Seafloor hydrothermal fluids play a critical role in regulating a wide range of geochemical and biological processes. Chemical and physical changes to venting fluids undoubtedly occur in response to geological events, at depth or near the seafloor, with corresponding effects on vent fluid chemistry, mineralization, and biological activity. However, the uncertain timing of such events makes it impractical to anticipate their occurrence. Thus, the temporal evolution of seafloor vent fluids and coexisting mineral deposits is most often inferred from observations made using conventional deep submergence assets that may be non-continuous for a specific vent field. The recent development of submarine volcanic observatories operated by Ocean Networks Canada (ONC) and the Ocean Observatory Initiative (OOI) in the U.S. that deliver power to instruments on the seafloor and communications via high-speed fiber optic cable permits real-time communication and monitoring of seafloor vents. Here, we describe the development and application of a remotely operated hydrothermal fluid sampler that enables repeat on-demand sampling of high-temperature vent fluids triggered through the internet. The sampler was deployed from September 2019 to June 2020 at the S&M vent area, in the southern portion of Main Endeavour Field (MEF), Juan de Fuca Ridge, on ONC's NEPTUNE ocean observatory. Nine vent fluid samples were acquired over a period of nine months. Analyses of these samples confirm the moderately high-temperature origin of the source fluid, as indicated, for example, by dissolved chloride depletion relative to seawater (vapor-rich) and moderately high silica concentrations. These and other dissolved species, such as methane, a known hallmark of MEF vent fluids, are in excellent agreement with the reported composition of S&M vent fluids over the past 15 years or more, suggesting overall chemical stability and rock-dominated alteration processes. On the other hand, short-term variability of dissolved Mg, sulfate and barium indicate entrainment of conductively heated and partially reacted seawater. Both before and after the incursion of the “secondary” seawater derived fluid, however, samples reveal the lowest Mg and sulfate concentrations yet reported for seafloor hydrothermal vent fluid samples. Accordingly, these data provide new insight on the solubility of these and other elements coexisting with minerals close to the Axial Magma Chamber (AMC), from which the primary source fluid is ultimately derived. A separate noteworthy event characterized by ∼20 °C decrease in temperature and associated 24–77% decrease in dissolved Fe concentration was observed toward the end of the deployment, with implications for hydrothermal transition metal fluxes and linked biogeochemical processes. On-demand remote acquisition of a continuous series of high-temperature vent fluid samples from a single vent permits the temporal evolution of heat and mass transfer processes to be studied with a heretofore unavailable perspective. Furthermore, deployment of this sampler within the NEPTUNE ocean observatory opens the door to future studies based on comparative analysis of contemporaneous datasets. 

http://manuscript.elsevier.com/S0967063722001224/pdf/S0967063722001224.pdf

Time series of hydrothermal vent fluid chemistry at Main Endeavour Field, juan de Fuca Ridge: Remote sampling using the NEPTUNE cabled observatory

The validation of ocean color satellite retrievals generally relies on analyzing match-ups between in situ measurements and satellite retrievals. These analyses focus on the quality of the satellite data, however, of the same importance is the quality of the in situ data. Here, we present the spatial variability of in situ above-water reflectance (Rrs(0+)) within the spatial resolution of different ocean color satellites—300, 900, 1500, and 3000 m spatial resolutions, mimicking Sentinel 3 OLCI and MODIS-Aqua satellites, and possible 3 × 3 and 5 × 5 windows. Radiometric data was acquired with autonomous radiometric sensors installed in the British Columbia Ferry Services Inc. vessel “Queen of Alberni” from May to September 2019, crossing the optically dynamic waters of the Strait of Georgia, Canada. The dataset followed optimal geometry of acquisition and processing, including corrections for skylight radiance signals, ship superstructure, the non-isotropic distribution of the water-leaving radiances, and quality control. A total of 33,073 spectra at full resolution, corresponding to 10 days, were considered for the analysis presented here. The results showed that, overall, the subpixel variability increased as the spatial resolution of the sensor or the window size increased, mainly in a linear fashion. Specifically, spatial variability of Rrs(0+) was the largest (∼18% and 68% for 900 and 3000 m pixel resolution, respectively) in Near Field Plume Interface waters, followed by in the Ocean Water Interface (∼28% and 35%, respectively), thus indicating spatial heterogeneity of interface waters. Further, we found that the estuarine waters showed higher subpixel Rrs(0+) variability (∼8% and 16% for 900 and 3000 m, respectively) compared with plume and oceanic waters. We showed that the high spatial variability in Rrs(0+) was primarily associated with the spatial dynamics of the optical water constituents, thus limiting the use of these datasets as Fiducial Reference Measurements and for validation of satellite-derived atmospherically corrected reflectance. We suggest that spatial variability of the in situ Rrs(0+) should also be considered in the selection criteria for good match-up data, especially for data acquired in coastal dynamic systems. As a result, it will advocate for the exclusion of interface or transition water pixel grids in order to avoid compromising the statistical result of satellite validation.

/pdf/spatial-variability-of-in-situ-above-water-reflectance-in-3q2mrbld.pdf

Spatial Variability of In Situ Above-Water Reflectance in Coastal Dynamic Waters: Implications for Satellite Match-Up Analysis

Ocean Networks Canada (ONC) runs a cabled ocean observatory that, among a wide variety of oceanographic instruments, has more than 30 hydrophones deployed. These hydrophones continuously send audio data back from the ocean floor, resulting in greater than 7 TB of data per hydrophone per year. To quickly visualize these acoustic data, power spectral density plots (spectrograms) are generated and stored in the ONC archive. While hydrophones generally perform well in the field, losses in sensitivity or ‘dropouts’ occasionally occur at isolated frequencies or across all bands for short periods of time. While such dropouts can be spotted on a spectrogram via visual inspection, the manual review of thousands of images is costly and impractical. With the current number of ONC’s hydrophones, it takes a data specialist over eight hours to inspect a day’s worth of spectrograms. We used a total of 2185 spectrogram images to train an autoencoder model. These images were split into three data sets: regular spectrograms without anomalies (1024 images that were separated in training and validation data sets), test data set (1000 unclassified images) and spectrograms containing anomalous signals (161 images). We used mean squared error (MSE) as a metric for the image reconstructions provided by the model and to establish the threshold of 0.006 to identify anomalous spectrograms. The trained model was capable of identifying most anomalous images from regular ones and correctly labelled approximately 93% of the images with anomalous features in the anomalous data sets. 

Anomaly detection in complex data: a practical application when outliers are few

Ocean Networks Canada (ONC) runs a cabled ocean observatory that, among a wide variety of oceanographic instruments, has more than 30 hydrophones deployed. These hydrophones continuously send audio data back from the ocean floor, resulting in greater than 7 TB of data per hydrophone per year. To quickly visualize these acoustic data, power spectral density plots (spectrograms) are generated and stored in the ONC archive. While hydrophones generally perform well in the field, losses in sensitivity or ‘dropouts’ occasionally occur at isolated frequencies or across all bands for short periods of time. While such dropouts can be spotted on a spectrogram via visual inspection, the manual review of thousands of images is costly and impractical. With the current number of ONC’s hydrophones, it takes a data specialist over eight hours to inspect a day’s worth of spectrograms. We used a total of 2185 spectrogram images to train an autoencoder model. These images were split into three data sets: regular spectrograms without anomalies (1024 images that were separated in training and validation data sets), test data set (1000 unclassified images) and spectrograms containing anomalous signals (161 images). We used mean squared error (MSE) as a metric for the image reconstructions provided by the model and to establish the threshold of 0.006 to identify anomalous spectrograms. The trained model was capable of identifying most anomalous images from regular ones and correctly labelled approximately 93% of the images with anomalous features in the anomalous data sets.

In accordance with the United Nations Decade of Ocean Science for Sustainable Development, Ocean Networks Canada (ONC) is increasingly standardizing data in compliance with globally accepted schemas and concepts such as Essential Ocean Variables (EOV)s as directed by the Canadian Integrated Ocean Observing System (CIOOS). One such EOV is biological abundance data and ONC aims to provide this to end users as the widely accepted format - Darwin Core Archives (DwC-A). This will be done in collaboration with the UN’s Ocean Biodiversity Information System (OBIS). Although biological datasets are regularly contributed to OBIS, there are few provided by ROV visual observations. ONC hopes to provide a novel method for deep-sea ROV visual observation data as inspiration for similar dataset contributions in the future. By using the latest recommendations from peer reviewed sources, ONC will provide both transect-based and opportunistic biological observations as DwC-A files in event core format. These tables will provide various environmental sensor information as taken from the Remote Operating Vehicle (ROV) at the point of observation for enriching of observation data. Datasets will be organized in a hierarchical manner, with an individual dataset consisting of a single dive event. Relational information will be available in both the body of the DwC-A and the dataset’s metadata for linking back to parent expeditions as a dataset collection. ONC has developed conventions for the annotations of biological observations within ONC’s in-house solution for video annotation and viewing, SeaTubeV3. This modular web-based solution for annotations is based on taxonomies and button-sets. Button-sets in particular are a great way to map out a list of commonly seen organisms for quick and efficient event annotations. Coupled with a World Register for Marine Species (WoRMS) taxonomy and potential attribute pre-set assignment (individual counts, open nomenclature codes), SeaTubeV3 can provide a template for rich observation data collection with little need to cross-walk to different taxonomies downstream. Annotations generated from the dives will go through a verification process in which subject matter experts can vet whether a certain taxonomic identification is accurate or not. The annotations that meet a certain threshold will then be passed to ONC task machines where an OBIS-compliant DwCA package will be generated. These packages will be automated to populate on the OBIS web page and other repositories like CIOOS for anyone to use. By using these novel techniques, ONC hopes to help pave the way for similar deep-water biological observation datasets to be accessible through open formats like OBIS.

Remote Operating Vehicle Observation Data Contributions to the Ocean Biodiversity Information System

The advent of large-scale cabled ocean observatories brought about the need to handle large amounts of ocean-based data, continuously recorded at a high sampling rate over many years and made accessible in near-real time to the ocean science community and the public. Ocean Networks Canada (ONC) commenced installing and operating two regional cabled observatories on Canada’s Pacific Coast, VENUS inshore and NEPTUNE offshore in the 2000s, and later expanded to include observatories in the Atlantic and Arctic in the 2010s. The first data streams from the cabled instrument nodes started flowing in February 2006. This paper describes Oceans 2.0 and Oceans 3.0, the comprehensive Data Management and Archival System that ONC developed to capture all data and associated metadata into an ever-expanding dynamic database. Oceans 2.0 was the name for this software system from 2006–2021; in 2022, ONC revised this name to Oceans 3.0, reflecting the system’s many new and planned capabilities aligning with Web 3.0 concepts. Oceans 3.0 comprises both tools to manage the data acquisition and archival of all instrumental assets managed by ONC as well as end-user tools to discover, process, visualize and download the data. Oceans 3.0 rests upon ten foundational pillars: (1) A robust and stable system architecture to serve as the backbone within a context of constant technological progress and evolving needs of the operators and end users; (2) a data acquisition and archival framework for infrastructure management and data recording, including instrument drivers and parsers to capture all data and observatory actions, alongside task management options and support for data versioning; (3) a metadata system tracking all the details necessary to archive Findable, Accessible, Interoperable and Reproducible (FAIR) data from all scientific and non-scientific sensors; (4) a data Quality Assurance and Quality Control lifecycle with a consistent workflow and automated testing to detect instrument, data and network issues; (5) a data product pipeline ensuring the data are served in a wide variety of standard formats; (6) data discovery and access tools, both generalized and use-specific, allowing users to find and access data of interest; (7) an Application Programming Interface that enables scripted data discovery and access; (8) capabilities for customized and interactive data handling such as annotating videos or ingesting individual campaign-based data sets; (9) a system for generating persistent data identifiers and data citations, which supports interoperability with external data repositories; (10) capabilities to automatically detect and react to emergent events such as earthquakes. With a growing database and advancing technological capabilities, Oceans 3.0 is evolving toward a future in which the old paradigm of downloading packaged data files transitions to the new paradigm of cloud-based environments for data discovery, processing, analysis, and exchange.

/pdf/the-oceans-2-0-3-0-data-management-and-archival-system-2l3djju8.pdf

Yan Chen

Papers

The Oceans 2.0/3.0 Data Management and Archival System