The use of remote sensing and GIS in the identification and vulnerability detection of coastal erosion as a hazard in False Bay, South Africa

[摘要] ENGLISH ABSTRACT: Coastal erosion is a worldwide hazard of which the consequences can only be mitigated viathorough and efficient monitoring of erosion and vulnerability to erosion. This study aimed toestablish the accuracy, efficacy and efficiency of various remote sensing techniques for thedetection and monitoring of coastal erosion and vulnerability occurring in False Bay, SouthAfrica. There is a need to monitor the erosion in this area as well as to determine the mosteffective techniques for monitoring the erosion in False Bay and other similar environmentsin the future. This study provides an assessment of the usefulness of different data sourcesand techniques for change detection in the coastal environment.The data sources used were Landsat TM/ETM+ imagery and aerial photographs. Imagedifferencing, tasselled cap transformations, vegetation index differencing, Boolean changedetection, and post-classification change detection were all performed on the Landsatimagery. The aerial photographs were assessed using the Digital Shoreline Analysis System(DSAS) add-on for ArcGIS which determines statistical differences in the shoreline positionas digitised in vector format.The results showed that while the resolution of the Landsat imagery was not sufficient toanalyse erosion along the beach itself, the larger area covered by the satellite imagesenabled vulnerability indicators to be seen. Notably, the post-classification change detectionindicated consistent increases in built-up areas, while sand dune, beach, and sand (notbeach) all decreased. NDVI differencing showed consistent decreases in NDVI indicatingdecreasing plant health and density. The results of image differencing with both band 4 andthe brightness band led to conclusions that vegetation health was decreasing while reflectivesurfaces such as bare sand and roads were increasing. All of these indicate an increasedvulnerability to coastal erosion. The Boolean change detection method was found not to beuseful in this case.Aerial photographs were studied on four focus areas: Bayview Heights, Macassar Beach,Strand, and Pringle Bay. The results showed erosion at all four areas, with Strandexperiencing only erosion (no accretion) at an average of 53 cm erosion per year. Erosion atMacassar Beach and Pringle Bay was also severe, with Bayview Heights being the leastsevere and showing a combination of erosion and accretion. The higher resolution availableon the aerial photographs was vital to view changes on the beach itself.In future studies requiring assessment of changes in the position or condition of the beachitself, aerial photographs or high resolution satellite data should be used. Studies ofvulnerability extending over the entire coastal zone may make use of Landsat TM images.Post-classification change detection provides powerful change direction information and canindicate the percentage of area change from one class to another. However, imagedifferencing and vegetation index differencing are much faster to perform and can provideinformation about general trends in the changes occurring. Therefore post-classificationchange detection might be used in areas of high and rapid change while image differencingand vegetation index differencing can be useful to cover vast areas where little change isexpected.

[发布机构] Stellenbosch University