Description: In this study ship- and AUV-based multibeam data from the German Mn-nodule license area in the Clarion-Clipperton Zone (CCZ; eastern Pacific) are linked to ground truth data from optical imaging. Photographs obtained by an AUV enable semi-quantitative assessments of nodule coverage at a spatial resolution in the range of meters. Together with high resolution AUV bathymetry this revealed a correlation of small-scale terrain variations (〈 5 m horizontally, 〈 1 m vertically) with nodule abundance. In the presented data set, increased nodule coverage could be correlated with slopes 〉 1.8° and concave terrain. On a more regional scale, factors such as the geological setting (existence of horst and graben structures, sediment thickness, outcropping basement) and influence of bottom currents seem to play an essential role for the spatial variation of nodule abundance and the related hard substrate habitat. AUV imagery was also successfully employed to map the distribution of re-settled sediment following a disturbance and sediment cloud generation during a sampling deployment of an Epibenthic Sledge. Data from before and after the "disturbance" allows a direct assessment of the impact. Automated image processing analyzed the nodule coverage at the seafloor, revealing nodule blanketing by resettling of suspended sediment within 16 hours after the disturbance. The visually detectable impact was spatially limited to a maximum of 100m distance from the disturbance track, downstream of the bottom water current. A correlation with high resolution AUV bathymetry reveals that the blanketing pattern varies in extent by tens of meters, strictly following the bathymetry, even in areas of only slightly undulating seafloor (〈 1 m vertical change). These results highlight the importance of detailed terrain knowledge when engaging in resource assessment studies for nodule abundance estimates and defining minable areas. At the same time, it shows the importance of high resolution mapping for detailed benthic habitat studies that show a heterogeneity at scales of 10 m to 100 m. Terrain knowledge is also needed to determine the scale of the impact by seafloor sediment blanketing during mining-operations.

Description: In this study, high-resolution bathymetric multibeam and optical image data, both obtained within the Belgian manganese (Mn) nodule mining license area by the autonomous underwater vehicle (AUV) Abyss, were combined in order to create a predictive random forests (RF) machine learning model. AUV bathymetry reveals small-scale terrain variations, allowing slope estimations and calculation of bathymetric derivatives such as slope, curvature, and ruggedness. Optical AUV imagery provides quantitative information regarding the distribution (number and median size) of Mn nodules. Within the area considered in this study, Mn nodules show a heterogeneous and spatially clustered pattern, and their number per square meter is negatively correlated with their median size. A prediction of the number of Mn nodules was achieved by combining information derived from the acoustic and optical data using a RF model. This model was tuned by examining the influence of the training set size, the number of growing trees (ntree), and the number of predictor variables to be randomly selected at each node (mtry) on the RF prediction accuracy. The use of larger training data sets with higher ntree and mtry values increases the accuracy. To estimate the Mn-nodule abundance, these predictions were linked to ground-truth data acquired by box coring. Linking optical and hydroacoustic data revealed a nonlinear relationship between the Mn-nodule distribution and topographic characteristics. This highlights the importance of a detailed terrain reconstruction for a predictive modeling of Mn-nodule abundance. In addition, this study underlines the necessity of a sufficient spatial distribution of the optical data to provide reliable modeling input for the RF.

Description: Marine imaging is transforming into a sensor technology applied for high throughput sampling. In the context of habitat mapping, imaging establishes thereby an important bridge technology regarding the spatial resolution and information content between physical sampling gear (e.g., box corer, multi corer) on the one end and hydro-acoustic sensors on the other end of the spectrum of sampling methods. In contrast to other scientific imaging domains, such as digital pathology, there are no protocols and reports available that guide users (often referred to as observers) in the non-trivial process of assigning semantic categories to whole images, regions, or objects of interest (OOI), which is referred to as annotation. These protocols are crucial to facilitate image analysis as a robust scientific method. In this article we will review the past observations in manual Marine Image Annotations (MIA) and provide (a) a guideline for collecting manual annotations, (b) definitions for annotation quality, and (c) a statistical framework to analyze the performance of human expert annotations and to compare those to computational approaches.

Description: Autonomous underwater vehicles (AUVs) offer unique possibilities for exploring the deep seafloor in high resolution over large areas. We highlight the results from AUV-based multibeam echosounder (MBES) bathymetry / backscatter and digital optical imagery from the DISCOL area acquired during research cruise SO242 in 2015. AUV bathymetry reveals a morphologically complex seafloor with rough terrain in seamount areas and low-relief variations in sedimentary abyssal plains which are covered in Mn-nodules. Backscatter provides valuable information about the seafloor type and particularly about the influence of Mn-nodules on the response of the transmitted acoustic signal. Primarily, Mn-nodule abundances were determined by means of automated nodule detection on AUV seafloor imagery and nodule metrics such as nodules m−2 were calculated automatically for each image allowing further spatial analysis within GIS in conjunction with the acoustic data. AUV-based backscatter was clustered using both raw data and corrected backscatter mosaics. In total, two unsupervised methods and one machine learning approach were utilized for backscatter classification and Mn-nodule predictive mapping. Bayesian statistical analysis was applied to the raw backscatter values resulting in six acoustic classes. In addition, Iterative Self-Organizing Data Analysis (ISODATA) clustering was applied to the backscatter mosaic and its statistics (mean, mode, 10th, and 90th quantiles) suggesting an optimum of six clusters as well. Part of the nodule metrics data was combined with bathymetry, bathymetric derivatives and backscatter statistics for predictive mapping of the Mn-nodule density using a Random Forest classifier. Results indicate that acoustic classes, predictions from Random Forest model and image-based nodule metrics show very similar spatial distribution patterns with acoustic classes hence capturing most of the fine-scale Mn-nodule variability. Backscatter classes reflect areas with homogeneous nodule density. A strong influence of mean backscatter, fine scale BPI and concavity of the bathymetry on nodule prediction is seen. These observations imply that nodule densities are generally affected by local micro-bathymetry in a way that is not yet fully understood. However, it can be concluded that the spatial occurrence of Mn-covered areas can be sufficiently analysed by means of acoustic classification and multivariate predictive mapping allowing to determine the spatial nodule density in a much more robust way than previously possible.

Description: Combining state-of-the art digital imaging technology with different kinds of marine exploration techniques such as modern AUV (autonomous underwater vehicle), ROV (remote operating vehicle) or other monitoring platforms enables marine imaging on new spatial and/or temporal scales. A comprehensive interpretation of such image collections requires the detection, classification and quantification of objects of interest in the images usually performed by domain experts. However, the data volume and the rich content of the images makes the support by software tools inevitable. We define some requirements for marine image annotation and present our new online tool Biigle 2.0. It is developed with a special focus on annotating benthic fauna in marine image collections with tools customized to increase efficiency and effectiveness in the manual annotation process. The software architecture of the system is described and the special features of Biigle 2.0 are illustrated with different use-cases and future developments are discussed.

Description: Marine researchers continue to create large quantities of benthic images e.g., using AUVs (Autonomous Underwater Vehicles). In order to quantify the size of sessile objects in the images, a pixel-to-centimeter ratio is required for each image, often indirectly provided through a geometric laser point (LP) pattern, projected onto the seafloor. Manual annotation of these LPs in all images is too time-consuming and thus infeasible for nowadays data volumes. Because of the technical evolution of camera rigs, the LP's geometrical layout and color features vary for different expeditions and projects. This makes the application of one algorithm, tuned to a strictly defined LP pattern, also ineffective. Here we present the web-tool DELPHI, that efficiently learns the LP layout for one image transect/collection from just a small number of hand labeled LPs and applies this layout model to the rest of the data. The efficiency in adapting to new data allows to compute the LPs and the pixel-to-centimeter ratio fully automatic and with high accuracy. DELPHI is applied to two real-world examples and shows clear improvements regarding reduction of tuning effort for new LP patterns as well as increasing detection performance.

Description: Cold-water coral (CWC) reefs are heterogeneous ecosystems comprising numerous microhabitats. A typical European CWC reef provides various biogenic microhabitats (within, on and surrounding colonies of coral species such as Lophelia pertusa, Paragorgia arborea and Primnoa resedaeformis, or formed by their remains after death). These microhabitats may be surrounded and intermixed with non-biogenic microhabitats (soft sediment, hard ground, gravel/pebbles, steep walls). To date, studies of distribution of sessile fauna across CWC reefs have been more numerous than those investigating mobile fauna distribution. In this study we quantified shrimp densities associated with key CWC microhabitat categories at the Røst Reef, Norway, by analysing image data collected by towed video sled in June 2007. We also investigated shrimp distribution patterns on the local scale (〈40 cm) and how these may vary with microhabitat. Shrimp abundances at the Røst Reef were on average an order of magnitude greater in biogenic reef microhabitats than in non-biogenic microhabitats. Greatest shrimp densities were observed in association with live Paragorgia arborea microhabitat (43 shrimp m−2, SD = 35.5), live Primnoa resedaeformis microhabitat (41.6 shrimp m−2, SD = 26.1) and live Lophelia pertusa microhabitat (24.4 shrimp m−2, SD = 18.6). In non-biogenic microhabitat, shrimp densities were 〈2 shrimp m−2. CWC reef microhabitats appear to support greater shrimp densities than the surrounding non-biogenic microhabitats at the Røst Reef, at least at the time of survey.

Description: High-resolution optical and hydroacoustic seafloor data acquired in 2015 enabled the reconstruction of disturbance tracks of a past Benthic Impact Experiment that was conducted in 1989 in the Peru Basin in the course of former German environmental impact studies associated with manganese nodule mining. Based on this information, the disturbance level of the experiment regarding the plough impact and distribution and re-deposition of sediment from the evolving sediment plume was assessed qualitatively. Through this, the evolution over the 26 years of a number of the total 78 disturbance tracks could be analyzed which highlights the considerable difference between natural sedimentation in the deep-sea and sedimentation of a resettled sediment plume. Such plumes are seen as one of the most concerning impact associated with potential Mn-nodule mining. Problems in data processing became eminent while dealing with old data from the late 80s, at a time when GPS was just invented and underwater navigation was in an infant stage. However, even today the uncertainties of underwater navigation and the use of a variety of acoustical and optical sensors at different resolutions require detailed post-processing in terms of absolute geographic positioning to improve the overall accuracy of the data. In this study, a ship-based bathymetric map of the survey area was used as absolute geographic reference and a workflow was applied successfully resulting in the most accurate geo-referenced dataset of the DISCOL Experimental Area to date. The new field data were acquired with sensors attached to GEOMARs AUV Abyss and the 0.5 × 1° EM122 multibeam system of RV SONNE during cruise SO242 -1 while the old data first needed to be found and compiled before they could be digitized and properly georeferenced for the presented joined analyses.

Description: With the mining of polymetallic nodules from the deep-sea seafloor once more evoking commercial interest, decisions must be taken on how to most efficiently regulate and monitor physical and community disturbance in these remote ecosystems. Image-based approaches allow non-destructive assessment of the abundance of larger fauna to be derived from survey data, with repeat surveys of areas possible to allow time series data collection. At the time of writing, key underwater imaging platforms commonly used to map seafloor fauna abundances are autonomous underwater vehicles (AUVs), remotely operated vehicles (ROVs) and towed camera “ocean floor observation systems” (OFOSs). These systems are highly customisable, with cameras, illumination sources and deployment protocols changing rapidly, even during a survey cruise. In this study, eight image datasets were collected from a discrete area of polymetallic-nodule-rich seafloor by an AUV and several OFOSs deployed at various altitudes above the seafloor. A fauna identification catalogue was used by five annotators to estimate the abundances of 20 fauna categories from the different datasets. Results show that, for many categories of megafauna, differences in image resolution greatly influenced the estimations of fauna abundance determined by the annotators. This is an important finding for the development of future monitoring legislation for these areas. When and if commercial exploitation of these marine resources commences, robust and verifiable standards which incorporate developing technological advances in camera-based monitoring surveys should be key to developing appropriate management regulations for these regions.

Description: Marine image analysis faces a multitude of challenges: data set size easily reaches Terabyte-scale; the underwater visual signal is often impaired to the point where information content becomes negligible; human interpreters are scarce and can only focus on subsets of the available data due to the annotation effort involved etc. Solutions to speed-up the analysis process have been presented in the literature in the form of semi-automation with artificial intelligence methods like machine learning. But the algorithms employed to automate the analysis commonly rely on large-scale compute infrastructure. So far, such an infrastructure has only been available on-shore. Here, a mobile compute cluster is presented to bring big image data analysis capabilities out to sea. The Sea-going High-Performance Compute Cluster (SHiPCC) units are mobile, robustly designed to operate with electrically impure ship-based power supplies and based on off-the-shelf computer hardware. Each unit comprises of up to eight compute nodes with graphics processing units for efficient image analysis and an internal storage to manage the big image data sets. The first SHiPCC unit has been successfully deployed at sea. It allowed us to extract semantic and quantitative information from a Terabyte-sized image data set within 1.5 h (a relative speedup of 97 compared to a single four-core CPU computer). Enabling such compute capability out at sea allows to include image-derived information into the cruise research plan, for example by determining promising sampling locations. The SHiPCC units are envisioned to generally improve the relevance and importance of optical imagery for marine science.