieri [1989] (Svalbard, Franz Josef Land, Barents Sea, Kara Sea, and Laptev Sea leads)b34.750.47169
Model results
   Becker and Bjørk [1996]∼34.680.5–0.89–155–9
Deep water
This studya
   Entire Laptev Sea flaw lead34.930.065--
   Lead sections A1–A334.930.037--
Other calculations
   Martin and Cavalieri [1989] (Svalbard, Franz Josef Land, Barents Sea, Kara Sea, and Laptev Sea leads)b35.10.213018

[47] After Dethleff et al. [1998, Table 4], an overall total of 9.63 m new ice (which is 30% of the local seasonal rate) was formed in lead sections A1–A3 in a 23 day freezing period of permanently offshore winds in December 1991. During these 3 weeks, which represented less than 10% of the annual freezing period, roughly one third of the seasonal local dense water was produced. This underscores that dense water formation is probably not a steady but rather an episodic process driven by time-dependent, severe freezing events in shelf leads. This was already shown for the St. Laurence Polynya [Schumacher et al., 1983; Danielson et al., 2006], where periods of offshore winds were followed by intense salinization events subsequent to polynya ice formation and was additionally proposed Arctic-wide by Winsor and Björk [2000] and suggested by Weingartner et al. [1998] for Alaskan flaw leads.

[48] In terms of lead sizes compared to the entire Laptev shelf area available for lead dense water production (35,900 km2; see Dethleff et al. [1998]; Table 1), sections A1–A3 together (total: 9000 km2; same authors) again are among the most effective brine producers. The three sections represent only 25% of the total Laptev lead area considered, but in model A, they produce ∼91% of the cold halocline water, ∼75% of the intermediate water, or ∼76% of the deep water. In model B, the fractions of dense water formation in sections A1–A3 are lower (46%, 47%, or 45%, respectively). The study by Winsor and Björk [2000; Table 2 and 3] also showed that these lead sections are among the biggest circum-Arctic salt producers related to their size, and Johnson and Polyakov [2001, Figure 4] proposed enhanced brine rejection in that area subsequent to intense new ice formation during the decade from the late 1970s to the late 1980s.

[49] Johnson and Polyakov [2001, Figure 4, right] suggested also strong brine rejection and related substantial salinity increase of the Laptev Sea due to anomalously high ice growth in polynyas on the central southern Laptev shelf and, thus, finally designated the Laptev Sea as a source of Arctic salinity changes for the period from 1989 to 1997. However, brine rejection calculated for the central southern Laptev lead sections A3, B1–B2, C1–C3, and D3 in the present study does not support this finding, at least for the 1991/1992 winter season. The present calculations were based on end-winter salinity measurements and on mean literature data (see Table 1). Accordingly, the above lead sections produced only 6% of the total Laptev lead 34.20 salinity water during that particular season, with lead sections B1–B2 and C1–C3 (Tables 2 and 3) being among the weakest dense water producers, which supports the findings of Winsor and Björk [2000]. Conversely, Johnson and Polyakov [2001, Figures 4 and particularly 5] modeled precisely these shelf (lead) sections to have the highest positive salinity anomaly due to intense freeze-related brine rejection from 1992 through 1995.

[50] Lead sections D3, E, and F (total: 6200 km2; see Dethleff et al. [1998]; Table 1) also produce considerable dense water volumes, which lie in the order of the production rates of sections A2 and A3 (Figure 4). A comparison shows that in model B all three sections D3, E, and F together (representing ∼17% of the lead surface available for dense water formation) account for either ∼29% of the cold halocline water, ∼28% of the intermediate water, or ∼27% of the deep water flux entirely rejected from the Laptev lead. In model A, the production rates are significantly lower and amount ∼6%, ∼16%, or ∼15%, respectively.

[51] Arctic shelf brine formation through ice extraction from high saline source water is evident from different studies [Aagaard et al., 1981; Martin and Cavalieri, 1989, Melling, 1993; Cavalieri and Martin, 1994; Churun and Timokhov, 1995; Schauer et al., 1997]. Vertically homogeneously distributed salinities >33.50 associated to turbulent freezing processes were, e.g., recorded during late winter in near coastal areas of the Chukchi Sea (Aagaard et al., 1985, Figure 6; Weingartner et al., 2005]), in the Canadian Beaufort Sea [Melling, 1993, Figure 5], and in parts of the Kara and Laptev seas [Pavlov et al., 1994, Figure 2.16; Churun and Timokhov, 1995, Table 1 and Figures 3–5]. This vertical homogeneity points to a well-mixed water column and suggests a constant vertical downward motion of dense brines toward the cold halocline in areas of intense shelf ice formation, as proposed by Melling [1993] and considered in model A of the present study.

[52] However, the assumption made by Melling [1993] that the entire lead water column must reach the desired salinity to contribute to the maintenance of the cold halocline may be questioned. If this assumption reflects the natural situation, all Laptev leads (or any other lead/polynya) potentially contributing to deep water formation must have late winter salinities at least equivalent to the salinity of the local upper halocline (provided that river discharge only partly freshens the shelf water throughout the winter), and this is evidently not the case (see Table 1). This means that all leads that do not reach late winter salinities of at least 33.50 must contribute “directly” to deeper water masses as proposed in model B by the production of dense water parcels. This conclusion is supported by the observation of vertically sinking, cold, salt-enriched water parcels as well as laterally moving shelf bottom brine tongues beneath Beaufort Sea leads [Muench et al., 1995], and by well defined, freeze-modified dense water plumes creeping from the Chukchi shelf over the slope and penetrating into the local cold halocline with only little lateral mixing during crossing of the shelf break [Weingartner et al., 1998].

[53] In conclusion, various processes such as continuous polynya dense water accumulation and buoyancy driven shelf-break eddie flux (Chapman and Gawarkiewicz studies), advective trans-shelf flow [Danielson et al., 2006], constant dense water mixing [Martin and Cavalieri, 1989; Cavalieri and Martin, 1994; Dethleff, 2010], and direct downward salt package flux [Muench et al., 1995; Shcherbina et al., 2003] may be responsible for final polynya dense water transport into deeper water masses. The dense water contribution from the Laptev flaw lead system appears to be driven by an unknown combination of all these processes. Model A (lead mixing) of this study may be considered as steady “background” mechanism of cold brine rejection subsequent to continuous daily ice formation, and model B (direct downward rejection of salt pockets and plumes) is instead attributed to highly salt-productive freezing events during intense short-term storm periods. As both model concepts cover parts of the above literature approaches of dense water production and downward transport, averaging the numerical results of models A and B seems the proper way to describe the Laptev Lead dense water production in the present study.

[54] Further, the present model outputs produce total water volumes for separate water masses (CHW, IMW, DW), but the (still in detail unclear) reality is that dense water produced by brine rejection from ice formation is partitioned in some unknown proportion between these three different water masses. Where the new dense water ends up is some function of how it is produced, i.e., dense, cold parcels formed during intense freezing events may form chimney-like plumes and thus are more likely to end up in deep water, whereas a constant flux of dense water produced by steady ice formation in the lead might mix more with ambient water, and so may end up less dense in the cold halocline.

4.2. Atlantic Water Upwelling

[55] Sverdrup [1929], Coachman and Barnes [1962], and Mountain et al. [1976] report on indications of Atlantic water upwelling at different North American and Eurasian shelf sites. Aagaard et al. [1981] suggested from direct observations in the Alaskan Beaufort Sea that cooling of upwelling Atlantic Water is a relevant mechanism of dense water formation on circum-Arctic shelves. Melling [1993] also reports such processes connected to extended, recurrent flaw leads episodically forced open between 1979 and 1991 on the Mackenzie shelf of the Beaufort Sea.

[56] Dynamical arguments on how shelf break upwelling-favorable winds create upward flow of Atlantic water masses in ice covered Arctic coastal zones are delivered by various studies. According to Bakun [1990] the general mechanism of costal upwelling response is globally forced by Ekman transport driven by alongshore wind, which deviates surface waters to the right on the Northern Hemisphere and cause upwelling if the surface deviation is directed offshore. In a long-term, Arctic-wide numerical study, Yang [2006] identified enhanced winter offshore Ekman transport in the Beaufort and Chukchi seas and in parts of the Laptev Sea, driven by regional alongshore and oblique offshore wind. Carmack and Chapman [2003] found that the intensity of coastal upwelling in Arctic seas is also connected to the position of the ice edge in relation to the shelf break.

[57] Dethleff et al. [1998] and Dethleff [1995] report prevailing alongshore and oblique offshore wind in the western and northwestern Laptev Sea, precisely in that region where the strongest dense water-producing lead sections A1–A3 are situated (Figure 1), during November 1991 and particularly for a period of about three weeks during December 1991 (Julian Day 339–361). During the latter period, the wind was predominantly blowing from southwesterly directions with an average direction of about 210° (where 0° is from the north) in lead sections A1 and A2, and with a tendency toward more southerly directions at the end of that period (200°–150°). In lead section A3, the wind was blowing from an average direction of 225°. The wind opened the leads and caused strong ice production [Dethleff, 1995]. As parts of lead sections A1–A3 lie over water depths allowing for the fully development of the Ekman spiral and related water mass transport, the surface lead water must have been deviated to the right, i.e., offshore toward the east. This suggestion is supported by Yang [2006, Figure 9] who identified enhanced eastward winter offshore Ekman transport in the shelf region off lead sections A1–A3 during November and December.

[58] The offshore Ekman surface water transport (Figure 6A; see small 1 in parenthesis in upper right corner) may then have been compensated by upwelling of higher salinity and warmer Atlantic Water (Figure 6a; see 2 in parenthesis), which was already proposed for that area by Zakharov [1966]. Schauer et al. [1997; see their Figure 1c, and profiles V to III in Figure 6] shows increasing intrusion of Atlantic water along the inner Laptev Sea shelf break from east to west between 135°E and 115°E. Further toward the northwest off the Severnaya Zemlya coast, Rudels et al. [2000, Figure 1] identified in turn decreasing intensity of Atlantic water intrusion so that strongest upwelling of Atlantic water can be suggested for the inner western Laptev Sea shelf break off lead sections A2 and A3, where Yang [2006] identified strongest winter offshore Ekman transport.

Figure 6.

Possible remnants of the proposed “pumping scenario” of upwelling Atlantic water and down going cold, dense water may be traced from (a) temperature and (b) salinity profiles recorded by Schauer et al. [1997] at the western Laptev shelf break close to lead section A1 in August through October 1993. As can be traced in the dashed box of the temperature profile in Figure 6a, the cold halocline is bulged down to a depth of as much as 250 m close to the slope bottom [cf. Schauer et al., 1994, Figure 2].

[59] Subsequent to lead-ice extraction from the potentially further upwelled higher salinity water or through remixing of the upwelled water type with dense water of the cold halocline fed by lead brines, the rejection of salt and the downslope movement of dense water may have been substantially intensified (Figure 6a; see 3 in parenthesis). This hypothesized mechanism represents a temporary “pumping system” modifying upwelling high-saline, relatively warm Atlantic water to descending high salinity cold shelf water (i.e., CHW or denser).

[60] The rejected cold, dense waters flowing down the slope may have mixed with lower intermediate and upper deep water masses resulting in the pronounced bulge of higher salinity and moderately cold water visible at 1000 m depth in lead section A1 (Figure 6b, dashed box). Both, the downward extended cold halocline (Figure 6b, solid box) and the up-bulging deep water masses are not at freezing, which suggests that this scenario is a remnant of local conditions from the 1992/1993 winter season but may also indicate the establishment of a starting “pumping system” due to beginning ice formation in September/October 1993 east of Severnaya Zemlya. Eicken et al. [1994] reported flaw lead (polynya) occurrence and intense ice formation (∼25 cm within few days) in that area during the Schauer et al. [1997] fall 1993 investigation period.

[61] Ivanov and Golovin [2007, Figure 11] described a similar scenario for that region in the 1984/1985 winter season. Accordingly, downslope descending cold and dense water in the western Laptev Sea penetrated below the warmer underlying intermediate Atlantic water and mixing with upwelling of warm Atlantic Water leading to a final distortion of the initial stratification. The authors interpreted this scenario as a result of cascading of modified surface water masses, as undisturbed stratification remained in deeper layers.

[62] As existing knowledge of dense water production in the Laptev Sea subsequent to lead ice extraction and/or upwelling of Atlantic Water masses is still rather weak and also contradictive, the model scenarios and crude estimates presented here may increase our understanding of the processes of dense water mass formation. More detailed regional field work and modeling efforts are needed to better understand dense water formation in the Laptev Sea, which may be, at least in part, one of the most important shelf seas consistently contributing to the renewal of Arctic halocline and deeper water masses.

5. Conclusions

[63] 1. Two model scenarios were presented to calculate dense water formation in the Laptev Sea flaw lead. In model A, salt rejected by lead ice formation is remixed into the ambient lead water until the salinity reaches concentrations required for the contribution to the midlayers of cold halocline (salinity: 34.20), the intermediate water (34.75), or the deep water (34.93). In model B, the rejected salt directly (i.e., without lead mixing) descends downward to the upper layer of the cold halocline, where it is remixed until the required salinities for deeper water mass contribution are reached.

[64] 2. Both models predict intense dense water formation in parts of the Laptev Sea flaw lead system. Main production areas are located on the northwestern and northeastern shelf, whereas southern lead sections contribute rather little to local dense water formation. Averaging both model calculations, the entire Laptev flaw lead annually produces as much as 0.161 Sv of cold halocline water, or 0.075 Sv of intermediate water, or 0.065 Sv of deep water, which represents considerable fractions of the total annual dense water flux from the shelves to the deep Arctic Ocean.

[65] 3. Lead sections A1–A3 east of Severnaya Zemlya and Taymyr as well as sections D3, E, and F north of Kotelnyy Island are the most important dense water-producing areas of the Laptev Sea. Sections A1 and A2 lie in the range of the typical dense brine volume flux of ∼0.04 Sv per 100 km coastal polynya water body. Averaged from both models, sections A1–A3 (representing ∼25% of the entire Laptev lead area available for dense water production) contribute ∼69% of the annual cold halocline water, ∼61% of the intermediate water, or ∼61% of the deep water masses produced, respectively. Sections D3, E, and F (∼17% of the lead area) produce in average ∼18% of the cold halocline water, ∼22% of the intermediate water, or ∼21% of the deep water rejected in the entire Laptev flaw lead, respectively.

[66] 4. Control calculations of model B show that slightly enhanced salinities of the upper layer of the cold halocline (from 33.5 over 33.8 to 34.0) cause each a roughly 90%–100% increase in 34.20 salinity water production, whereas both the 34.75 and 34.93 salinity dense water formation increase only roughly between 15%–35% and 15%–25%, respectively. Decreasing upper halocline salinities would cause reduction of dense water formation.

[67] 5. It is conceivable that polynya dense water eddies produce lateral brine injections into the Arctic cold halocline thereby amplifying the dense water formation process. Buoyancy equilibrium estimates (time until offshore eddie-driven bottom brine flux may occur) for the highly brine-productive, narrow Laptev lead sections A1 and A2 are much shorter (∼13 d) than timescales described in literature for wider polynyas (∼30 d) so that buoyancy equilibrium strongly depends on the width of polynyas.

[68] 6. Compared to literature data of dense water formation in various parts of the Arctic shelves, the Laptev Sea flaw lead may contribute ∼6%–23% of the total annual flux of Arctic cold halocline water, ∼9%–16% of the intermediate water, or as much as ∼30% of the deep water. Lead sections A1–A3 alone may contribute ∼4%–16% of the annual cold halocline water flux, 5%–9% of the intermediate water, or as much as 18% of the deep water.

[69] 7. As the precise process of brine production in the Laptev Sea flaw lead system is unclear yet, a combination of both processes described in model A (lead mixing) and model B (direct contribution) is proposed to drive the dense water formation, namely by a mixture of permanent freeze-related brine rejection, and temporary strong downward salt expulsions through short-term extreme freezing events. The model outputs produce total water volumes either for CHW, IMW, or DW; however, proportions of individual water masses separation are unknown in reality. Besides lead contribution, Atlantic water upwelling seems another likely mechanism for dense water formation at least in the northwestern Laptev Sea.

Acknowledgments

[70] The author is thankful to Ed Kempema for improving the manuscript. Erk Reimnitz and Dirk Nürnberg are thanked for multiple help during the Laptev Sea field work in 1992. The author appreciates the helpful comments of two anonymous reviewers.

Ancillary