Remote sensing techniques for mangrove mapping

Mangroves, important components of the world's coastal ecosystems, are threatened by the expansion of human settlements, the boom in commercial aquaculture, the impact of tidal waves and storm surges, etc. Such threats are leading to the increasing demand for detailed mangrove maps for the purpose of measuring the extent of the decline of mangrove ecosystems. Detailed mangrove maps at the community or species level are, however, not easy to produce, mainly because mangrove forests are very difficult to access. Without doubt, remote sensing is a serious alternative to traditional field-based methods for mangrove mapping, as it allows information to be gathered from the forbidding environment of mangrove forests, which otherwise, logistically and practically speaking, would be extremely difficult to survey. Remote sensing applications for mangrove mapping at the fundamental level are already well established but, surprisingly, a number of advanced remote sensing applications have remained unexplored for the purpose of mangrove mapping at a finer level. Consequently, the aim of this thesis is to unveil the potential of some of the unexplored remote sensing techniques for mangrove studies. Specifically, this thesis focuses on improving class separability between mangrove species or community types. It is based on two important ingredients:(i) the use of narrow-band hyperspectral data, and(ii) the integration of ecological knowledge of mangrove-environment relationships into the mapping process.Overall, the results of this study reveal the potential of both ingredients. They show that delicate spectral details of hyperspectral data and the spatial relationships between mangroves and their surrounding environment help to improve mangrove class separability at the species level. Despite the optimism generated by the overall results, it was found that appropriate data treatments and analysis techniques such as spectral band selection and noise reduction were still required to harness essential information from both hyperspectral and ecological data. Thus, some aspects of these data treatments and analysis techniques are also presented in this thesis. Finally, it is hoped that the methodology presented in this thesis will prove useful and will be followed for producing mangrove maps at a finer level