Project Topics
Get Complete Project Material File(s) Now! »
S-ADCPs configuration during RREX2015 cruise
VADCP were measured during the RREX2015 cruise from two S-ADCPs operating at 38 kHz (OS38) and at 150 kHz (OS150). As shown in Figure 2.4, the maximum depth reached by the pulses was 1300 – 1400 m for the OS38 and 200 – 250 m for the OS150.
Data Acquisition System VMDAS is used to configure S-ADCP data. The configuration parameters are specified in Table 2.1. The number of vertical cells (called bins) was set to 85 for OS38 and 38 for OS150. The vertical sizes of bins were 24 m for OS38 with the middle of the first bin at 47.06 m, and 8 m for OS150 with the middle of the first bin at 20.28 m. No data is available in the first 35 m for OS38, and the first 16 m for OS150, because of the delay between emission and reception. To avoid interferences, the two S-ADCP emissions were synchronized. The resulting ping rate was 4.27 seconds for both instruments. Pings were averaged by VMDAS over 2-minute periods referred hereinafter as 2-minute ensembles.
The two instruments can operate in unmodulated Narrow Band mode (NB) or in modulated Broad Band mode (BB). The NB mode allows long-range emission while BB mode allows higher precision velocity measurement at the expense of the range. For a given accuracy, the vertical resolution in BB mode is better than in NB mode. A drawback of the BB mode is its strong sensitivity to ambient acoustic noise and interference with other sonars (Firing & Hummon, 2010), but this was dealt with on board by synchronizing acoustic emissions. As seen in Table 2.1, OS38 was used in NB mode for maximal range. Because OS150 only reaches 200 – 300 m, high precision was preferred over depth range and the BB mode was used. The BB mode was also used for OS38 during Bottom-Tracking (BT) in shallow waters. In BT, one ping over two is used to measure the ship velocity with respect to ocean bottom, which allows estimation of S-ADCP misalignment.
S-ADCP data acquired during the RREX2015 cruise were processed with Cascade Version 7.0 software (« Chaine Automatisée de Suivi des Courantomètres Acoustiques Doppler Embarqués », http://www.umr-lops.fr/en/Technology/Software/Cascade-V7.1-a-matlab-software-to-process-Vessel-Mounted-ADCP-data) developed by LOPS (Laboratoire d’Océanographie Physique et Spatiale, Brest, France) since 1998. This software is designed to qualify, correct, fill gaps in, filter, and select final S-ADCP data acquired by VMDAS (file.STA).
The S-ADCP data processing was done in two stages summarized in Table 2.2 (Le Bot et al., 2011). The processing is done on the absolute flow velocity of 2-minute ensemble data points. In first stage, ETOPO1 bathymetry (Amante & Eakins, 2009) was used with statistical tests to detect doubtful or bad data. Then barotropic tides were removed based on tidal currents generated by the OSU tidal prediction software tpxo8.0 (Egbert & Erofeeva, 2002). The model resolution is 1/6° for the open ocean and it resolves the tidal components M2, S2, N2, K2, K1, O1, P1, Q1, M4, MS4, MN4, MM and MF. Finally, attitude, amplitude and misalignment corrections were estimated. The same steps were followed in a second stage except that the latter corrections were applied before data qualification and removal of the barotropic tides. At the end, filters were applied to the data and the gaps were filled in.
Attitude correction
The pitch and roll given by the ship navigation system were used by VMDAS for real time correction of S-ADCP measurements. We thus considered that the remaining attitude error, which depends on the position of the S-ADCP on the ship, was constant during the cruise. Cascade computed the remaining attitude angle between ship and S-ADCP from the mean vertical velocity averaged over the cruise. Indeed, vertical velocities should be of the order of 10−3 m s−1 and not affected by the ship motion. Without attitude correction, vertical velocities estimated from OS38 are positively biased (yellow zones in Figure 2.5, upper panel), with large value of mean vertical velocities averaged over the cruise (0.026 m s−1) and with a root-mean-square (RMS) of 5.4 10−3 m s-1. Assuming that the mean vertical velocity was induced by a projection of ship horizontal velocity onto the vertical, the attitude correction was obtained by dividing S-ADCP vertical velocity by the ship horizontal velocity. Indeed, for low angle values, the sinus of the angle can be considered as being a good approximation of the angle. The corrections are 0.3° for OS38 and 0.1° for OS150. The correction removes the bias of vertical velocities and was applied to our data set. For instance, once the OS38 data were corrected, the mean vertical velocity averaged over the cruise was 10-fold lower (−0.0048 m s−1) and the RMS (4.5 10−3 m s-1) was barely changed (Figure 2.5, lower panel).
Amplitude and misalignment correction
To correct misalignment and amplitude errors, S-ADCP data should be calibrated using a Water-Tracking (WT) method or a Bottom-Tracking (BT) method. The WT calibration minimizes the root-mean-square differences between ocean velocities estimated by S-ADCP and by GPS during ship accelerations and decelerations, assuming a constant ocean current in a reference layer during these periods. The BT calibration compares the ship velocity estimated by GPS, to the ship velocity estimated from the S-ADCP bottom ping. The latter calibration is most reliable even though it requires specific conditions. Indeed, the BT calibration must be realized in shallow water (the acoustic pulse has to reach the bottom), which was the case at the beginning and end of the cruise, as well as on the northern part of the Reykjanes Ridge.
An amplitude and misalignment correction was thus estimated using Bottom-Tracking (BT) data in Cascade. Coefficients of amplitude and misalignment corrections were respectively associated with the difference of amplitude and direction between GPS and BT ship velocities for rectilinear motion and uniform speed of the ship. These computations should only take into account data recorded while the ship was moving (ship velocities > 2.5 m s−1). Indeed, linear regression of BT versus GPS ship velocities used for the determination of amplitude correction in Figure 2.6 highlights the outliers at low ship speed. Statistical tests were implemented in Cascade to remove outliers. We considered data for which Cship> 2.5 m s−1 and data for which the amplitude differences between BT and GPS ship velocities were less than 2.7 times the standard deviation. By following this procedure, we obtained a correlation coefficient of 0.99 between the BT and GPS ship velocity estimates. The amplitude correction was estimated as the slope of the linear regression between the two estimates (Figure 2.7, Table 2.5). Similar statistical tests were implemented for the misalignment computation (Figure 2.8, Table 2.5).
To refine the misalignment correction, OS38 data were compared with OS150 data for several corrections (Table 2.6). The aim was to determine whether these two data sets were compatible and if varying the misalignment correction within the confidence interval minimized a possible bias between these two data sets.
Without misalignment correction (0/0 in Table 2.6), there is a negative bias of -0.0099 cm s−1 between the two S-ADCP ocean velocity estimates averaged between the surface and 250 m.
The OS38 alignment has then a positive trigonometric angle α with OS150, which means that OS38 orthogonal velocities are on the left side of OS150 orthogonal velocities.
Applying the misalignment corrections provided by Cascade (0.07 for OS38 and -0.06 for OS150 in Table 2.6) divides the bias between OS38 and OS150 by three. Nevertheless, velocities perpendicular to the ship track still have a positive bias of 0.0032 m s−1. By varying the misalignment corrections of OS38 and OS150 within their respective confidence intervals, we found that the smallest biases (10−3 m s−1) were obtained with corrections 0.05/-0.04, 0.06/-0.04 and 0.05/-0.05. For these 3 pairs of corrections, bias values are very close. Figure
2.9 shows that the vertical average of cumulated differences is similar whatever the choice. The maximum difference is 0.00073 m s−1 for the Ridge Section, 0.00017 m s−1 for the North Section, 0.001 m s−1 for the South Section and 0.0001 m s−1 for the Ovide Section. Figure 2.9 shows the rapid convergence of these differences after averaging over about 100 km. Because the misalignment correction 0.05° for OS38 and -0.04° for OS150 are among the best choices for the full RREX2015 cruise, we applied these corrections to our data sets.
Filtering and gap filling
As described in Table 2.2, the corrections previously determined (attitude, misalignment and amplitude) are applied before filtering and gap filling the data.
To filter the data, a running average is used on 3 horizontal and vertical points following the [¼ ½ ¼] rule. Note that when the average includes more than 2 suspicious data (flag = 2), the resulting data is flagged as suspicious (Table 2.4). Missing data (white areas of the raw data in Figure 2.10, upper panel) are replaced by the average of two surrounded good data and are flagged as suspicious (flag = 2). Figure 2.10 (bottom panel) shows the filtering and gap filling of the OS38 data proposed by Cascade. Because the averaged velocity only changed by 10−4 m s−1 before and after this stage, the interpolation does not impact the final result and was applied to our data set.
Instrumental errors
Main instrumental errors come from the S-ADCP and GPS, which affect VADCP and VShip, respectively. Firstly, the S-ADCP error depends on the S-ADCP frequency, calibration, and configuration (such as the BT, bin size…) reported in Table 2.1. For OS38, profiles were mainly acquired in NB mode with a bin size of 24 m. As stated by the manufacturer, the measurement error on a single ping velocity is 23 cm s−1. By averaging the velocity profiles over 2-minute ensemble of 29 pings, the velocity error decreases to εOS38 = 23/√29 = 4.27 cm s−1. In BT, 2-minute ensemble was associated with 17 pings resulting in a larger velocity error of 5.58 cm s−1. The OS150 was configured in BB mode with a bin size of 8 meters, which is associated with a velocity error of 9 cm s−1 per ping. For a 2-minute ensemble of 29 pings, the velocity error decreases to εOS150 = 1.67 cm s−1. In BT, 2-minute ensemble was associated with 17 pings resulting in a velocity error of 2.18 cm s−1.
The second main instrumental error comes from the GPS. As shown by King & Cooper (1993), a 0.5° error in the ship heading affects the ship velocity of about 1%. For a ship moving at 5 m s−1 the induced error is 5 cm s−1. During the RREX2015 cruise, GPS HDS800 gave geographical coordinates of the ship. Because the GPS system is identical to Chafik et al. (2014), its accuracy was estimated by using the same calculation, which show a standard error of εGPS = 1 cm s−1 for 2-minute averaged GPS derived ship velocity.
Table of contents :
1 Introduction
1.1 Role of the North-Atlantic Ocean on the climate system
1.2 Mean circulation in the northern North-Atlantic Ocean
1.3 State of the art of North-Atlantic water masses
1.3.1 SubPolar Mode Water
1.3.2 Intermediate Water
1.3.3 Labrador Sea Water
1.3.4 Icelandic Slope Water
1.3.5 Iceland-Scotland Overflow Water
1.4 Impact of the topography on the North-Atlantic SubPolar Gyre: some key elements
1.4.1 Impact of topographic features on the flow
1.4.2 The Reykjanes Ridge
1.4.3 Cross-ridge flow
1.4.4 Along-ridge flow
1.5 The Reykjanes Ridge Experiment Project
1.6 Aims of the PhD thesis
2 Data and methods
2.1 Data
2.1.1 CTDO2 data
2.1.2 Lowered-ADCP data
2.1.3 Shipboard-ADCP data
2.1.3.1 S-ADCPs configuration during RREX2015 cruise
2.1.3.2 S-ADCP data processing
2.1.3.3 Instrumental errors
2.1.3.4 Conclusion
2.1.4 The AVISO data set
2.1.5 Atmospheric reanalysis
2.2 Computation of geostrophic transports
2.2.1 General Principle
2.2.2 Bottom triangles
2.2.3 Computation of the absolute reference velocities
2.2.4 Determination of the absolute reference layer
2.2.5 Conclusion
3 First direct estimates of volume and water mass transports across the Reykjanes Ridge
3.1 Introduction
3.2 Data and Methods
3.2.1 Description of the cruise
3.2.2 Data sets
3.2.3 S-ADCP referenced geostrophic velocities
3.2.4 Transport estimates and errors
3.2.5 Water mass characterization
3.3 Results: transports across the Reykjanes Ridge
3.3.1 The top-to-bottom cross-ridge flow
3.3.2 Water mass transports across the Reykjanes Ridge
3.4 Discussion
3.4.1 Circulation across the Reykjanes Ridge
3.4.2 NAC water masses
3.4.3 Subpolar Mode Water and Intermediate Water
3.4.4 Iceland-Scotland Overflow Water
3.4.5 Water mass transformations
3.5 Conclusion
4 Formation and evolution of the East Reykjanes Ridge Current and Irminger Current
4.1 Introduction
4.2 Data and Methods
4.2.1 Data sets
4.2.2 S-ADCP referenced geostrophic velocities and transport estimates
4.2.3 Water mass characterization
4.3 Results: Connections between the Iceland Basin and the Irminger Sea
4.3.1 Horizontal and vertical structures of the along-ridge currents
4.3.2 Hydrography of the eastern flank of the Reykjanes Ridge
4.3.3 Hydrography of the western flank of the Reykjanes Ridge
4.3.4 Circulation in density layers
4.4 Discussion
4.4.1 Large-scale circulation of the ERRC
4.4.2 Large-scale circulation of the IC
4.5 Conclusion
5 Deep through-flow in the Bight Fracture Zone
5.1 Introduction
5.2 Data and Methods
5.2.1 Bathymetry of the Bight Fracture Zone
5.2.2 Hydrographic sections
5.2.3 Deep-Arvor floats
5.3 Results: Through-flow in the Bight-Fracture Zone
5.3.1 The eastern sill of the Bight Fracture Zone
5.3.2 The rift valley of the Reykjanes Ridge
5.3.3 Exit of ISOW toward the Irminger Sea
5.3.4 Circulation of ISOW through the BFZ
5.3.5 Deep-Arvor float trajectories in the BFZ
5.4 Discussion
5.5 Conclusion
6 Conclusions and perspectives
6.1 Estimation of geostrophic transports
6.2 Intensity and structure of the subpolar gyre across the Reykjanes Ridge
6.3 Link between distribution of the cross-ridge flow and large-scale circulation of the subpolar gyre
6.4 Circulation and evolution of Iceland-Scotland Overflow Water across the Reykjanes Ridge
6.5 Formation, connection and evolution of the East Reykjanes Ridge Current
6.6 Connections between Irminger Current and cross-ridge flow
