Abstract
The deep ocean is vast and challenging to observe; however, it is key to knowledge of the sea and its impact on global climate. Fixed sea observing points (such as the EMSO observing nodes) provide a limited view and are complemented by expensive oceanographic campaigns with systems demanding high logistical requirements such as deep-sea ROVs.  These costs not only limit our capability for key ocean data collection in the deep but also introduce their own environmental costs.Emerging challenges in knowledge and pressure on the exploration of the deep ocean demand new technological solutions for monitoring and safeguarding the marine ecosystem.Innovative robotic technologies such as the TURTLE robotic deep-sea landers can combine long-term permanence at the seabed with mobility and dynamic reconfigurability in spatial and temporal deep-sea observation.Robotic systems of a heterogeneous nature (from conventional gliders, AUVs, or robotic landers) can be combined with standard and new sensing systems, such as bottom-deployed sensor nodes, moored systems, and cabled points when feasible.These systems can provide underwater localization services for the different assets, energy supply and high bandwidth data transfer with robotic docking stations for other mobile elements. An example of the synergy obtained with these new systems is the possibility of using robotic landers as carriers of EGIM (EMSO Generic Instrument Module) sensor payloads, providing power and data storage and flexibility in the deployment and recovery process.This approach, partly taken in the EU-funded Trident project to develop technical solutions for cost-effective and efficient observation of environmental impacts on deep seabed environments, allows for a substantial reduction in the operational and logistic requirements for deep-sea observation, greatly reducing the need for costly oceanographic campaigns or the use of expensive (economic and logistical) deep sea ROV systems.In this work, we present some of the new developments and discuss the transition from existing technological solutions to new ones integrating these recent developments.

Abstract
Abandoned, Lost, or otherwise Discarded Fishing Gear (ALDFG) has a great impact on marine ecosystems. This is not only due to the direct contribution to marine litter production with particular emphasis on plastics but also to the effects of ghost fishing.The Nettag+ project aims to reduce these impacts by acting on three main lines of action: prevention, avoidance, and mitigation. In the first line, direct action and collaboration with fishers and nature protection organizations around Europe aim to establish the fishermen community as guardians of the ocean. These actions with active fishers' collaboration range from training and dissemination activities related to marine litter and ocean protection to direct measures in day-to-day work to minimize and recover litter from the sea.In the prevention line, an acoustic tag designed explicitly for the location of ALDFG was developed in collaboration with research institutions and fishing gear manufacturers. It can be integrated into the fishing equipment for future tracking and recovery. This tool can reduce lost fishing gear retrieval costs and is complemented with robotic solutions to support retrieving operations.To mitigate the effects of existing untagged ALDFG, multisensorial  detection algorithms are being developed to detect and map ALDFG on the sea and to take advantage of autonomous and robotic systems to perform this task.

Abstract
This study addresses the variability of gamma radiation measurements over the Atlantic Ocean. The analysis of back trajectories shows that the path of the air masses is the main factor determining gamma radiation levels over the ocean, rather than the distance to the coast. Different gamma values were recorded at different times in the same location as a result of the distinct origin of the corresponding air masses. Higher counts observed in the northeast Atlantic in winter compared with the spring values result from air masses coming from Europe and the African continent. In general, gamma radiation values over the ocean increase with increasing continental influence on the air mass above. A predictive classifica-tion model is developed showing that marine gamma observations can be used to classify marine boundary layer air masses according to the degree of continental influence.

Abstract
This paper presents the development and implementation of learning-based detection and tracking methods using multibeam data to detect marine litter in the water column. The presented work encompasses (i) the creation of acoustic videos and the application of multiple post-processing techniques; (ii) the training of multiple You Only Look Once (YOLO) detection models, specifically YOLOv8, across different variants, acoustic frequencies, and input types (both raw and post-processed); (iii) and the development of a marine litter tracking system based on DeepSORT. The results include a multibeam multi-frequency data study demonstrating the potential of acoustic image sensing for detecting and tracking marine litter materials in the water column.

Abstract
The impact of wildfires is increasing worldwide. The root causes of these effects are manifold, encompassing among others climate change and the accumulation of fuels and increasing settlements in wildland-urban interfaces (WUI). Reports and initiatives to better understand and govern these developments have been launched and call for more integrated approaches to wildfire risk management, including the use of targets or Key Performance Indicators (KPIs). However, despite some examples such as Portugal, wildfire risk management targets are still mainly lacking in Europe. This is surprising since they find wider application in the U.S. and are also more widely applied for flooding in Europe. This perspective hence takes a closer look at the use of targets in reducing disaster risk for different hazards worldwide and reflects about the opportunities and challenges for wildfire risk reduction targets for Europe. It concludes with some suggestions for the application of wildfire risk reduction targets for Europe.

Abstract
Crowdsourced data streams are popular and extremely valuable in several domains, namely in tourism. Tourism crowdsourcing platforms rely on past tourist and business inputs to provide tailored recommendations to current users in real time. The continuous, open, dynamic and non-curated nature of the crowd-originated data demands specific stream mining techniques to support online profiling, recommendation, change detection and adaptation, explanation and evaluation. The sought techniques must, not only, continuously improve and adapt profiles and models; but must also be transparent, overcome biases, prioritize preferences, master huge data volumes and all in real time. This article surveys the state-of-art of adaptive and explainable stream recommendation, extends the taxonomy of explainable recommendations from the offline to the stream-based scenario, and identifies future research opportunities.