Monitoring outcomes of rehabilitation of large-scale exploration disturbance: what is the role of remote sensing?

Abstract

Mineral exploration in Australia can extend across vast areas of land, often in remote locations. The nature of disturbance within exploration areas varies but typically includes vegetation clearing along tracks and around drill pads. In Western Australia, the Department of Mines, Industry Regulation and Safety recently announced a strengthening of its approach to rehabilitation compliance monitoring in relation to exploration (Government of Western Australia 2020). Compliance generally focuses on the prompt completion of activities such as plugging drill holes, filling sumps, removing waste, redistributing topsoil and surface scarification. However, less emphasis has been paid to the long-term (beyond six months) environmental outcomes of exploration rehabilitation, compared to rehabilitation of mine sites more broadly.Approaches to ecological monitoring are evolving in line with increasing access to high quality imagery with a range of spatial, spectral and temporal resolution, and advances in image analysis techniques. In 2020, Rio Tinto engaged Stantec to trial the use of a remote sensing approach for compliance monitoring of 33 exploration areas distributed across the Pilbara bioregion. The areas had been disturbed for exploration in line with conditions associated with exploration approvals and were distributed across an area exceeding 200,000 hectares (ha).The first step in the development of a remote sensing assessment involves choosing a suitable data capture approach. In this study, given the importance of accurately delineating live vegetation cover, and the largescale of the area to be assessed, multispectral (4-band) satellite imagery (SkySat, Planet) at 50 cm spatial resolution was selected. An object-based image analysis (OBIA) approach was applied to the captured imagery at each site in order to accurately quantify ground cover. Cover was allocated into separate classes: live vegetation, bare or sparsely vegetated ground, litter, shadow and any remaining 'other' features. Given that the spatial data available prior to the analysis was limited to drill pad points, rather than the entire disturbance boundary, OBIA was also used to delineate a disturbance area. The remaining area was categorised as analogue and its classification data was extracted separately for comparative purposes.The 33 areas assessed ranged in size from 30 to 39,000 ha. The proportion of each area detected by OBIA as 'disturbance' ranged from 0.1% to 30%. The number of drill pads located within each site ranged from six to nearly 600. At some of the smaller areas, the majority of drill pads were located within the OBIA-delineated disturbance area. However, at other sites, up to 90% of the drill pads were located outside the disturbance area, indicating that they had sufficient vegetation attributes to align more closely with the areas delineated as 'analogue'. A number of sites contained a range of drill pad ages, and at selected sites, a positive correlation between vegetation cover and rehabilitation age was established, an in...