Coastal erosion mapping through intergration of SAR and Landsat TM imagery

It is important to monitor long-term coastal erosion in countries such as Australia given the majority of our population live in coastal regions. However, ground-based surveys are labour-intensive and involve significant cost when large spatial areas need to monitored. This paper presents a complementary, cost-effective approach for mapping the eroded shoreline, through integrating Landsat multispectral (MS) imagery data and synthetic aperture radar (SAR) imagery data. This integration method overcomes the problem of data shortage associated with using a single data source. Moreover, it can extract the instant land-water interface at sub-pixel resolution. Because the extracted land-water boundaries are dynamic, a tidal model has to be applied to define the high water line. Wavelet transform and linear spectral unmixing are the two major algorithms used for extracting the land-water interface. Several inundated areas are identified at the selected study area along the coast of East Gippsland Basin, Victoria, Australia. Naturally occurring erosion is believed to be the major factor for these inundated areas. Theoretically the extracted shorelines for defining erosion can reach 1m - 2m resolution approximately.