The Use of Spectral Unmixing of Landsat Imagery in Estimating Surface Runoff

Author: Ngoc Anh Nguyen

Nguyen, Ngoc Anh, 2016 The Use of Spectral Unmixing of Landsat Imagery in Estimating Surface Runoff, Flinders University, College of Science and Engineering

Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact with the details.


Urbanization and urban expansion has been a widely recognized problem throughout the world due to its negative influences on of land-use/land-cover changes. The growth of urbanization leads to impervious surfaces replacing natural landscapes. As impervious surfaces increase, surface runoff produced from rainfall and other sources also increases. This is one of the primary cause of urban flooding. Adelaide metropolitan area is the capital city of South Australia. The city is in a lowlying region and upstream catchments flow across the urban area. It still has much of the original stormwater drainage infrastructure, which was built in the period of 1940s to 1980s and the drainage systems have not been frequently reconstructed. This leads to an inadequate flow capacity or lack of drainage infrastructure, resulting in the increase of surface runoff and vulnerability to flooding. WetSpass model developed by Batelaan and De Smedt in 2001 is used for estimating surface runoff based on a water balance calculation. It assumes that each pixel in a landuse map contains proportions of land cover types (endmember), that are more or less conducive to surface water runoff; these include vegetation, impervious surfaces, bare soil, and open water. The proportions of those contributions are assumed to be the same for every pixel in a land-cover category. This could be subjective when applying the model to different study areas due to different characteristics of the landscape. This research introduces the use of Spectral Unmixing of Landsat imagery to produce fractional maps of these endmembers as substitute for using the land-use category alone in order to improve the efficiency of WetSpass model in estimating surface runoff over the Adelaide metropolitan area. The result of surface runoff estimation is used to help in indicating the areas vulnerable to flooding in the study area.

Keywords: RS, GIS, Remote sensing, Geographical Information System, WetSpass, Flood, Spectral Unmixing, Landsat, SUA, LULC, Adelaide

Subject: Environmental Science thesis

Thesis type: Masters
Completed: 2016
School: College of Science and Engineering
Supervisor: Stephen Fildes