Cross-Domain Submesoscale Eddy Detection Neural Network for HF Radar

With the rapid development of deep learning, the neural network becomes an efficient approach for eddy detection. However, previous work employs a traditional neural network with a focus on improving the detecting accuracy only using limited data under a single scenario. Meanwhile, the experience of detecting eddies from one experiment is not directly inherited from the detection model for other experiments. Therefore, a cross-domain submesoscale eddy detection neural network (CDEDNet) based on the high-frequency radar (HFR) data of the Nansan and Xuwen region is proposed in this paper. Firstly, a fundamental deep eddy detection architecture CDEDNet-0 is constructed with a fully convolutional network (FCN). Secondly, for solving the problem of insufficient labeled eddy data, an instance-based domain adaption method is adopted in CDEDNet-1 to increase training samples. Thirdly, for tackling the problem of unable to inherit previous detection experience, parameter-based transfer learning is incorporated in CDEDNet-2 for multi-scene eddy detection. The experiment results demonstrate CDEDNet-1 and CDEDNet-2 perform better than CDEDNet-0 in terms of accuracy. Meanwhile, eddy characteristics including eddy type, radius, occurring time, merger, and dynamic trajectory are analyzed for the Nansan and Xuwen regions.

Three mesoscale eddy detection and tracking methods: Assessment for the South China Sea

Complex topography and the Kuroshio eddy shedding process produce active mesoscale eddy activity in the South China Sea (SCS). Three eddy detection and tracking methods, the Okubo-Weiss (O-W), Vector-Geometry (V-G), and Winding-Angle (W-A) algorithms, have been widely applied for eddy identification. This study provides a comprehensive assessment of the O-W, V-G and W-A methods in the SCS, including their detection, statistical analysis, and tracking capabilities. The mean successful detection rates (SDRs) of the O-W, V-G and W-A methods are 51.9%, 56.8% and 61.4%, respectively. The O-W and V-G methods preferentially detect eddies with medium radii (1/2°-1°), while the W-A method is tend to be larger radii (>1°). The V-G method identifies an excessive number of weak (radius<1/3°) eddy-like structures in the SCS, accounting for 48.2% of the total eddy number. The highest mean excessive detection rate (EDR) of the V-G method biases the data on eddy number, probability and propagation direction. With the lowest mean successful tracking rate (STR), the O-W method might not be suitable for tracking long-lived eddies in the SCS. The V-G method performs well regarding the over-tracking issue and has the lowest mean questionable tracking rate (QTR) of 1.1%. Among the three methods, the W-A method tracks eddies most accurately, with the highest mean STR of 80.6%. Overall, the W-A method produces reasonable statistical eddy characteristics and eddy tracking results. Each method has advantages and disadvantages, and researchers should choose wisely according to their needs.

Characterization of the ocean mesoscale eddies in the Antarctic Circumpolar Current from in situ, model and remotely sensed data

&lt;p&gt;Mesoscale variability and associated eddy fluxes play crucial roles in the ocean dynamics, transport of water mass properties and ecology of the upper ocean. In the Southern Hemisphere, where the nearly zonal flow of the Antarctic Circumpolar Current (ACC) acts as a barrier to the direct poleward transport toward the Antarctica, the eddy flux across the ACC is the main mechanism that guarantees the heat budget and distributes physical and biogeochemical properties between subtropical and polar regions. We focused on a high dynamical region located between the South-West Indian Ridge and the South Pacific Ridge. In this area, the interaction between the ACC and the major bathymetric features produces relatively large values of eddy kinetic energy and eddy heat fluxes as well as a relevant forcing for the ACC path.&lt;/p&gt;&lt;p&gt;The aim of this study is to evaluate the actual efficiency of mesoscale eddies to exchange heat and other properties across the different ACC fronts and to describe the vertical properties of the eddies, their tracks and evolution. To this end, we used in-situ and satellite data in conjunction with a hindcast simulation from 1958 to 2018 performed with a 1/10&amp;#176; ocean biogeochemistry model.&lt;/p&gt;&lt;p&gt;Eddies are identified and tracked in both the model output and altimetry data while their thermohaline properties and vertical extension are described using model outputs and in situ data, which include available repeated XBT sections (i.e. New Zealand &amp;#8211; Antarctica and Hobart &amp;#8211; Antarctica) and Argo float profiles located inside these structures.&lt;/p&gt;&lt;p&gt;Thanks to the joint analysis of model and observational data, we are able to 1) assess the ability of the 1/10&amp;#176; ocean model of simulating the eddy field properties, and to 2) better interpret the spatial and temporal variability of the observed eddy characteristics in the larger and longer framework of the ocean simulation.&lt;/p&gt;

Understanding eddy field in the Arctic Ocean from high-resolution satellite observations

&lt;p&gt;The Arctic Ocean is a host to major ocean circulation systems, many of which generate eddies that can transport water masses and corresponding tracers over long distances from their formation sites. However, comprehensive observations of critical eddy characteristics are currently not available and are limited to spatially and temporally sparse in situ observations.&lt;/p&gt;&lt;p&gt;Here we use multi-mission high&amp;#8208;resolution spaceborne synthetic aperture radar (SAR) measurements to detect eddies over open ocean and marginal ice zones (MIZ) of Fram Strait and Beaufort Gyre regions. We provide the first estimate of eddy properties, including their locations, size, vorticity sign and monthly distribution during summer period (from June to October). The results of historical Envisat ASAR observations for 2007 and 2011 are then compared to Sentinel-1 and ALOS-2 PALSAR-2 measurements acquired in 2016 and 2018, to infer the possible changes in the intensity and locations of eddy generation over the last decade.&lt;/p&gt;&lt;p&gt;The most prominent feature of the obtained results is that cyclonic eddies strongly dominate over anticyclones. Eddies range in size between 0.5 and 100 km and are frequently found over the shelf and near continental slopes but also present in the deep basin. For MIZ eddies, the number of eddies clearly depends on sea ice concentration with more eddies detected at the ice edge and over low ice concentration regions. The obtained results clearly show that eddies are ubiquitous in the Arctic Ocean even in the presence of sea ice and emphasize the need for improved ocean observations and modeling at eddy scales.&lt;/p&gt;&lt;p&gt;A special focus is also given to infer eddy dynamics over the Arctic marginal ice zones. The use of sequential Sentinel-1 SAR images enables to retrieve high-resolution velocity field over MIZ on a daily basis and observe eddy-driven MIZ dynamics down to submesoscales. The obtained eddy orbital velocities are in agreement with historical observations and may reach up to 0.5-0.7 m/s. We believe that this information is critical for better understanding of the key dynamical processes governing the MIZ state, as well as for improving and validation of sea ice and coupled ice-ocean models.&lt;/p&gt;&lt;p&gt;The analysis of eddies in this work was supported by RFBR grant 18&amp;#8208;35&amp;#8208;20078. Processing and analysis of Sentinel&amp;#8208;1 and ALOS&amp;#8208;2 Palsar&amp;#8208;2 data were done within RSF grant 18&amp;#8208;77&amp;#8208;00082. Software development for data analysis in this work was made under the Ministry of Science and Higher Education of the Russian Federation contract 0555&amp;#8208;2019&amp;#8208;0001.&lt;/p&gt;

Mesoscale eddy characteristics in the Labrador Sea from observations and a 1/60° numerical model

&lt;p&gt;Mesoscale eddies play an important role in lateral property fluxes. Observational studies often use sea level anomaly maps from satellite altimetry to estimate eddy statistics (incl. eddy kinetic energy). Recent findings suggest that altimetry derived eddy characteristics may suffer from the low spatial resolution of past and current satellite-tracks in high-latitude oceans associated with small Rossby radii. Here we present results of an eddy reconstruction based on a nonlinear damping Gauss-Newton optimisation algorithm using ship based current profiler observations from two research expeditions in the Labrador Sea in 2014 and 2016. Overall we detect 14 eddies with radii ranging from 7 to 35 km.&lt;/p&gt;&lt;p&gt;In order to verify the skill of the reconstruction we used the submesoscale permitting NATL60 model (1/60&amp;#176;) as a reference data set. Spectral analysis of the horizontal velocity implies that the mesoscale regime is well represented in NATL60 compared with the observations. The submesoscale regime in the model spectra shows deviations to the observations at scales smaller than 10km near the ocean surface. The representation of the submesoscale flow further decreases in the model with increasing depth.&lt;/p&gt;&lt;p&gt;By subsampling the NATL60 model velocities along artificial ship tracks, applying our eddy reconstruction algorithm, and comparing the results with the full model field, a skill assessment of the reconstruction is done. We show that the reconstruction of the eddy characteristics can be affected by the location of the ship track through the velocity field.&lt;/p&gt;&lt;p&gt;In comparison with the observed eddies the NATL60 eddies have smaller radii and higher azimuthal velocities and thus are more nonlinear. The inner core velocity structure for observations and NATL60 suggests solid body rotation for 2/3 of the radius. The maximum azimuthal velocity may deviate by up to 50% from solid body rotation.&lt;/p&gt;&lt;p&gt;The seasonality of the submesoscale regime can be seen in the data as the power spectrum is reduced from spring to summer in both the ship-based measurements and model.&lt;/p&gt;

Lagrangian Drifter to Identify Ocean Eddy Characteristics

Deterministic–stochastic empirical mode decomposition (EMD) is used to obtain low-frequency (non-diffusive; i.e., background velocity) and high-frequency (diffusive; i.e., eddies) components from a Lagrangian drifter‘s trajectory. Eddy characteristics are determined from the time series of eddy trajectories from individual Lagrangian drifters such as eddy radius, eddy velocity, eddy Rossby number, and the eddy–current kinetic energy ratio. A long-term dataset of the Sound Fixing and Ranging (RAFOS) float time series obtained near the California coast by the Naval Postgraduate School from 1992 to 2004 at depth between 150 and 600 m is used as an example to demonstrate the capability of the deterministic–stochastic EMD.

Lagrangian Drifter to Identify Ocean Eddy Characteristics

Deterministic-stochastic empirical mode decomposition (EMD) is used to obtain low-frequency (non-diffusive, i.e., background velocity) and high-frequency (diffusive, i.e., eddies) components from a Lagrangian drifter&lsquo;s trajectory. Eddy characteristics are determined from the time series of eddy trajectories from individual Lagrangian drifter such as the eddy radial scale, eddy velocity scale, eddy Rossby number, and eddy-background kinetic energy ratio. A long-term dataset of the SOund Fixing And Ranging float time series obtained near the California coast by the Naval Postgraduate School from 1992 to 2004 at depth between 150 and 600 m (http://www.oc.nps.edu/npsRAFOS/) is used as an example to demonstrate the capability of the deterministic-stochastic EMD.