Improving marine habitat mapping using high-resolution acoustic data; a predictive habitat map for the Firth of Lorn, Scotland

• Published in: Continental Shelf Research
• Main Authors: Boswarva, Karen, Butters, Alyssa, Fox, Clive, Howe, John, Narayanaswamy, Bhavani
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2018
• Subjects: Marine predictive habitat mapping; Supervised classification; ecosystem-based managemnet; Maximun likelihood classification (MCL); Principal component analysis (PCA); Harbour porpoise; Phocoena phocoena
• Online Access: https://pure.uhi.ac.uk/en/publications/37f26f10-480a-4e2a-8679-f23b4cea467a; https://doi.org/10.1016/j.csr.2018.09.005; https://pureadmin.uhi.ac.uk/ws/files/3308476/1_s2.0_S0278434318302474_main.pdf

Habitat mapping is an important tool for marine spatial planning, enabling anand is required for most ecosystem-based management approaches to be adopted. The Firth of Lorn Special Area of Conservation (SAC), west Scotland, was originally designated for its rocky reef habitat but. iIt is also an area of high importance for the Flapper skate (Dipturus intermedius) and Harbour porpoise (Phocoena phocoena). Here we present a predictive habitat map for part of the area that improves on the previous predictive habitat map created for the SAC consultation by, utilising multibeam backscatter and bathymetry data collected as part of the Ireland, Northern Ireland and Scotland Hydrographic Survey Project, that . improves on the previous predictive habitat map created for the SAC consultation. Backscatter, bathymetry and benthic bathymetric derivatives were analysed using Principal Component Analysis (PCA) and acoustic signatures were created from biotope drop-down video habitat location data from drop-down video surveys. A predictive habitat map was created from Maximum Likelihood Classification using the PCA. Dominant habitat types identified included; moderate energy circalittoral rock (CR.MCR), sublittoral mixed sediment (SS.SMx) and sublittoral sand and muddy sand (SS.SSa). Drop down video showed variable accuracy with 0 to 100% correctly classified biotopes habitats due to the small sample size. The initial Vvalidation points were added and the model was rerun. In areas with no previous ground truthsea truthing points, some of the predicted output changed with the dominant habitat predictions changeding to SS.SMx at depths >100m and SS.SSa in depths <50m, suggesting that the model accuracy can be improved with a greater depth-range of ground truthsea truthing data . ModernBroad scale acoustic surveys undertaken for other reasons, such as navigational charting, can thus be used to generate Ppredictive habitat maps in a which are cost and time effective manner. Such mapssuch as this have the potential for a wide ...