Search

Mapping benthic habitats with bathymetric, acoustic, and spectral data requires georeferenced ground-truth information about habitat types and characteristics. New technologies like autonomous underwater vehicles (AUVs) collect tens of thousands of images per mission making image-based ground truthing particularly attractive. Two types of machine learning (ML) models, random forest (RF) and deep neural network (DNN), were tested to determine whether ML models could serve as an accurate substitute for manual classification of AUV images for substrate type interpretation. RF models were trained to predict substrate class as a function of texture, edge, and intensity metrics (i.e., features) calculated for each image. Models were tested using a manually classified image dataset with 9-, 6-, and 2-class schemes based on the Coastal and Marine Ecological Classification Standard (CMECS). Results suggest that both RF and DNN models achieve comparable accuracies, with the 9-class models being least accurate (~73–78%) and the 2-class models being the most accurate (~95–96%). However, the DNN models were more efficient to train and apply because they did not require feature estimation before training or classification. Integrating ML models into benthic habitat mapping process can improve our ability to efficiently and accurately ground-truth large areas of benthic habitat using AUV or similar images. [ABSTRACT FROM AUTHOR]

Light Detection and Ranging (LiDAR), a comparatively new technology in the field of underwater surveying, has principally been used for taking precise measurement of undersea structures in the oil and gas industry. Typically, the LiDAR is deployed on a remotely operated vehicle (ROV), which will "land" on the seafloor in order to generate a 3D point cloud of its environment from a stationary position. To explore the potential of subsea LiDAR on a moving platform in an environmental context, we deployed an underwater LiDAR system simultaneously with a multibeam echosounder (MBES), surveying Kingston Reef off the coast of Rottnest Island, Western Australia. This paper compares and summarises the relative accuracy and characteristics of underwater LiDAR and multibeam sonar and investigates synergies between sonar and LiDAR technology for the purpose of benthic habitat mapping and underwater simultaneous localisation and mapping (SLAM) for Autonomous Underwater Vehicles (AUVs). We found that LiDAR reflectivity and multibeam backscatter are complementary technologies for habitat mapping, which can combine to discriminate between habitats that could not be mapped with either one alone. For robot navigation, SLAM can be effectively applied with either technology, however, when a Global Navigation Satellite System (GNSS) is available, SLAM does not significantly improve the self-consistency of multibeam data, but it does for LiDAR. [ABSTRACT FROM AUTHOR]

In this work, multibeam echosounder (MBES) and dual frequency sidescan sonar (SSS) data are combined to map the shallow (5–100 m) benthic habitats of the National Marine Park of Zakynthos (NMPZ), Greece, a Marine Protected Area (MPA). NMPZ hosts extensive prairies of the protected Mediterranean phanerogams Posidonia oceanica and Cymodocea nodosa, as well as reefs and sandbanks. Seafloor characterization is achieved using the multi-frequency acoustic backscatter of: (a) the two simultaneous frequencies of the SSS (100 and 400 kHz) and (b) the MBES (180 kHz), as well as the MBES bathymetry. Overall, these high-resolution datasets cover an area of 84 km2 with ground coverage varying from 50% to 100%. Image texture, terrain and backscatter angular response analyses are applied to the above, to extract a range of statistical features. Those have different spatial densities and so they are combined through an object-based approach based on the full-coverage 100-kHz SSS mosaic. Supervised classification is applied to data models composed of operationally meaningful combinations between the above features, reflecting single-sonar or multi-sonar mapping scenarios. Classification results are validated against a detailed expert interpretation habitat map making use of extensive ground-truth data. The relative gain of one system or one feature extraction method or another are thoroughly examined. The frequency-dependent separation of benthic habitats showcases the potentials of multi-frequency backscatter and bathymetry from different sonars, improving evidence-based interpretations of shallow benthic habitats. [ABSTRACT FROM AUTHOR]

This study was an evaluation of the spectral signature generalization properties of coral across four remote Pacific Ocean reefs. The sites under consideration have not been the subject of previous studies for coral classification using remote sensing data. Previous research regarding using remote sensing to identify reefs has been limited to in-situ assessment, with some researchers also performing temporal analysis of a selected area of interest. This study expanded the previous in-situ analyses by evaluating the ability of a basic predictor, Linear Discriminant Analysis (LDA), trained on Depth Invariant Indices calculated from the spectral signature of coral in one location to generalize to other locations, both within the same scene and in other scenes. Three Landsat 8 scenes were selected and masked for null, land, and obstructed pixels, and corrections for sun glint and atmospheric interference were applied. Depth Invariant Indices (DII) were then calculated according to the method of Lyzenga and an LDA classifier trained on ground truth data from a single scene. The resulting LDA classifier was then applied to other locations and the coral classification accuracy evaluated. When applied to ground truth data from the Palmyra Atoll location in scene path/row 065/056, the initial model achieved an accuracy of 80.3%. However, when applied to ground truth observations from another location within the scene, namely, Kingman Reef, it achieved an accuracy of 78.6%. The model was then applied to two additional scenes (Howland Island and Baker Island Atoll), which yielded an accuracy of 69.2% and 71.4%, respectively. Finally, the algorithm was retrained using data gathered from all four sites, which produced an overall accuracy of 74.1%. [ABSTRACT FROM AUTHOR]