The analysis and validation of landform stability using unmanned aerial vehicles

Abstract

Mine closure relies on proof of best practice in both design and performance of rehabilitation. Field techniques have been the traditional approach for producing detailed supporting empirical evidence for mine closure. Although field sampling provides a detailed snapshot of key performance criteria within small areas, these areas themselves may not be representative of the overall performance of rehabilitation. Additionally, their limited scale may miss broader spatial characteristics that could further strengthen arguments for relinquishment. Remote sensing is a complimentary approach to field sampling that can produce an entire census of a rehabilitation site at a reduced scale. However, uncertainty still surrounds the adoption of a remote sensing approach, and whether such techniques can capture key performance indicators accurately and consistently. This paper provides both a demonstration of capacity, and quantification of accuracy, of remotely sensed data analytics for the production of empirical evidence to support mine closure management. Using rehabilitated landforms in the Western Australian Goldfields as case studies, remote sensing was adopted in two supporting roles: The validation of landform construction, and the ongoing monitoring of landform performance. The geometry of constructed landform surfaces was measured through photogrammetric techniques and assessed against design specifications. Ongoing monitoring assessed both vegetative colonisation and relative stability of established landform surfaces. Coupled together, the broader scale impact of non-compliant areas upon local rehabilitation performance was explored and discussed. Underpinning these data analytics is the accuracy of remotely sensed data. The quantification of uncertainty within the data was derived through a comparison against precision field measurements. Quantification of this uncertainty allowed the establishment of confidence intervals on derived measurements. Furthermore, the impact of changing environmental complexity upon analysis performance was quantified. This allowed for the modelling of compensation factors that dynamically counterbalance the increased uncertainty of complex environments. The result of the study demonstrates the capacity for a remote sensing approach to empirically support mine closure and relinquishment.