Remotely Operated Vehicles (ROVs) are positioned relative to an underwater acoustic positioning system: Ultra-Short BaseLine (USBL). The accuracy is affected by refraction artifacts, caused by the variations of the sound velocity in the water column. Accurate positioning would require continuous measuring of the sound velocity profile (SVP) which is unpractical and preferably to be minimized. It requires alternative or inversion methods to obtain SVP information. In this thesis, two inversion methods are introduced: HISOM (Hull In Situ Ocean Model) and OMBES (Overlapping MultiBeam EchoSounder).

HISOM analysis whether the SVP can be approximated via other sources: 1) a constant profile based on the in situ surface (ISS) sound velocity, or 2) a profile derived from ocean model data in combination with the ISS sound velocity. These simplifications inevitably introduce a refraction error but for fallpipe ROVs operating at small incident angles (e.g. θ<6°) this error may be within acceptable margins (e.g. threshold of 0.2 m). The applicability is assessed by estimating the horizontal refraction error with a ray-tracing technique using daily-mean SVPs, derived from freely-available ocean model data. A spatiotemporal quantification for the North Sea yielded maps of sea areas where SVP measurements are necessary, and areas where constant surface SVPs suffice. The latter are the shallow parts of the North Sea (<80 m), where the error is always smaller than 0.2 m. For deeper locations, the gradients of the SVP cannot be neglected. Then, the model-based SVPs can be used with or without the use of ISS sound velocity data. Comparing a collection of observed SVPs in the North Sea revealed that these daily-mean model-based SVPs are accurate enough for the positioning of fallpipe ROVs at least up to 370 m depth.

OMBES uses the synchronous overlap in depth measurements between two dual-head multibeam echosounders (MBES). In previous studies, the overlap is obtained by sailing two adjacent tracks with one MBES on a ship. Here, we propose to use two MBESs deployed on the same ship, thereby reducing the uncertainty of the ship’s motions that affect the quality of the depth measurements. The mean sound velocity can be inverted by mathematically minimizing the depth differences in the overlap. This technique completely minimizes the refraction error with frequent updates of the inverted mean sound velocity on the flight. Subsequently, the mean sound velocity can be used to locate ROVs that operate close to the seafloor in shallow water (<80 m), even for large incident angles (e.g. θ~65°). Simulations showed that the best performance of the inversion technique is established when maximizing the distance between the multibeam heads, and by inward-tilting one or two heads. Practically, it means the deployment of MBESs on either side of the ship, rather than one pair of MBESs at mid-ship.

With increasing access to reliable ocean data, HISOM methods can potentially be run on ocean forecast model data to assess where and when refraction artifacts become dominant for fallpipe ROV positioning. This automatic assessment tool supports the SVP measuring strategy in subsea rock installation projects. OMBES can potentially be used as monitoring tool by comparing near real-time updates of the inverted mean sound velocity with the measured SVP. This data-driven decision tool can assist when to take an additional SVP. OMBES also improves the accuracy of multibeam bathymetric surveys by automatic collection of SVPs.

The production of water masses formed by convection in the Labrador Sea (i.e. Labrador Sea Water, LSW) and its variability contributes to the variability of the Atlantic Meridional Overturning Circulation (AMOC). Several studies put the role of the Labrador Sea under renewed debate, and suggest a rather complex interplay between the production of the LSW, the boundary current and the eddy field. To this end, an increased effort is put in understanding the variability of the LSW, its export routes and associated export timescales. In this study, the effects of variations in boundary current strength on the export pathways of convected water masses are investigated. The same idealized eddy-resolving numerical model is used as Georgiou et al. (2019) which has proven to be capable of capturing the key dynamics of the Labrador Sea, like the annual cycle of convection, the process and timescales of restratification, and properties of the mesoscale eddy field. Model simulations are set-up with different scenarios of the density structure of the boundary current at inflow location (i.e. southern tip of Greenland). The variations result in respectively a 5% strengthening and 5% weakening of the boundary current, which corresponds to interannual variability of observed surface velocities. The model output demonstrates that boundary current variations start a chain of reactions, significantly changing the dynamics of the Labrador Sea. This has implications for deep convection processes in the interior of the basin and thus the export product. With a passive tracer analysis it is shown that convected water masses formed in the convection area are laterally steered along isopycnals by an eddy-induced shear flow from the interior towards the boundary current at the West-Greenland coast in deeper layers. A strengthening (weakening) of the boundary current yields a lighter (denser) water mass to be exported at shallower (deeper) layers out of the interior. The most intense entrainment into the boundary current occurs where both the density and depth of the convected water masses match the local water mass properties of the boundary current, and where eddies detach from the boundary current. The associated export timescales can be linked to the location where eddies detach, and to the strength of the eddy-induced shear flow. This study further highlights the implications for linking variability in the LSW production and export to AMOC variability as the total export of convected waters in the Labrador Sea is a mixture of multiyear convected waters. Based on density alone, measurements of water masses at the exit do not directly reveal the past-year dynamical state of the Labrador Sea. This emphasizes that a proper representation of mesoscale eddies in models is necessary for representing the export timescales and water mass properties of the LSW, and their response to changing forcing.