Application of ALOS AVNIR-2 for the detection of seaweed and seagrass beds on the northeast of Brazil.

Seaweed and seagrass beds play a multiplicity of functions within ecosystems, and have both ecological and economical value. However, anthropogenic activities, as well as climate change have contributed on the degradation. This study used data from Advanced Land Observing Satellite (ALOS) Advanced Visible and Near Infrared Radiometer Type 2 (AVNIR-2) orbital images to detect seaweeds and seagrasses. Qualitative and quantitative samples were used to validate the images, classified using the kappa coefficient (κ). The supervised classifications performed by the algorithms fuzzy logic – Fuzclass, maximum likelihood – Maxlike, minimum distance to means – Mindist, parallelepiped – Piped, showed an accuracy level of 0.93, 0.84, 0.83 (both excellent), and 0.62 (substantial), respectively. The results of the hard classifiers (Maxlike, Mindist, and Piped) submitted a new classification based on fuzzy logic (Fuzclass) demonstrated accuracy level of 0.74, 0.61 (both substantial), and 0.50 (moderate), respectively. Considered superior to the others, the Fuzclass classifier exhibited the best tendency in representing reef bottom–type distribution. Maxlike generated a map of seaweed and seagrass spatial distribution and abundance of the Maracajaú reef, identifying seven classes: (1) dense seaweeds; (2) sand; (3) dense seagrasses; (4) sparse seagrasses; (5) calcareous seaweeds; (6) sparse seaweeds; and (7) fine sand. The map of Maracajaú reef bottom type showed that it was possible to apply image processing and digital classification methodologies to distinguish submerged organisms, revealing information to help in planning and management of these ecosystems, enabling future monitoring. [ABSTRACT FROM AUTHOR]