Mapping coral-algal dynamics in a seasonal upwelling area using spaceborne high resolution sensors

PROBA/CHRIS is one of the first satellite sensors to offer both high spatial and spectral resolutions. We explored the potential of this sensor to map the dynamics of seaweed and coral cover in an area influenced by seasonal upwelling in the Arabian Sea. Quantitative field assessments coincided with image acquisitions. After removal of sensor noise and atmospheric effects, maximum likelihood supervised classification yielded a tau accuracy of 64.09 for the summer monsoon dataset. Clearer waters and a lower spatial heterogeneity in the winter monsoon dataset resulted in a tau accuracy of 71.45. Post-classification comparison and vegetation indices illustrated the conspicuous turnover from dense macroalgal stands covering nearly all coral communities during summer to bare rock or turf communities during winter, with coral becoming the predominant bottom type. These results were further analysed using a novel maximum entropy sub-pixel approach and were shown to consistently outperform results from Landsat 7 ETM+ imagery.