Multivariate detection and attribution of land-cover changes in the Central Pilbara, Western Australia
Taylor and Francis Inc.
School of Engineering
ECU, Edith Cowan University
The Multivariate Alteration Detection (MAD) method was applied to locate areas where land-cover changes occurred between 2003 and 2009 in the Central Pilbara, Western Australia. It was demonstrated that each of the six MAD variates contained information of land-cover changes at various spatial scales. This allowed attribution of the identified changes to particular stressors such as climate variability, fire events, and mining activity in the area. The results were analysed and interpreted using time series of multispectral normalized difference vegetation index, normalized difference wetness index, and normalized burn ratio grids derived from Landsat data observed over the study period. In addition, various ground truth data such as fire maps, historical climate data, and the available information about mine operations and water management, which could lead to alteration of natural water regime, were utilized.