Coastal Remote Sensing: Monitoring intertidal and saltmarsh vegetation in coastal lakes using high resolution satellite imagery – a case study at Wallis Lake, NSW.
The impact of climate change and increasing population pressure on the extent and health of saltmarsh and mangrove communities in NSW estuaries needs to be quantified to establish baselines and track temporal trends. These tasks require mapping that is comprehensive, standardised and objective, and repeated over time for the 150 estuaries in the state which have these communities. Previously, repetitive mapping has been undertaken through labour intensive field-based observations, but the acquisition of satellite data, combined with a representative spectral database, might provide an alternative approach.
An overpass of Wallis Lake by a QuickBird satellite in Sept. 2008 produced an image (2.6m spatial resolution) that was validated with field observations of homogenous patches of estuarine vegetation, a priori knowledge and historical vegetation maps. Twelve vegetation classes were initially determined in the intertidal and supratidal zones using a standard supervised spectral angle mapping (SAM) technique. These were later combined into five output classes based on a need to separate the saltmarsh and mangrove vegetation from other vegetation. Classification results were compared to field observations to determine accuracy. Some types of vegetation were consistently mis-classified (e.g., mangrove was confused with other forest species, especially swamp oak) but saltmarsh was more consistently mapped.
The development of a method by which the extent of saltmarsh can be estimated with relative accuracy from high resolution satellite imagery can be applied to historical, current and future images, with limited or no additional fieldwork, to enhance the efficiency of monitoring and detecting change in coastal habitats. However, because mangrove is spectrally similar to other estuarine vegetation its classification technique needs additional refinement before reliable change detection can be established. In the near future, it is hoped that sensors with higher spectral resolution will enhance the separability of estuarine and terrestrial macrophytes, particularly if the improved imagery is combined with pattern recognition techniques.