Mapping benthic macroalgal communities in the coastal zone using CHRIS-PROBA mode 2 images

• Published in: Estuarine, coastal and shelf science coastal and shelf science, 2011-09, Vol.94 (3), p.281-290
• Main Authors: Casal, G., Kutser, T., Domínguez-Gómez, J.A., Sánchez-Carnero, N., Freire, J.
• Format: Article
• Language: English
• Subjects: Acoustic microscopes; Animal and plant ecology; Animal, plant and microbial ecology; benthic mapping; Biological and medical sciences; Brackish; Brackish water ecosystems; CHRIS; Classifications; Coastal; coastal zone; Communities; Construction; Differentiation; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; hyperspectral; macroalgae; remote sensing; Sand; Synecology; Teledetection and vegetation maps
• ISSN: 0272-7714; 1096-0015
• DOI: 10.1016/j.ecss.2011.07.008

The ecological importance of benthic macroalgal communities in coastal ecosystems has been recognised worldwide and the application of remote sensing to study these communities presents certain advantages respect to in situ methods. The present study used three CHRIS-PROBA images to analyse macroalgal communities distribution in the Seno de Corcubión (NW Spain). The use of this sensor represent a challenge given that its design, build and deployment programme is intended to follow the principles of the “faster, better, cheaper”. To assess the application of this sensor to macroalgal mapping, two types of classifications were carried out: Maximum Likelihood and Spectral Angle Mapper (SAM). Maximum Likelihood classifier showed positive results, reaching overall accuracy percentages higher than 90% and kappa coefficients higher than 0.80 for the bottom classes shallow submerged sand, deep submerged sand, macroalgae less than 5 m and macroalgae between 5 and 10 m depth. The differentiation among macroalgal groups using SAM classifications showed positive results for green seaweeds although the differentiation between brown and red algae was not clear in the study area. ► Three CHRIS images acquired in mode 2 were analysed to map macroalgae communities. ► Parametric and non-parametric rules were used to classify the images. ► Maximum likelihood classifier showed overall accuracies higher than 90%. ► Green macroalgae are differentiable from red and brown macroalgae. ► Differences between red and brown macroalgae are not clear.