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

Date
2014-04
Authors
Callaghan, Kerry Lee
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Coastal erosion is a worldwide hazard of which the consequences can only be mitigated via thorough and efficient monitoring of erosion and vulnerability to erosion. This study aimed to establish the accuracy, efficacy and efficiency of various remote sensing techniques for the detection and monitoring of coastal erosion and vulnerability occurring in False Bay, South Africa. There is a need to monitor the erosion in this area as well as to determine the most effective techniques for monitoring the erosion in False Bay and other similar environments in the future. This study provides an assessment of the usefulness of different data sources and techniques for change detection in the coastal environment. The data sources used were Landsat TM/ETM+ imagery and aerial photographs. Image differencing, tasselled cap transformations, vegetation index differencing, Boolean change detection, and post-classification change detection were all performed on the Landsat imagery. The aerial photographs were assessed using the Digital Shoreline Analysis System (DSAS) add-on for ArcGIS which determines statistical differences in the shoreline position as digitised in vector format. The results showed that while the resolution of the Landsat imagery was not sufficient to analyse erosion along the beach itself, the larger area covered by the satellite images enabled vulnerability indicators to be seen. Notably, the post-classification change detection indicated consistent increases in built-up areas, while sand dune, beach, and sand (not beach) all decreased. NDVI differencing showed consistent decreases in NDVI indicating decreasing plant health and density. The results of image differencing with both band 4 and the brightness band led to conclusions that vegetation health was decreasing while reflective surfaces such as bare sand and roads were increasing. All of these indicate an increased vulnerability to coastal erosion. The Boolean change detection method was found not to be useful 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 Strand experiencing only erosion (no accretion) at an average of 53 cm erosion per year. Erosion at Macassar Beach and Pringle Bay was also severe, with Bayview Heights being the least severe and showing a combination of erosion and accretion. The higher resolution available on 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 beach itself, aerial photographs or high resolution satellite data should be used. Studies of vulnerability extending over the entire coastal zone may make use of Landsat TM images. Post-classification change detection provides powerful change direction information and can indicate the percentage of area change from one class to another. However, image differencing and vegetation index differencing are much faster to perform and can provide information about general trends in the changes occurring. Therefore post-classification change detection might be used in areas of high and rapid change while image differencing and vegetation index differencing can be useful to cover vast areas where little change is expected.
AFRIKAANSE OPSOMMING: Kus-erosie is ‘n wêreldwye gevaar waarvan die gevolge slegs deur deeglike en doeltreffende monitering van erosie en kwesbaarheid vir erosie verminder kan word. Hierdie studie poog om die akkuraatheid, doeltreffendheid en effektiwiteit van verskillende afstandswaarneming tegnieke vas te stel vir die opsporing en monitering van kus-erosie en kwesbaarheid in Valsbaai, Suid Afrika. Daar is ‘n behoefte aan die monitering van erosie in hierdie area, sowel as om die mees doeltreffende tegnieke van die monitering hiervan in Valsbaai en ander soortgelyke omgewings in die toekoms te bepaal. Hierdie studie bied ‘n evaluering van die nut van verskillende data-bronne en tegnieke vir die opsporing van verandering in ‘n kusomgewing. Die data-bronne wat gebruik is, is Landsat TM/ETM+ beelde asook lugfoto’s. Beeld differensievorming, “tasselled cap” transformasies, plantegroei indeks differensievorming, Boolse verandering en post-klassifikasie verandering is toegepas op die Landsat beelde. Die lugfotos is ge-evalueer deur die Digitale Kuslyn Analise Stelsel (Digital Shoreline Analysis System – DSAS). DSAS is ‘n bykomstige sagteware vir ArcGIS wat statistiese verskille in gedigitaliseerde kuslyn posisie bepaal. Die resultate toon dat terwyl die resolusie van die Landsat beelde nie voldoende was om strand-erosie self te analiseer, die groter area wat deur die satellietbeelde gedek word toegelaat het om kwesbaarheid aanwysers te ontleed. Spesifiek die post-klassifikasie verandering het aangedui dat konsekwente toenames in beboude areas voorkom, terwyl afnames in sandduine, strand en sand-areas voorgekom het. NDVI differensievorming het konsekwente afnames in NDVI getoon, wat dui op afnames in die gesondheid en digtheid van plantegroei. Die resultate van die beeld differensievorming met beide Landsat Band 4 en die helderheid-band het gelei tot die gevolgtrekking dat die gesondheid van plantegroei afgeneem het, terwyl reflektiewe oppervlaktes soos oop sand en paaie aan die toeneem is. Al hierdie resultate dui op die verhoogde kwesbaarheid vir kus erosie. Die Boolse verandering metode is bevind om nie van nut te wees in hierdie geval nie. Lugfoto’s van vier fokus-areas is bestudeer: Bayview Heights, Macassar Strand, Strand en Pringlebaai. Resultate van die DSAS analise het gevind dat oorwegend erosie by al vier areas plaasvind, met Strand die enigste area wat slegs erosie (geen aanwas) ervaar teen ‘n gemiddelde koers van 0.53 m per jaar. Erosie by Macassar Strand en Pringlebaai was ook ernstig, terwyl Bayview Heights die minste erosie ervaar het, met ‘n kombinasie van erosie en aanwas. Die hoër resolusie beskikbaar deur die lugfoto’s was noodsaaklik om veranderinge in strand areas waar te neem. In toekomstige studies wat die assessering van verandering in die posisie of toestand van strande noodsaak behoort lugfotos of hoë-resolusie satellietbeeld data gebruik te word. Studies oor die kwesbaarheid van ‘n hele kusstreek kan wel gebruik maak van Landsat data. Post-klassifikasie verandering bied kragtige informasie oor die rigting van verandering en kan die persentasie van verandering van een klas na ‘n ander aandui. Beeld en NDVI differensievorming is egter veel vinniger om uit te voer en kan informasie rakende die algemene tendense in verandering lewer. Post-klassifikasie verandering kan dus gebruik word in gebiede van vinnige en beduidende verandering plaasvind, terwyl beeld en NDVI differensievorming nuttig kan wees om groot areas te dek waar min verandering verwag word.
Description
Thesis (MSc)--Stellenbosch University, 2014.
Keywords
Dissertations -- Geography and environmental studies, Theses -- Geography and environmental studies, Coast changes -- Geographic information systems -- South Africa -- False Bay, Erosion -- Geographic information systems -- South Africa -- False Bay, Coast changes -- South Africa -- False Bay -- Remote sensing, Erosion -- South Africa -- False Bay -- Remote sensing, UCTD
Citation