Passive spectral bathymetry using satellite remote sensing in Cockburn Sound, W.A.

Abstract

Conventional bathymetric surveying is a costly and time consuming business. Even today many areas of shallow inshore ocean, some which encompass potential oil and gas fields, are only minimally charted. There is a need for reconnaissance systems which can effectively direct more expensive detailed surveys to best effect. Remote spectral bathymetry is one such system.A review of candidate sensor systems and processing algorithms highlighted problems due to changing bottom cover types and water quality parameters. A method, proposed and theoretically validated by other workers, was chosen for further investigation. This method develops an approximate relationship between the spectral content of the satellite data and water depths and then, by an iterative phase in the spatial domain, seeks to minimise the effect of spatially dependant variations.A study site in Cockburn Sound, Western Australia was chosen for a demonstration of this method. Spectral data are from the Landsat Thematic Mapper instrument and depth data are taken from Admiralty Charts. A variation on the originally proposed algorithm introduces spatial preprocessing phase, in which the image is segmented into zones where spectral relationships are expected to be more uniform. Two different methods of spatial mapping are used.The results demonstrate the capability of spatial modelling to improve remotely sensed depth estimates in the depth range of 5 to 12 m. The need for further research to better understand the shallow water spectral relationships is identified.