SALTMARSH HABITAT CLASSIFICATION FROM SATELLITE IMAGERY

Saltmarsh habitats provide a good indicator of the condition of the coastal margin environment with respect to changing sea levels, saltwater inundation, air and water temperature, and land management practices. Fragmented saltmarsh vegetation is often mapped by field techniques, however these are both labour and time intensive and not amenable to the objective monitoring of continuous change. Airborne remote sensing requires extensive logistical preparation, and commissioning of overflights for small areas is expensive. Satellite imagery however is relatively easy to acquire, and once the processing chain has been established can be repeated on a regular basis and at varied sites allowing interannual variations in extent and dynamics to be identified. This research seeks to identify the optimum spatial, temporal and spectral resolutions required to discriminate key indicator habitats from satellite imagery, at sites in northern England and Wales. Pixel-based and object-based methods of distinguishing vegetation classes are compared for imagery products from a range of sensors and platforms. The potential benefits of integrating LiDAR data with the multispectral imagery are investigated, as are results obtained through both hard and fuzzy classification methods. These investigations are used to frame guidelines and recommendations for the analysis of saltmarsh habitats from space.