Remote sensing of vegetation surrogates for regolith landform mapping

In many environments, vegetation is known to reflect the mineralogy of the soils in which it grows. Similarly, regolith landforms, which are a function of underlying geology and localized geomorphological processes, affect vegetation community composition and structure. Eleven primary regolith landform units were mapped during fieldwork in the Goldfields region of Western Australia, and the overlying vegetation units were characterised in detail. The presence and abundance of gold in the landscape is closely associated with particular regolith landform units. Results revealed quantifiable differences in species composition and diversity, vegetative cover, vegetation frequency and observable differences in soils and leaf litter accumulation for each of the regolith units. Spectral analysis of all vegetation components was also undertaken using a laboratory spectrometer. Landsat TM and ASTER imagery were used to assess how well the various regolith landform units could be mapped through remote sensing of the vegetation. Very high classification accuracies of approximately 90% were achieved in mapping regolith units through the remotely sensed imagery used in the study. Despite high levels of vegetative cover over most units, spectral unmixing of scene components proved to be non-viable. Remote sensing of vegetation as a mapping surrogate for regolith materials in the Goldfields shows significant promise, as well as being environmentally benign and a far less costly and laborious means of determining where to direct gold exploration efforts.