Research Progress on Ocean Observations Technology and Information Systems

The oceans play a crucial role in the global ecosystem; they shape trends in the climate, weather, water management, and health (including biogeochemical cycles). Although innovative technologies have been developed for the marine environment to better understand the ocean’s processes, several constraints remain in ocean observation, hindering the replacement of laboratory and routine monitoring methods. The development of smart in situ marine sensors—capable of being integrated into existing fixed units (such as landers and mooring buoys) as well as in mobile units (such as AUVs, ROVs, ships of opportunity, marine drones, Argo floats, gliders)—is ongoing in the form of various European and international projects. In recent years, a lot of progress has been made in ocean observation technologies and information systems; for example, cost-effective and miniaturised sensing devices with very low power consumption have been developed, which can be easily integrated in agnostic platforms. Additionally, a lot of effort has been made to develop acoustic communication methods and modules to transmit the data in the ocean as well as cellular systems for transmitting the data of marine sensors in near-real-time mode. In order to improve our understanding of global ocean processes (such as weather monitoring and forecasting, climate variability, sea level rise, natural hazards, ocean acidification, ocean health, pollution and ecosystem functioning, energy, economic development and coastal management, public safety, security, training and education), new research is ongoing with the hopes of improving and optimizing infrastructures and models at international level. State-of-the-art in situ marine sensors—which can be easily integrated into ocean platforms and can be combined with innovative communication and information systems for real-time data transmission—are emerging as new features of both forecasting methods and smart emergency systems, in place to protect humans. This Special Issue promotes new insights into the integration of sensors into platforms and cabled observatories for the continuous monitoring of requested parameters. A brief overview of all the contributions is provided in this Editorial.

In this Special Issue, a multidisciplinary study is performed in the frame of an IAEA campaign for the evaluation of sediment dynamics using in situ radiometric methods combined with theoretical models; this research is conducted in terms of numerical simulations by an open-source model considering land–sea interactions and meteorological conditions [1]. Sediment dynamics assessments were performed based on the mapped activity concentrations of gamma-ray emitters of ²³⁸U and ²³²Th progenies and ⁴⁰K along the shoreline of the beach. The validated results showed that the spatial distribution of radionuclides reflects the selective transport processes of sediments in the system. Special attention is also given to using unmanned aerial and autonomous surface vehicles (UAVs and ASVs) for the mapping of coastal and marine environments. The sea-bottom composition is resolved with very high fidelity in shallow waters using UAV photogrammetry. Tests showed that, at greater depths, acoustic methods exhibit far better propagation properties than optical methods. The main output of this technology is the application of a sea-bottom classification methodology for mapping the protected habitat of Mediterranean Sea grass Posidonia oceanica through a machine learning scheme [2]. Due to the accuracy issues, the proposed hybrid mapping approach is promising to mitigate the depth-induced bias in UAV-only models.

Underwater in situ gamma-ray spectrometry was also used to monitor natural radionuclides at the Nice Slope for over a month, exploiting the radon daughters for tracing groundwater movement from a coastal aquifer to a nearshore continental shelf [2]. Groundwater movement leads to the production of submarine groundwater discharge and sediment weakening combined with slope failure. The time series of hydrological parameters identified tidal fluctuations in freshwater input, highlighting the crucial role of the groundwater source in better understanding the stability of failure-prone slopes. The radionuclide monitoring time series data provide a valuable contribution in efforts to charge permeable sub-bottom areas [3]. In situ systems were also deployed to continuously monitor radioactivity levels in the North Sea using an existing cabled subsea network (COSYNA network) in order to identify and to assess potential anthropogenic and/or natural hazards [4]. The system is operated via an online mode controlled by the operational centre and provides continuous gamma-ray spectra and activity concentrations of key radionuclides that were enriched in seawater during the monitoring period. The effectiveness of the method is validated together with the operability of the underwater sensor, and quantification methods were optimised. The data analysis proved that the system is stable in terms of voltage stability optimising the data analysis via summing up efficiently in time to produce statistical results. Additionally, a review manuscript related to cabled ocean observatories (COOs) is enabled to provide real-time in situ ocean observations, facilitating oceanic understanding and exploration by compiling and discussing typical COOs worldwide in terms of system configurations and state-of-the-art technology, including network structures, power supply modes, and communication capabilities [5]. The main characteristics of line, ring, star, and grid networks and their applicability in COOs are elucidated, and the advantages and disadvantages of various power supply modes, as well as the opportunities brought by the development of communication technologies, are described.

In order to predict time series for marine scientific research, the Hierarchical Temporal Memory (HTM) model proved to be more suitable than traditional recurrent-neural-network-based models due to its online learning and prediction capabilities [6]. This study proposes a new, improved HTM model, incorporating Gated Recurrent Units neurons into the temporal memory algorithm. The capacities and advantages of the proposed model are validated and evaluated on time series data collected from the Xiaoqushan Seafloor Observatory in the East China Sea. It is proven that the mechanism of online learning has certain advantages in predicting ocean observation data. Moreover, a neural network method was developed to predict the position of Argo buoys, improving target tracking and emergency support capabilities [7]. Based on a deep learning framework using a Simple Recurrent Unit, a new Time–Space Feature Fusion Method based on an Attention Mechanism (TSFFAM) model to accurately predict the target trajectory, avoiding the disadvantages of traditional Long Short-Term Memory models (which are time-consuming and difficult to train). The TSFFAM model is able to better capture multiscale ocean factors, leading to more accurate and efficient buoy trajectory predictions, and it aims to shed light on the mechanism of the joint multi-element and multiscale effects of laminar and surface currents. Experimental verification is conducted in the Pacific Ocean using buoy trajectory data, proving practical value for the field of maritime activities and needs, including rescue operations and efficient target tracking.

Another topic of the Special Issue is related to simulating ocean dynamics parameters by integrating ocean numerical models into the sea clutter spectrum estimation, exploiting existing filter parameters [8]. Sea clutter introduces a typical background signal due to the echo signals received by radar; thus, target detection and identification become complicated. The Weather Research and Forecasting model is employed to simulate the ocean dynamic parameters within the radar detection area, utilising hydrological data to calibrate the parameterisation scheme. Based on the simulated ocean dynamics parameters, empirical formulas are used to calculate the sea clutter spectrum. The filter coefficients are updated in real-time mode using the sea clutter spectral parameters, enabling the precise suppression of sea clutter and providing a drastically improved factor.

The first highlight of this Special Issue is the optimised method to perform in situ measurements directly into the marine environment, replacing lab-based methods for routine monitoring (which are time-consuming and costly). Such activities concern the study of sediment dynamics near the shoreline by exploiting natural radionuclides as tracers, installing a georeferenced detection system in mobile platforms to identify the sediment transport. The mapping processes of marine habitat are also demonstrated via a transfer-learning framework using data from mobile platforms. These mapping processes enable seagrass habitats (Posidonia oceanica) beyond the depth limits of optical imaging. The second highlight is related to the identification and characterisation of submarine groundwater discharge processes using in situ underwater detection systems. The dataset improves our understanding of the hydrogeological conditions that influence coastal landslide hazards by linking submarine groundwater flow to submarine slope instability processes. Furthermore, a similar underwater detection system is integrated into an existing cabled coastal observing network to monitor and inspect potential threats through continuous real-time monitoring of marine radioactivity, especially in rainfall periods. Moreover, a comprehensive state-of-the-art review of cabled ocean observatory technology is presented, emerging systematic classification of network architectures, power and communication systems and sensor integration strategies. Identification of future technological challenges and development pathways for large-scale ocean observing infrastructures (knowledge synthesis, strategic technological roadmap) are also discussed. The third highlight is the use of neural networks for several ocean applications. The development of a sea clutter suppression algorithm that explicitly incorporates ocean dynamic parameters, simulated with the WRF model, demonstrates that numerical weather/ocean models can improve radar signal processing by introducing a physics-informed clutter suppression approach. Hierarchical Temporal Memory (HTM) algorithms, used for real-time oceanographic data prediction, can improve online learning capability for streaming ocean time series data and demonstrate higher predictive performance for ocean monitoring data compared with traditional time series models (AI algorithm enhancement, online prediction for ocean monitoring systems, physics-informed signal filtering). In the development of a trajectory prediction model for Argo floats using multiscale oceanographic drivers, the authors introduced a spatiotemporal Attention Mechanism; this allows the modelling if ocean current influences and the integration of physical ocean parameters and deep learning for improved drift prediction.

Long-term ocean observation systems increasingly rely on technological breakthrough solutions; these are widely applied as they are cost-effective tools which can be used in mapping the marine environment, thereby supporting the protection of humans from threats that are a result of climate change, pollution, and biodiversity. For instance, radiotracers are powerful tools which have been used in the study of the Aegean Sea [9] (water mass circulation, particle transport, and biogeochemical processes); researchers have used them to study the behaviour of anthropogenic radionuclides using systematic sampling and analysis. At present, the activity concentration of Cesium in the deep basins is relatively low; however, there is a need to acquire data from the deep sea, and researchers are employing observation tools [10] to correlate observed measurements with oceanographic, geophysical and biological parameters. The goal of these efforts is to contribute to a better understanding of how anthropogenic tracers propagate in complex marginal seas and how they can be used to trace long-term processes and system interaction. Recent advances in artificial intelligence (AI) and machine learning are also opening new possibilities for ocean processes. For instance, researchers have recently performed radiometric analyses of the automated processing of gamma-ray spectra acquired by low-resolution in situ detectors [11]; these analyses were made possible by an AI-assisted spectrum approach, which can be successfully applied in low-level spectra analyses and can be used in support of near-real-time data processing within marine monitoring systems. The integration of such data-driven approaches with ocean observation infrastructures can further strengthen the capability of tracer measurements to provide reliable and continuous information on marine environmental conditions. The derived data can support environmental monitoring programs, contribute to the validation of ocean models, and enhance the capability of marine observation networks to detect and interpret long-term environmental changes.