| Algemene beschrijving herkomst |
Data lineage information was taken from "https://s3-us-west-2.amazonaws.com/bluehalometadata/Barbuda-Metadata-Habitat-Mapping-v1.htm:" IMAGE DATA: IKONOS-2 satellite imagery collected by GeoEye (now DigitalGlobe) in 2012 is the basis for the layer. The images have a per pixel spatial resolution of 4 m by 4 m, thus covering a 16 sq. m area, and four spectral bands covering the wavelength range of 400 - 1050 nm. Three of the four bands are in the visible (VIS) spectrum (400 - 750 nm), and the remaining band is in the near infra-red (NIR) spectrum (750 - 1050 nm). The IKONOS image was evaluated for quality prior to purchase. Scenes with excessive sea-surface-glint, cloud cover, or other factors that obscured seafloor features, were avoided. Imagery was delivered as a georectified product. The images were converted from 16-bit digital numbers (DN) to remote sensing reflectance (%) just above the water's surface. Only light between 400 nm and 700 nm penetrates the water column sufficiently to provide usable information on benthic composition, thus only the three spectral bands within this region are used for benthic habitat mapping. Land and cloud were masked out of imagery and a correction for sea-surface glint was applied prior to habitat classification.CLASSIFICATION: The product was created through object-oriented mapping using Definiens eCognition software allowing spectral, textural, edge-detection, and landscape properties of the seafloor features to be applied in classification workflow. Spectral information draws on the separation of benthic habitats based on differences in reflectance characteristics. Texture based classification considers the systematic variation of brightness within a group of pixels and is a function of the seafloor feature. Edge-detection is a process whereby boundaries are identified within an image corresponding to where brightness changes sharply across a narrow spatial threshold. Edge-detection was used principally to streamline processing through identifying objects with clearly defined (i.e. crisp) boundaries. Landscape contextual editing draws on the fact that geomorphological and ecological zonation across a depositional system follows generic and logical rules (near-shore sediments for example are not encountered on the reef-edge). Within eCognition, a multi-resolution segmentation algorithm groups neighboring pixels, based on their spectral properties, to form polygons representing observed seafloor features. The resulting polygons are assigned classes representing different benthic habitats, as determined from available on-ground observations with known geographic positioning, either automatically through spectral and textural thresholding or manually by producer assignment. The polygons are exported from eCognition as ESRI shapefiles (*.shp) to create the final polygons of the benthic habitat map. This final map maintains the native 4 m resolution of the IKONOS imagery. Description Thematic classification of benthic habitats identified through remote sensing using the IKONOS sensor. Benthic habitats are categorized into one of the classes of the fourteen defined for this project with the filenames outlined below. Land - LAND, Continuous reef pavement-low relief - CONT_LR, Continuous reef pavement-medium relief - CONT_MR, Continuous reef pavement-high relief - CONT_HR, Isolated patch reef-low relief - ISOL_LR, Isolated patch reef-medium relief - ISOL_MR, Isolated patch reef-high relief, - ISOL_HR, Sand - SAND, Dense seagrass - DENSE_SEAGRASS, Sparse seagrass - SPARSE_SEAGRASS, Dead corals - DCOR, Gorgonian Hardgrounds - GORG, Low relief hard bottom - LRHB, Hard bottom pavement with cracks - LRHB-CRACK, For the purposes of marine spatial planning in Barbuda (the Barbuda Blue Halo Initiative) in 2013, these data were further classified with the following schema: New Classification - Original Classification, Sand - Sand, Seagrass - Dense seagrass, sparse seagrass, Patch Reef - Isolated patch reef (low, medium and high relief) Continuous Reef - Continuous reef pavement (medium and high relief), gorgonian hardgrounds, Hard Bottom - Continuous reef pavement (low relief), Low relief hard bottom, Hard bottom pavement with cracks, dead corals. The dataset AG004002 uses the new classification schema. For publication to EMODnet, SAERI reprojected the data to EPSG:4326 - WGS 84, cleaned the data using mapshaper.org (https://mapshaper.org/) to eliminate gaps and overlaps, and processed the structure to fit the EMODnet data exchange format. All work was completed in QGIS v3.28.4. |
EMODnet Product Catalogue