CC BY
12
RUSSIAN JOURNAL OF EARTH SCIENCES, VOL. 22, ES5001, doi:10.2205/2022ES000783, 2022
On spreading of Antarctic Bottom Water in fracture zones of the Mid-Atlantic Ridge at 7-8°N
I. Yu. Dudkov1'2, M. V. Kapustina1, and V. V. Sivkov1'2
Received 9 Oct. 2021; accepted 26 May 2022; published 24 September 2022.
A Data-Interpolating Variational Analysis in n-dimensions was used to describe a potential temperature distribution in the bottom layer of the fracture zones of the Mid-Atlantic Ridge at 7-8°N. This analysis was based on a new digital terrain model obtained by supplementing the STRM15+ bathymetry data with multibeam echo sounding data from the 33rd cruise of the research vessel Akademik Nikolaj Strakhov (2016) and oceanological data from the World Ocean Database, supplemented with CTD profiles and reversing thermometer data measured in scientific cruises of the Shirshov Institute of Oceanology, Russian Academy of Sciences in 2014-2016. A 2D model of near-bottom potential temperature distribution in the study area was calculated based on the analysis. The model allows us to propose the Antarctic Bottom Water propagation pattern through the Doldrums, Vernadsky, and Pushcharovsky fracture zones. It is shown that bottom water warms up when passing fracture zones from 1.4°C in Pushcharovsky Fracture Zone up to 1.6-1.7°C in Vernadsky Fracture Zone. Bottom water from Pushcharovsky and Vernadsky fractures propagates in two directions. Northernly, it propagates to the Doldrums Fracture Zone, where its temperature reaches about 1.9-2.0°C. Easterly, it flows along Pushcharovsky Fracture Zone and raising the temperature up to 1.8-2.0°C. We propose the absence of Antarctic Bottom Water's overflow with a temperature less than 1.8°C to the East Atlantic in the study area. KEYWORDS: Antarctic Bottom Water; Mid-Atlantic Ridge; fracture zones; bottom topography; multibeam echo sounding; potential temperature.
Citation: Dudkov, I. Yu., M. V. Kapustina, and V. V. Sivkov (2022), On spreading of Antarctic Bottom Water in fracture zones of the Mid-Atlantic Ridge at 7-8°N, Russ. J. Earth. Sci., 22, ES5001, doi:10.2205/2022ES000783.
Introduction
Antarctic Bottom Water (AABW) is the densest of the deep-water masses in the open ocean. The main region of AABW formation is the Weddell Sea shelf. Leaving the shelf, AABW flows into the South Atlantic through the South Sandwich Trench and the Scotia Sea [Seabrooke et al., 1971]. Here it is modified, mixing with the deep waters originat-
1 Shirshov Institute of Oceanology RAS, Moscow, Russia
2Immanuel Kant Baltic Federal University, Kaliningrad, Russia
Copyright 2022 by the Geophysical Center RAS. http://rjes.wdcb.ru/doi/2022ES000783-res.html
ing elsewhere on the Antarctic shelf and the deep circumpolar current. This mixing leads to a temperature gain of about 1°C [Morozov et al., 2010].
Leaving the Scotia Sea, AABW enters the Argentine Basin through two deep passages: the Georgia Passage [Locarnini et al., 1993] and the Shag Rocks Passage. Further, it propagates to the Brazil Basin through the Vema Channel [Morozov and Tarakanov, 2014] where, after overflowing the channel sill, it spreads almost to the entire Atlantic Ocean basins [Morozov et al., 2010, 2021; Whitehead and Worthington, 1982].
AABW enters the East Atlantic basins through the fracture zones of the Mid-Atlantic Ridge (MAR), mainly the Romanche Fracture Zone (FZ), the Chain FZ [Demidov et al., 2011; Frey et al.,
ES5001
1 of 18
Figure 1. A general scheme of the AABW propagation in the Atlantic Ocean (after [Morozov et al., 2010, 2012 2017]). Bottom topography from [Tozer et al., 2019]. The red square displays the study area boundaries.
2019; Mercier and Morin, 1997; Mercier and Speer, 1998] and the Vema FZ [Vangriesheim, 1980] (Figure 1). Following from this, minor fractures zones are also interesting as AABW pathways, including: Doldrums, Vernadsky, Pushcharovsky (prior to 2020, a nameless fracture at 7°28' N [Morozov et al., 2017; Pushcharovsky and Raznitsin, 1991]), Bogdanov, and Strakhov [Morozov et al., 2017].
The first evidence of possible AABW flow through the fracture zones at 7-8°N was found during the 39th-41st cruises of the research vessel (R/V) Akademik Sergey Vavilov in 2014-2016 [Morozov et al., 2018]. These studies included CTD (conductivity, temperature, and pressure) and ADCP (water current velocity) profiling at several oceanographic stations. However, while
39°W 36°W 33°W
0 25 50 75 100 km <2400 2400 3300 4200 5100 >5100
Figure 2. Study area. 1 - MBES survey site; 2 - CTD stations and 3 - high-precision thermometer measurements stations made in 2016 in the 33rd cruise of the R/V Akademik Nikolaj Strakhov; 4 - CTD stations made in 2014-2016 in the 39th-41st cruises of the R/V Akademik Sergey Vavilov; 5 - CTD stations from the WOD18. Bathymetry from [Tozer et al., 2019].
it is widely known that bottom topography plays a key role in the propagation of bottom water [Rhein et al., 1998], previous studies in the 7-8°N region do not factor in the effect of bottom topography. Given that consideration of the topography is a principal requirement for the spatial-temporal analysis of temperature inhomo-geneity in the bottom layer, no explicit scheme of AABW flow through the Doldrums, Vernadsky, and Pushcharovsky exists.
Therefore, using results of these previous works in conjunction with new data collected during the 33rd cruise of the R/V Akademik Nikolaj Strakhov (2016), this work aims to clarify the AABW pathways in the 7-8°N segment of the MAR using statistical analysis of the potential temperature distribution across the Doldrums, Vernadsky, and Pushcharovsky FZs taking into account the influence of bottom topography.
Materials and Methods
In order to track the spreading of AABW across the northern equatorial MAR, our approach consisted of a statistical analysis of the potential temperature field over three FZs: Pushcharovsky (7°28' N), Vernadsky (7°43' N), and Doldrums (8° 14' N). Potential temperature was derived from CTD profiling data which were analysed within the framework of new global bathymetric models [GEBCO, 2020; Tozer et al., 2019] and additional original multibeam echosounder survey data (MBES).
Variational analysis (interpolation) was applied to the dataset using the Data-Interpolating Variational Analysis in n-dimensions (DIVAnd) software [Barth et al., 2014]. DIVAnd is an open-source software written in Julia and widely used for data interpolation purposes. It can be used for in situ data analysis [Dai et al., 2020], mapping of the climate atlases and calculation of the oceanographic databases [Belgacem et al., 2021; Korablev et al., 2014; Troupin et al., 2010], and ocean currents research [Barth et al., 2021].
Oceanographic data. The CTD dataset was compiled from data from the World Ocean Database (WOD18) [Boyer et al., 2018] in the date range 2015-2018 with data collected during the scientific cruises of the Shirshov Institute of Oceanol-ogy, Russian Academy of Sciences (IO RAS). The IO RAS research cruise data included the 39th-41st cruises of the R/V Akademik Sergey Vavilov conducted in 2014-2016 [Morozov et al., 2017] and from the 33rd cruise of the R/V Akademik Niko-laj Strakhov conducted in 2016 (Figure 2, Supplementary material). Overall, the analyzed dataset included 61 CTD profiles (Figure 3).
Oceanographic measurements during three research cruises on the R/V Akademik Sergey Vavilov in 2014-2016 were acquired using the CTD profiler Sea-Bird SBE 19plus SeaCAT [Morozov et al., 2013, 2015, 2018]. The precision of the measurements was up to 0.005°C. The probe was calibrated in the Sea-Bird Electronics Laboratory, and additional intercalibration was carried out during two cruises with the high-precision reversing ther-
Figure 3. Scatterplot of the potential temperature values of the waters deeper than 3500 m at the 61 stations within 7-8°N segment of the MAR area.
mometer RTM 4002 X. The intercalibration results showed a mean difference between the probes of about 0.003°C.
The CTD profiles collected during the 33rd cruise of the R/V Akademik Nikolaj Strakhov (2016) were acquired using a Neil Brown Instrument Systems Mark III CTD profiler. The accuracy of the CTD profiler according to the specification is ~ 0.005°C for the temperature sensor and 6.5 dbar for the pressure sensor. Calibration of the CTD profiler's sensors was performed at the Atlantic Branch of the IO RAS. Empirical coefficients and formulas for the profiler's measurements correction were derived. Corrected measurements had mean conductivity, temperature, and pressure uncertainties of about 0.005 mSm/cm, 0.005°C, and 8 dbar, respectively. Additionally, the data of the high-precision reversing thermometers SIS RTM 4002 and barometers SIS RPM 6000X were used to verify the results of DIVAnd analysis, though the data itself were not included in the analysis. The thermometers were accurate up to ~ 0.003°C within the range between -2.0°C and 40°C. Barometers had an accuracy of about 0.1% of the water column pressure.
The depth and potential temperature were derived from the conductivity, temperature, and pressure measurements using the Thermodynamic Equation of SeaWater 2010 (TE0S-10). The equations are implemented in the Gibbs-SeaWater (GSW) Oceanographic Toolbox software [McDougall and Barker, 2011].
Bathymetry. During the 33rd cruise of the R/V Akademik Nikolaj Strakhov a bathymetric survey was carried out in the rift valleys of the Doldrums and Vernadsky FZs, the rift valley between them and areas of the Pushcharovsky FZ valley (see Figure 2). The survey's acquisition system consisted of the 12 kHz multibeam echosounder Reson 7150 for bathymetry data acquisition and the integrated positioning system Ap-planix POS MV, which included two GNSS antennas used for the position, and heading computation Trimble DSM132 and an inertial measurement unit used for attitude and heave measurements. Data acquisition and processing were done with PDS2000 v.3.7.0.47 software package. Water column speed of sound profiles used for correct-
ES5001
ing bathymetric measurements were derived from the temperature and salinity profiles collected in the study area (Supplementary material, stations ANS-33043-33046). In accordance with the 6th edition of IHO Standards for Hydrographic Surveys (IHO Standards for Hydrographic Surveys, Edition 6.0.0, (2020), https://iho.int/en/standards-and-specifications, July, 2021), an analysis of both total vertical and total horizontal uncertainties (TVU and THU, respectively) was done to evaluate the quality of the collected data. Analysis of the initial data showed that for the given survey the average indicators of vertical and horizontal variability constituted ~ 55 and ~ 220 m, respectively, which corresponds well with the IHO Standards for Hydrographic Surveys for mean depths of 4150 m (IHO Standards for Hydrographic Surveys, Edition 6.0.0, (2020), https://iho.int/en/standards-and-specifications, July, 2021).
To perform a DIVAnd analysis, we required continuous and accurate (as much as possible) survey of seabed surface in the entire research area. The study area, at present, lacks sufficiently high-resolution bathymetric data and the multibeam survey carried out during the 33rd cruise of the R/V Akademik Nikolaj Strakhov was patchy. Therefore, two appropriate global DTMs were used: SRTM15+v2 [Tozer et al, 2019] and GEBC02020 [GEBCO, 2020] grids. Both have a 15 arcsecond resolution and both are essentially a combined product of single-beam, multibeam, and satellite-altimetry bathymetry. To define which DTM had the best representation of the seafloor in this area, we made two junctions of the MBES survey surface with SRTM15+v2 and GEBC02020 surfaces. The junctions and their statistical analysis were made using QGIS software [QGIS Development Team, 2021] and Python [van Rossum and Drake, 1995] libraries - GDAL [GDAL/OGR contributors, 2021] and NumPy [Harris et al., 2020]. The junction procedure included: median filtering of the MBES DTM to match the resolution of both global DTMs; surface difference derivation between the MBES and global DTMs; a simple statistical analysis of the difference surfaces - calculation of the mean and standard deviation of the difference surfaces; and creation of the histogram graph of the difference value distribution.
Multidimensional variational analysis DI-VAnd. DIVAnd is a software tool for spatial interpolation (gridding) of oceanological parameters
ES5001
using in situ data [Beckers et al., 2014]. DIVAnd is the implementation of the Variational Inverse Method (VIM), which resembles Optimal Interpolation (OI) methods (e.g., spline or kriging) [Beckers et al., 2014; Troupin et al., 2012]. DIVAnd has the following advantages:
• It is able to take into account boundaries (isobaths or coastlines), sub-basins, and various constraints (e.g., advection constraint).
• It was purposely designed for oceanographical in situ data analysis.
• It can provide quality control of the analysis field and estimate the used data quality [Barth et al., 2014].
• DIVAnd is written in the modern programming language, Julia. It is easy to use and has a computationally effective gridding algorithm.
In the present work statistical analysis was performed on 55 horizons in the depth range of 35004600 m with vertical and horizontal resolutions of 20 m and 500 by 500 m, respectively. The analysis result was a regular three-dimensional (3D) grid for the 30 vertical layers (4000-4600 m) with a resolution of 500 x 500 x 20 m.
Determination of the water mass boundaries. The benthic thermocline between AABW and Lower North Atlantic Deep Water (LNADW) vanishes when water passes through the MAR fractures thus, in the East Atlantic the boundary between AABW and LNADW becomes difficult to distinguish [Mercier and Morin, 1997]. The proportion of AABW in the bottom layer of the East Atlantic near the study area is estimated at 10-20% [Johnson, 2008].
Consequently, the position of the upper boundary of AABW in the East Atlantic can be determined as the boundary from [Rhein et al., 1995, 1998] where a density anomaly of a4 = 45.90, which relates to temperatures of about 1.8°C, is assumed to be the boundary of the two water masses. This boundary is located at a depth of about 39004000 m at 7-8°N in the Guiana Basin west of the MAR. Besides, d = 1.8°C surface has been used as the AABW upper boundary in the western Atlantic in [Friedrichs and Hall, 1993; Friedrichs et al., 1994; Molinari et al., 1992]. In the East Atlantic, the d = 2.0°C isotherm at 11°N is accepted
dudkqy et al.: on spreading of antarctic bottom water
Table 1: DIVAnd Analysis Parameters for Selected Layers
Parameter
Depth layer, m
4000 4200 4400 4600
Initially estimated correlation length, km 75.57 72.89 61.49 189.68
Initially estimated Epsilon2 0.08
Optimized correlation length, km 429.18 917.63 774.17 91.99
Optimized Epsilon2 0.564 0.506 0.007 0.629
Weighting kernel size for closely 1 .5' x 1.5'
spaced measurements, arcmins
Estimation of the background field 1.823 1.706 1.636 1.595
value of the potential temperature, °C
Figure 4. Potential temperature distribution over the WOCE A06 line at 7.5°N [WOCE Hydrographie Programme, 2002]. Isopycnic referenced to 4000 m (0-4 = 45.90) and isotherms are shown. Inset - the WOCE A06 line location. Bathymetry from [GEBCO, 2020].
as the AABW upper boundary in Vema FZ [McCartney et al., 1991].
In the study area, the potential density 04 = 45.90 and the 1.8°C isotherm are located shallower than in the Guiana Basin, according to the potential temperature distribution over the World Ocean Circulation Experiment (WOCE) A06 line at 7.5°N [WOCE Hydrographic Programme, 2002] (Figure 4).
DIVAnd Epsilon2 and L estimation. The
key function of the DIVAnd tool is DIVAndrun, which requires the next inputs: in situ data repre-
sented by the anomalies of the analyzing parameter (value minus mean value of the potential temperature within a selected layer (background field value), Table 1), the analysis boundaries (mask), the correlation length parameter L, and the parameter Epsilon2, which is a qualitative estimation of the used dataset uncertainty. Epsilon2 and L parameters were defined based on potential temperature values and geographic position of the stations.
The initial parameter Epsilon2 is contextual and represents a qualitative assessment of the data and its uncertainty. The main factors influencing the uncertainty of the data within the study area are
temporal data irregularity and methodological error (different accuracy of the instruments used, difference in measurement techniques, etc.). We accepted the initial value of Epsilon2 as 0.08.
The initial correlation length parameter L was calculated using the DIVAnd_fitforhorlen function, which estimates the average correlation distance (in km) by determining the nature of the mutual influence of potential temperature measurements (i.e., the degree of their spatial coherence). This function does not consider the physical boundaries and analyzes only the dataset values and the initial parameter Epsilon2. Although it is an estimate, the initial parameter L obtained may approach reality if there are no boundaries between the measurements.
The DIVAnd_cv function optimized the primary parameters of Epsilon2 and L. The function is based on the cross-validation taking into account isobaths. As an illustration, the optimized Ep-silon2 and L parameters for some of the analyzed layers are presented in Table 1. As it turned out, the initial and optimal Epsilon2 parameters differed on average by less than an order of magnitude, while the correlation length parameter L differed by slightly more than an order of magnitude. Such variations may be caused by the analyzed dataset's low spatial and temporal homogeneity within a particular depth layer and the complexity of the bottom topography.
The optimized Epsilon2 values of the closely spaced measurements were weighted to avoid the negative impact on the interpolation result. The weighting kernel size was 1.5 x 1.5 arcmins
3 km2). The background field value of potential temperature was defined for each layer to calculate potential temperature anomalies.
Quality Control. The analysis quality control (QC) was performed using the QC DIVAnd functions: DIVAnd_residualobs (residuals analysis), DIVAnd_qc (standard cross-validation), and DIVAnd_aexerr (almost exact error field associated with the analysis grid cells, further - error field). QC included: (a) calculation of the difference between the in situ measurements and the analysis results at the measurement stations (i.e., residuals derivation); (b) cross-validation of the data. Observations with a cross-validation value greater than 2 were considered "suspicious", i.e., poorly compatible with neighboring measurements. Such
Figure 5. Scatterplot of residuals as a function of potential temperature. The dot color corresponds to the unitless observation quality value derived by cross-validation. In 6 layers (4000-4600 m in 100 m) there are 294 potential temperature values. Observations with a depth more than the DTM's depth were not included in the analysis and not shown on the graph.
inconsistencies could be a consequence of either a methodological error in the observations or inaccuracy of the seabed topography or a result of hy-drodynamic processes at the area of the "suspicious" measurement (e.g., temporal variability of oceanological characteristics); (c) variational analysis of the randomly located measurements and calculating the error field.
Based on (a) and (b) QC steps, we created a scatterplot of the residuals as a function of the potential temperature observations for 6 depth layers within the 4000-4600 m range with 100 m step (Figure 5). The average residual values were no more than ±0.05°C. Observations with residual values above the 0.05°C confidence intervals were considered suspicious. Possible causes included: inconsistency between the observations and neighboring measurements or poor optimization of the parameters Epsilon2 and L for the observations.
The QC of the potential temperature data and performed analysis was assumed to be successful - (a) and (b) steps, and scatterplot analysis did not reveal any significant issues with the chosen Epsilon2 and L parameters and the used dataset. Hence, the optimized parameters Epsilon2 and L and the observations dataset were deemed appropriate and the performed DIVAnd analysis of the 55th depth layers within the range of 3500-
Table 2: The Filtering Statistics
Grids Grid cells
Number of %-age of the Number of the %-age of the rejected
non-empty cells, cells of the rejected cells, cells of the
thousands initial number thousands initial number
Initial (unfiltered) DIVAnd 5173 (initial) 100% (78.41%) 0 0%
analysis grid
Filter-A grid 2268 43.84% 2905 56.15%
Filter-B grid 1686 32.56% 582 11.25%
Overall filtering stats 1686 32.56% 3487 67.40%
4600 m was considered acceptable. This depth range exceeded the investigated depth range (40004600 m), intending to remove the possible vertical inconsistency. The error field calculated in step (c) could be used if the analysis parameters, dataset, and therefore performed analysis are deemed acceptable. The error field grid had the same dimensions and resolution as the analysis grid and characterized the analysis uncertainty.
Filtering. As the final step filtration of the results of the analysis consisted of the removal of grid cells values that had a vertical inconsistency (the temperatures of the overlying cells were colder than the underlying ones (Filter-A)) and removal of those cell values for which the error field value exceeded 0.05°C and more (Filter-B). The filtering statistics are shown in Table 2.
Bottom temperature analysis. As a secondary product of the multidimensional analysis, we calculated the two-dimensional (2D) grid of the bottom potential temperature distribution using the same DIVAnd gridding algorithm. A new dataset for 2D analysis was computed by sampling the 3D grid's potential temperature values. The sampling criteria were: (a) cell values of the 3D grid should have a depth that differs by no more than 300 m from the DTM's depth; (b) this value should belong to the deepest cell of cell values selected according to the rule (a). The depth source was presented by the same combined DTM as for 3D analysis. The Epsilon2 and Correlation Length
L parameters used were 0.8 and 100 km, respectively.
Results and Discussion
Combined DTM. The result of the MBES survey performed in 2016 by R/V Akademik Nikolaj Strakhov was a DTM of the seabed of 100 x 100 m spatial resolution covering Doldrums, Pushcharovsky, and Vernadsky FZs (Figure 6).
The junctions performed between MBES DTM and two global DTMs - SRTM15+v2 [Tozer et al., 2019] and GEBCO2020 [GEBCO, 2020] resided by analyzing two grids of the depth value differences: (1) between MBES DTM and SRTM15+V2 grids; (2) between MBES DTM and GEBCO2020 grids (Figure 7) showed us next: (a) the mean depth difference for case (1) was 36% less than (2); (b) the standard deviation of the depth difference values distribution for case (1) was 21.5% less than (2). Both (a) and (b) revealed explicit evidence that SRTM15+V2 DTM is more representative of the seabed within the research area compared to GEBCO2020. So, in combination with MBES DTM, SRTM15+V2 provided the most accurate seabed surface representation available.
Flows of AABW in the fracture zones of
the MAR. At some stations the values of potential temperature were deemed "suspicious" by the DIVAnd analysis (Figure 8). Most of the "suspicious" values are related to the difference in the
Figure 6. The DTM based upon the bathymetric survey of the 33rd cruise R/V Akademik Nikolaj Strakhov overlayed on the SRTM15+v2 grid [Tozer et al., 2019]. Digits define the following seabed features: (1) Doldrums FZ; (2) Vernadsky FZ; (3) Pushcharovsky FZ; (4) rift valley between Doldrums and Vernadsky FZs; (5, 6) nodal depressions; (7) deep gap in the western part of the Vernadsky FZ; (8) the median ridge between the rift valley and Doldrums FZ; sills in western (9) and central (10) parts of the Vernadsky FZ, and (11) wide sill between Vernadsky and Pushcharovsky FZs.
Difference, m .....
0 50 200 400 > 400
75 100 km
m
Figure 7. Grids of the differences between MBES DTM from the 33rd cruise of the R/V Akademik Nikolaj Strakhov bathymetric data and the global DTMs (A) SRTM15+V2 [Tozer et al., 2019] and (B) GEBC02020 [GEBCO, 2020], and histograms of the depth difference values distributions.
Figure 8. Potential temperature distribution in Doldrums, Vernadsky, and Pushcharovsky FZs for (a) 4000 m, (b) 4200 m, (c) 4400 m, and (d) 4600 m layers. Areas of varying uncertainty of analysis results (error field) are shown by hatching. Yellow-marked measurements have been labeled as "suspicious".
potential temperature values at the closely located stations. For the "suspicious" values associated with the sills and narrow passages of fracture zones these differences could be explained by the complex hydrological conditions in these bathymetric features. Besides, such differences may be associated with temporal variability in the temperature associated with tidal currents [Morozov, 1995; Mo-
rozov et al., 2008]. For example, it is known that in the neighboring Vema FZ, the variability of the potential temperature at the 4000-m isobath was found to be 0.06°C [Demidov et al., 2007], while the diurnal variability of the AABW temperature [Demidov et al., 2020] reached 0.05°C. The position of the 1.45 and 2.00°C isotherms in the Vema FZ could vary daily within 0-300 m and 0-200 m, re-
spectively. This fact suggests that a similar pattern of variability may be observed in the neighboring fracture zones.
A few "suspicious" values are related to the fact that the depth of the measurement used in the analysis is greater than the DTM depth at this station, which is due to the lack of accuracy of the bathymetric data making it impossible to analyze these measurements.
Figure 8 shows the potential temperature distribution in several layers based on the results of the analysis. The uncertainty of potential temperature analysis increases in areas poorly supported by initial data. In addition, the analysis uncertainty increases where physical (isobath) boundaries are lacking.
The largest uncertainty is found in the northeast of the study area for the depth of 4600 m. This relates to the absence of the temperature measurements. Nevertheless, the initial data and the values of the parameters Epsilon2 and L allowed us to perform an analysis characterized by a sufficiently high degree of spatial consistency of the grid cell values with each other and with the surrounding relief. Filtering of the analysis results allowed us to eliminate both horizontal and vertical inconsistencies in the values of the 3D grid.
The results of the 2D analysis of the potential temperature distribution, taking into account bathymetry (Figure 9a), made it possible to develop a scheme of AABW propagation below 4000 m (Figure 9b). The minimal values of the potential temperature obtained at the oceanographic stations were also taken into consideration. The propagation scheme was made for "unmodified" AABW with a potential temperature up to 1.8°C [Rhein et al., 1995, 1998], and "modified" AABW with a temperature of 1.8-2.0°C.
In the western part of the Pushcharovsky FZ, the minimum values of potential temperature within the bottom layer (в < 1.4°C) were recorded at depths exceeding 4250 m. Sill 1 (Figure 9a, 37.08°W, depth 4290 m) prevents the penetration of the coldest water further eastward. At the same time, some fraction of the AABW flows into the Vernadsky FZ to the north over sill 2 (37.88°W, 4130 m). Both eastward of sill 1 and northward of sill 2, AABW in the bottom layer becomes slightly warmer, 1.5-1.6°C and 1.6-1.7°C, respectively.
After flowing over sill 1, the majority of AABW in the Pushcharovsky FZ spreads over sill 3 (36.73°W, 4580 m) and propagates into the Ver-nadsky FZ. A minor fraction of AABW propagates into the East Atlantic along the Pushcharovsky FZ and passes through sill 4 (4030 m), where it is likely that the flow splits into two branches: part of the flow goes eastward over sill 5 (4200 m), and another part spreads northward, towards the shallow sills 6-8 (4060-4160 m). In the eastern part of the FZ, the AABW reaches sill 9 (34°W, 4010 m). Antarctic waters with potential temperature less than 1.8°C cannot propagate past this sill.
As previously mentioned, a slightly colder AABW enters the Vernadsky FZ via sill 3 compared to sill 2 (Figure 9a). This water has a temperature of 1.5-1.6°C. Along the Vernadsky FZ itself, AABW with a temperature of 1.6-1.7°C enters from the west through sills 10 (38.53°W, 4360 m) and 11 (38.3°W, 4530 m). With such a combination of sources after sill 3, AABW not only continues moving eastward to 35.53°W, but also turns backward - westward, crossing sills 12-14 with depths up to 4410-4450 m. This multi-source AABW fills the basin below 6000 m deep. The temperature here is about 1.6-1.7° C.
From the area of the nodal depression, the AABW spreads northward (over sill 15) to the rift valley between the Doldrums and Vernadsky FZs where it most likely turns to flow along the north walls of the rift valley and the shallow sill 16 (4110 m), returning back to the area of the nodal depression. The absence of the densest water flow over the sill is reflected in the potential temperature measured over sill 16 (1.96°C) and beyond this sill (1.89°C), with temperatures within the rift valley reaching 1.8°C). The overflow across the sill between the rift valley and the Doldrums FZ (sill 16 in Figure 9a) is likely accompanied by a strong bottom current, so-called the underwater cataract Similar cataracts are described in [Morozov et al., 2012; Whitehead, 1998]. "Modified" AABW with a temperature greater than 1.8°C, crossing this sill, flows eastward to 38.2°W and westward over sill 17 (4170 m), along the Doldrums FZ. After flowing over sills 16 and 17, "modified" AABW flows further eastward along the Doldrums FZ into the East Atlantic (Figure 9b).
In the western part of the Doldrums FZ, there is a wide sill at a depth of approximately 3800 m.
39°0'W 38°0'W 37°0'W 36°0'W 35°0'W 34°0'W
39°0'W 38°0'W 37°0'W 36°0'W 35°0'W 34°0'W
Figure 9. Distribution of the bottom potential temperature in the Doldrums, Vernadsky, and Pushcharovsky FZs. a - DIVAnd analysis results, the main sills with its numbers and depths. The white dotted line shows the location of the oceanographic section; b - DIVAnd analysis results, stations with the minimum potential temperature at the bottom and AABW spreading scheme (solid arrows - water with a potential temperature of less than 1.8°C, dotted lines - with a temperature of 1.8-2°C). Isobaths - every 200 m.
This sill may block AABW in its westward migration. It is likely that water with temperatures less than 1.85-1.9°C does not leave the Doldrums FZ. It is also possible that water in the eastern part of the Doldrums FZ mixes with AABW coming both from the Pushcharovsky FZ and from the Vema FZ. This is indirectly indicated by the presence of the shallow sills 7 and 8 with depths of 40604160 m. Those sills may prevent the spread of water with temperatures below 1.9°C, and changes in the 1.9° C isobath depth over the section along the Pushcharovsky FZ east of 35°W (Figure 10). This may be associated with both the flow of AABW from the Vema FZ and with tidal processes [Moro-zov, 1995; Demidov et al., 2020].
The presented analysis suggests that water with a potential temperature of 1.8-2.0°C flows into the East Atlantic from the studied fracture zones. The DTM shows that to the south and to the south of this area, there are sills with depths of less than 3800 m, which prevent the flow of AABW eastward (Figure 9). To the north of studied fractures, water with a potential temperature of 1.85° C was recorded at a depth of 3955 m in the Arkhangelsky FZ to the west of the 3700-m sill [Morozov et al., 2017]. That indicates no overflow of AABW with a temperature less than 1.8°C to the East Atlantic.
Conclusions
The Data-Interpolating Variational Analysis in n-dimensions statistical analysis of the potential temperature distribution was based on the original DTM obtained by supplementing the STRM15+ DTM with data from the 33rd cruise of the R/V Akademik Nikolaj Strakhov. CTD data from the WOD18 [Boyer et al., 2018] from 1995, supplemented with data from IO RAS cruises (39th-41st cruises of the R/V Akademik Sergey Vav-ilov in 2014-2016 and the 33rd cruise of the R/V Akademik Nikolaj Strakhov (in 2016) were used as initial temperature data. A total of 61 oceanographic stations were included in the analysis.
The result of the potential temperature distribution analysis carried out with DIVAnd was a three-dimensional grid of the potential temperature with a spatial resolution of 500 x 500 x 20 m. The analysis QC allowed us to quantify the uncertainty in each individual grid cell. Filtering of grid cells on the basis of these QC allowed us to completely eliminate the areas of spatial inconsistency in the analysis and thus provide an objective model of potential temperature field distribution in the region of the Mid-Atlantic Ridge fractures at 7-8°N.
Figure 10. Potential temperature distribution over the section in Pushcharovsky FZ. Isopycnic referenced to 4000 m (white lines) and isotherms (black lines) are shown.
Based on the DIVAnd analysis, a 2D model of near-bottom potential temperature distribution in the study area was calculated. This model allows us to propose the pattern of AABW propagation through the Doldrums, Vernadsky, and Pushcharovsky FZs. The main flow of "unmodified" AABW (with potential temperature less than 1.8°C) in these fractures propagates through the Pushcharovsky FZ. Part of the water, overcoming sills up to 4030 m deep, warming up to 1.8°C, propagates eastward along this FZ to 34° W; part of the AABW penetrates into the Vernadsky FZ through transverse sills 2 and 3, where it mixes with the AABW coming into this FZ from the west. Then, AABW from different sources spreads northward along the rift valley (up to 5300 m deep) to the sills 16 and 17. Through these sills, only "modified" AABW spreads eastward and in the eastern part of the Doldrums FZ probably mixes with water outflowing from the Vema FZ and from the Pushcharovsky FZ.
Taking into consideration the bathymetry improves our understanding of the flow of AABW in Doldrums, Vernadsky, and Pushcharovsky FZs. AABW warms up when passing fracture zones from 1.4°C in Pushcharovsky FZ up to 1.9-2.0°C in Doldrums FZ. There is no overflow of AABW with a potential temperature less than 1.8°C to the East Atlantic in this area.
Acknowledgments. The expedition and the primary processing of the data obtained during the 33rd cruise of the R/V Akademik Nikolaj Strakhov (2016) were supported by the state assignment of IO RAS, no. FMWE-2021-0012. The analysis and interpretation of the data were funded by the project of the Russian Science Foundation no. 1917-00246. The data obtained during cruises 39th-41st of the R/V Akademik Sergey Vavilov were analyzed within the framework of the project of the Russian Foundation for Basic Research no. 20-0800246. The authors are grateful to Prof. E. G. Mo-rozov (IO RAS) for providing oceanographic data obtained during the 39th-41st cruises of the R/V Akademik Sergey Vavilov and to Prof. S. A. Do-brolyubov (Lomonosov Moscow State University) for providing thermometers for the 33rd cruise of the R/V Akademik Nikolaj Strakhov. The authors also express their gratitude to the participants of the 33rd cruise of the R/V Akademik Nikolaj Strakhov: I. A. Mironov (Lomonosov Moscow State University), V. M. Pyatakov, and A. B. Demenina (IO RAS) for their help in collecting of oceanographic and bathymetric data and the crew of the R/V for their assistance in carrying out the scientific work. The authors are grateful to Dr. D. V. Dorokhov (IO RAS) for fruitful discussion and valuable recommendations and to T. Glazkova (Royal Holloway, University of London) for English corrections.
Appendix A
Supplementary material
CTD-profiling and thermometers stations used in the study. Underlined are the depth (DTM) that do not exceed the measurement horizon.
Potential Bottom depth (DTM), m
No. Station Latitude, °N Longitude, °W Date temperature, °C Instrument depth, m
1 ANS-33043 07° 43.15' 38° 10.00' Nov. 30, 2016 1.775 4235 4391
2 ANS-33044 08° 13.45' 37° 59.81' Dec. 1, 2016 1.806 3954 4048
3 ANS-33045 08° 16.62' 38° 06.83' Dec. 1, 2016 1.927 4129 4333
4 ANS-33046 08° 15.60' 37° 59.73' Dec. 1, 2016 1.885 4495 4648
5 ASV-2556 07° 56.50' 38° 01.90' Oct. 10, 2014 1.700 4327 4298
6 ASV-2557 07° 43.30' 37° 25.00' Oct. 11, 2014 1.660 4623 4647
7 ASV-2558 07° 28.70' 36° 59.10' Oct. 11, 2014 1.508 4644 4666
8 ASV-2573 08° 15.80' 36° 53.40' Sept. 23, 2015 1.871 4866 4502
9 ASV-2574 07° 36.50' 36° 41.60' Sept. 23, 2015 1.571 4825 4745
10 ASV-2575 07° 49.20' 35° 32.50' Sept. 24, 2015 1.835 4073 4113
11 ASV-2576 07° 23.10' 35° 08.00' Sept. 24, 2015 1.758 4524 4216
12 ASV-2597 07° 23.00' 35° 08.00' Apr. 19, 2016 1.765 4526 4180
13 ASV-2598 07° 24.50' 35° 08.00' Apr. 19, 2016 1.765 4424 4233
14 ASV-2599 07° 28.10' 35° 01.50' Apr. 19, 2016 1.799 4370 4106
15 ASV-2600 07° 34.80' 34° 57.00' Apr. 20, 2016 1.839 4216 4088
16 ASV-2601 07° 46.40' 34° 57.10' Apr. 20, 2016 1.869 4413 4420
17 ASV-2602 07° 37.50' 36° 44.20' Apr. 21, 2016 1.564 4619 4578
18 ASV-2603 07° 37.50' 36° 42.20' Apr. 21, 2016 1.558 4625 4595
19 ASV-2604 07° 42.00' 38° 15.00' Apr. 21, 2016 1.606 4804 4885
20 ASV-2655 08° 16.10' 38° 25.20' Oct. 23, 2016 1.836 4034 3655
21 ASV-2658 07° 35.00' 37° 55.70' Oct. 24, 2016 1.671 4081 4065
22 WOD-3351448 07° 29.28' 38° 28.86' Aug. 20, 1983 1.260 4493 4760
23 WOD-7744331 07° 27.07' 35° 00.00' Feb. 22, 1993 1.956 3888 3963
24 W0D-7909476 07° 29.95' 38° 00.03' Sept. 17, 1995 1.436 4381 4434
25 W0D-7909478 07° 27.96' 36° 39.97' Sept. 18, 1995 1.678 4292 4344
26 W0D-7909480 07° 29.85' 35° 30.20' Sept. 19, 1995 1.829 4342 4208
27 WOD-7909481 07° 29.93' 35° 00.00' Sept. 19, 1995 1.943 3825 3941
28 WOD-7909565 07° 29.96' 37° 59.99' Apr. 24, 1996 1.379 4358 4434
29 WOD-7909570 07° 30.72' 35° 30.12' Apr. 25, 1996 1.827 4411 4213
30 WOD-7909571 07° 30.32' 35° 00.14' Apr. 26, 1996 1.932 3816 4012
31 WOD-8078420 07° 29.68' 38° 39.93' Sept. 17, 1995 1.297 4545 4591
32 WOD-8078423 07° 29.86' 37° 20.00' Sept. 18, 1995 1.428 4473 4575
33 WOD-8078426 07° 30.13' 36° 00.03' Sept. 18, 1995 1.761 4633 4706
34 WOD-8078643 07° 30.43' 38° 40.10' Apr. 24, 1996 1.268 4551 4467
35 WOD-8078644 07° 30.58' 37° 19.89' Apr. 24, 1996 1.386 4487 4706
36 WOD-8078645 07° 30.52' 36° 40.39' Apr. 25, 1996 1.525 5522 5085
37 WOD-8078646 07° 31.29' 35° 59.60' Apr. 25, 1996 1.750 4740 4865
(Continuation of Supplementary material) CTD-profiling and thermometers stations used in the study. Underlined are the depth (DTM) that do not exceed the measurement horizon.
Potential Bottom depth (DTM), m
No. Station Latitude, °N Longitude, °W Date temperature, °C Instrument depth, m
38 WOD-11957615 07° 30.00' 35° 04.38' Dec. 7, 2000 1.801 4384 4392
39 WOD-11957616 08° 06.07' 35° 04.12' Dec. 7, 2000 1.886 4901 5058
40 WOD-11957617 08° 29.99' 35° 04.32' Dec. 7, 2000 1.892 4569 4576
41 WOD-11957624 07° 22.61' 35° 43.93' Dec. 10, 2000 1.768 4870 4804
42 WOD-11957726 07° 26.84' 38° 19.32' June 14, 2002 1.264 4843 4905
43 WOD-11957727 07° 27.13' 37° 47.51' June 14, 2002 1.274 4686 4737
44 WOD-11957728 07° 27.81' 37° 19.13' June 15, 2002 1.405 4759 4572
45 WOD-11957729 07° 28.64' 36° 52.39' June 15, 2002 1.513 4524 4536
46 WOD-11957731 07° 25.02' 34° 17.33' June 16, 2002 1.783 4682 4508
47 WOD-11957919 07° 24.58' 34° 17.17' June 1, 2003 1.776 4705 4562
48 WOD-11957920 07° 28.36' 06° 46.98' June 2, 2003 1.536 4724 4800
49 WOD-13096323 07° 30.00' 39° 00.00' Apr. 30, 2010 1.374 4438 4743
50 WOD-13096332 07° 30.00' 38° 29.90' Apr. 30, 2010 1.297 4594 4683
51 WOD-13096341 07° 30.00' 38° 00.06' Apr. 30, 2010 1.424 4340 4434
52 WOD-13096354 07° 30.00' 37° 30.00' May 1, 2010 1.403 4671 4386
53 WOD-13096367 07° 30.00' 37° 00.00' May 1, 2010 1.692 4126 4103
54 WOD-13096377 07° 29.94' 36° 30.06' May 1, 2010 1.560 4730 4678
55 WOD-13096388 07° 30.06' 36° 00.00' May 2, 2010 1.766 4555 4705
56 WOD-13096397 07° 30.00' 35° 30.00' May 2, 2010 1.826 4301 4219
57 WOD-13096444 07° 30.00' 34° 00.00' May 3, 2010 1.929 4594 4176
58 WOD-13096458 07° 30.00' 33° 30.00' May 3, 2010 1.918 4418 5186
59 WOD-13096474 07° 30.00' 33° 00.00' May 3, 2010 1.966 4184 4740
60 WOD-13096496 07° 30.06' 32° 20.04' May 4, 2010 1.935 4516 4404
61 WOD-19080078 07° 29.92' 35° 00.07' May 12, 2015 1.872 3853 3968
62 ANS-33040 (T) 07° 43.20' 37° 22.20' Nov. 28, 2016 1.74 4376 4356
63 ANS-33041 (T) 07° 43.20' 37° 24.00' Nov. 29, 2016 1.73 4617 4550
64 ANS-33042 (T) 07° 44.40' 37° 21.00' Nov. 29, 2016 1.68 4572 4616
65 ANS-33043 (T) 07° 43.20' 38° 10.20' Nov. 30, 2016 1.71 4282 4391
66 ANS-33044 (T) 08° 13.20' 38° 00.00' Dec. 1, 2016 1.96 3937 4016
References
Barth, A., J. M. Beckers, et al. (2014), Divand-1.0: n-dimensional variational data analysis for ocean observations, Geoscientific Model Development, 7, No. 1, 225-241, Crossref Barth, A., C. Troupin, et al. (2021), Variational interpolation of high-frequency radar surface currents using DIVAnd, Ocean Dynamics, 71, No. 3, 293308, Crossref Beckers, J. M., A. Barth, et al. (2014), Approximate and efficient methods to assess error fields in spatial gridding with data interpolating variational analysis (DIVA), Journal of Atmospheric and Oceanic Technology, 31, No. 2, 515-530, Crossref Belgacem, M., K. Schroeder, et al. (2021), Clima-tological distribution of dissolved inorganic nutrients in the Western Mediterranean Sea (1981-2017), Earth System Science Data Discussions, 1-49, Crossref Boyer, T. P., O. K. Baranova, et al. (2018),
World Ocean Database 2018, 207 pp. NOAA, NESDIS.
Dai, J., H. Wang, et al. (2020), Observed
spatiotemporal variation of three-dimensional structure and heat/salt transport of anticyclonic mesoscale eddy in Northwest Pacific, Journal of Oceanology and Limnology, 1-22, Crossref Demidov, A. N., S. A. Dobrolyubov, et al. (2007), Transport of bottom waters through the Vema Fracture Zone in the Mid-Atlantic Ridge, Doklady Earth Sciences, 416, No. 7, 1120-1124, Crossref Demidov, A. N., A. A. Ivanov, et al. (2020),
Transport of deep and bottom waters through the Mid-Atlantic Ridge in the Vema fracture zone, Doklady Earth Sciences, 494, No. 1, 735-740, Crossref Demidov, A. N., E. G. Morozov, R. Y. Tarakanov (2011), Structure and transport of bottom waters through the chain fracture zone of the Mid-Atlantic Ridge, Russian Meteorology and Hydrology, 36, No. 8, 542-548.
Frey, D. I., E. G. Morozov, et al. (2019), Regional modeling of Antarctic Bottom Water flows in the key passages of the Atlantic, Journal of Geophysical Research: Oceans, 124, No. 11, 8414, Crossref Friedrichs, M. A., M. M. Hall (1993), Deep
circulation in the tropical North Atlantic, Journal of Marine Research, 51, No. 4, 697-736, Crossref Friedrichs, M. A., M. S. McCartney, M. M. Hall (1994), Hemispheric asymmetry of deep wa-
ter transport modes in the western Atlantic, Journal of Geophysical Research: Oceans, 99, No. C12, 25,165-25,179, Crossref GDAL/OGR contributors (2021), GDAL/OGR
Geospatial Data Abstraction Software Library, Open Source Geospatial Found., URL https://gdal.org. GEBCO Bathymetric Compilation Group (2020), The GEBCO-2020 Grid - a continuous terrain model of the global oceans and land, British Oceanographic Data Centre, National Oceanography Centre, NERC, UK.
Harris, C. R., K. J. Millman, et al. (2020), Array programming with NumPy, Nature, 585, No. 7825, 357-362, Crossref Johnson, G. C. (2008), Quantifying Antarctic bottom water and North Atlantic deep water volumes, Journal of Geophysical Research: Oceans, 113, C5, Crossref
Korablev, A., O. K. Baranova, et al. (2014), Clima-tological atlas of the Nordic Seas and northern North Atlantic, NOAA Atlas NESDIS, 77, 122. Locarnini, R. A., T.Whitworth, W.D.Nowlin (1993), The importance of the Scotia Sea on the outflow of Weddell Sea Deep Water, Journal of Marine Research, 51, No. 1, 135-153, Crossref McCartney, M. S., S. L. Bennett, et al. (1991),
Eastward flow through the Mid-Atlantic Ridge at 11 N and its influence on the abyss of the eastern basin, Journal of Physical Oceanography, 21, No. 8, 10891121, Crossref McDougall, T. J., P. M. Barker (2011), Getting started with TEOS-10 and the Gibbs Seawater (GSW) oceanographic toolbox, SCOR/IAPSO WG, 127, 1-28.
Mercier, H., K. G. Speer (1998), Transport of bottom water in the Romanche Fracture Zone and the Chain Fracture Zone, Journal of Physical Oceanography, 28, No. 5, 779-790, Crossref Mercier, H., P. Morin (1997), Hydrography of the Romanche and Chain fracture zones, Journal of Geophysical Research: Oceans, 102, No. C5, 10,37310,389, Crossref Molinari, R. L., R. A. Fine, E. Johns (1992), The deep western boundary current in the tropical North Atlantic Ocean, Deep Sea Research Part A. Oceano-graphic Research Papers, 39, No. 11-12, 1967-1984, Crossref
Morozov, E. G. (1995), Semidiurnal internal wave global field, Deep Sea Research Part I: Oceanographic Research Papers, 42, No. 1, 135-148, Crossref Morozov, E. G., R. Yu. Tarakanov (2014), The Flow of Antarctic Bottom Water from the Vema Channel to the Brazil Basin, Doklady Earth Sciences, 456, No. 2, 227-230, Crossref Morozov, E. G., A. N. Demidov, R. Yu. Tarakanov (2008), Transport of Antarctic waters in the
deep channels of the Atlantic Ocean, Doklady Earth Sciences, 423, No. 1, 1286, Crossref Morozov, E. G., A. N. Demidov, et al. (2010),
Abyssal Channels in the Atlantic Ocean: Water Structure and Flows, Springer Science & Business Media, Crossref
Morozov, E. G., R. Y. Tarakanov, et al. (2012), Abyssal cataracts in the Romanche and Chain fracture zones, Doklady Earth Sciences, 446, No. 2, 1211, Crossref Morozov, E. G., R. Yu. Tarakanov, H. van Haren (2013), Transport of AABW through the Kane Gap, tropical NE Atlantic Ocean, Ocean Science, 9, 825-835, Crossref Morozov, E. G., R. Y. Tarakanov, N. I. Makarenko (2015), Flows of Antarctic bottom water through
fractures in the southern part of the North Mid-Atlantic Ridge, Oceanology, 55, No. 6, 796-800, Crossref
Morozov, E. G., R. Y. Tarakanov, et al. (2017), Flows of bottom water in fractures of the North Mid-Atlantic Ridge, Doklady Earth Sciences, 4-74, No. 2, 653-656, Crossref Morozov, E. G., R. Y. Tarakanov, et al. (2018), Bottom water flows in the tropical fractures of the Northern Mid-Atlantic Ridge, Journal of Oceanography, 74, No. 2, 147-167, Crossref Morozov, E. G., R. Yu. Tarakanov, D. I. Frey (2021), Bottom Gravity Currents and Overflows in Deep Channels of the Atlantic Ocean. Observations, Analysis, and Modeling, 483 pp. Springer International Publishing, Crossref Pushcharovsky, Yu. M., Yu. N. Raznitsin (1991), Structure of the Doldrums Fracture Zone, Central Atlantic, 224 pp. Nauka, Moscow. (in Russian) QGIS Development Team (2021), Bottom water flows in the tropical fractures of the Northern Mid-Atlantic Ridge, QGIS, CC BY-SA. (http://qgis.osgeo.org) Rhein, M., L. Stramma, G. Krahmann (1998), The spreading of Antarctic bottom water in the tropical Atlantic, Deep Sea Research Part I: Oceanographic Research Papers, 45, No. 4-5, 507-527, Crossref Rhein, M., L. Stramma, U. Send (1995), The Atlantic deep western boundary current: Water masses and transports near the equator, Journal of Geophysical Research: Oceans, 100, No. C2, 2441-2457, Crossref
Seabrooke, J. M., G. L. Hufford, R. B. Elder (1971), Formation of Antarctic bottom water in the Weddell Sea, Journal of Geophysical Research, 76, No. 9, 2164-2178, Crossref Tozer, B., D. T. Sandwell, et al. (2019), Global bathymetry and topography at 15 arc seconds: SRTM
15+, Accepted Earth and Space Science (accessed on 20.01.2021), (https://topex.ucsd.edu/cgi-bin/get_ srtm15.cgi) Crossref Troupin, C., A. Barth, et al. (2012), Gen-
eration of analysis and consistent error fields using the Data Interpolating Variational Analysis (DIVA), Ocean Modelling, 52-53, 90-101, Crossref Troupin, C., F. Machin, et al. (2010), Highresolution climatology of the northeast Atlantic using Data-Interpolating Variational Analysis (DIVA), Journal of Geophysical Research: Oceans, 115, C8, Crossref
van Rossum, G., F. L. Drake, Jr. (1995), Python Tutorial (Vol. 620), Centrum voor Wiskunde en Informatica, Amsterdam. Vangriesheim, A. (1980), Antarctic bottom water flow through the Vema fracture zone, Oceanologica Acta, 3, No. 2, 199-207. Whitehead, J. A. (1998), Topographic control of oceanic flows in deep passages and straits, Reviews of Geophysics, 36, No. 3, 423-440, Crossref Whitehead, J. A., L. V. Worthington (1982), The flux and mixing rates of Antarctic Bottom Water within the North Atlantic, Journal of Geophysical Research: Oceans, 87, No. C10, 7903-7924, Crossref WHP (2002), WOCE Hydrographic Programme,
WHP: Physical oceanography during L'ATALANTE cruise 35A3CITHER1.2 on section A06. PANGAEA, Crossref
Corresponding author:
I. Yu. Dudkov, Shirshov Institute of Oceanology RAS, 36, Nahimovskiy Pr., 117997 Moscow, Russia. ([email protected])
