Logo: to the web site of Uppsala University

Economic and technological advancements driven by the mining sector have come at a significant environmental cost, exacerbated by past and present regulatory shortcomings. Growing awareness has led to increased research on monitoring and restoring legacy mine sites. Hyperspectral imaging (HSI) has emerged as a valuable tool in the mining life cycle, especially post-mining environments. By capturing detailed spectral data, HSI detects variations in crystal structures and chemical attributes, enhancing data-driven mapping and providing detailed insights into site conditions. 

This work investigates the potential of hyperspectral imagery at multiple scales for monitoring the legacy Uranium mine, Mary Kathleen, in Queensland, Australia. Specifically, it examines hyperspectral data from Uncrewed Aerial Systems (UAS, i.e., drones) and satellite (Environmental Mapping and Analysis Program – EnMAP) to assess their effectiveness in mapping surface patterns related to contamination dispersal in key areas of the tailings storage facility (TSF), evaporation pond, and surrounding environment. UAS data was captured over the site in September 2023 with the HySpex Mjolnir VS620 across the visible and near-infrared (VNIR) and short-wave infrared (SWIR) wavelength ranges (400 – 2500𝑛𝑚), EnMAP data was collected in November 2023 and May 2024. While UAS captures high-resolution details over limited areas, the use of EnMAP data extends surveys across larger areas and enables multi-temporal observations. Using knowledge-based (Band Ratios – BR and Minimum Wavelength Mapping – MWL) and data-driven (Spectral Angle Mapper – SAM) spectral processing techniques, UAS- and EnMAP-based hyperspectral data were analyzed. The resulting maps demonstrate the ability of HSI to highlight subtle variations in surface properties, and also reveal limitations in both UAS- and satellite-based HSI data. The analyses show HSI’s potential to provide valuable insights and to narrow down monitoring efforts to areas of potential environmental concern. 

The UAS imagery yields reliable maps for all the investigated key minerals and their surface distribution patterns. The accumulation of evaporitic sediments suggests surface reactivity across the site and can be effectively mapped using BR, highlighting areas of interest. Simultaneously, SAM distinguishes different mineral endmembers across the site, revealing through occurring spatial association that some minerals may be intermediary products in evaporite formation. MWL emerged as a reliable method for mapping clay distribution, with occurrences identified in several mixtures. Localized accumulations of ferric oxides were effectively mapped using BR, although limitations in correction methods for the co-registration of data in the VNIR and SWIR wavelength ranges currently restrict the analysis. The EnMAP data shows potential for depicting patterns associated with reactive surfaces, however the products were found to be limited in their ability to complement and extend insights beyond the sites surveyed with the UAS. The main challenge relates to the impact of vegetation cover and the reduced number of effective masked pixels. Furthermore, a temporal effect of data correction between scenes from 2023 and 2024 highlights the need for more standardization in data collection and correction protocols. At the time of publication, the mapping outputs were verified based on site-specific expert knowledge provided by the Geological Survey of Queensland (GSQ), while mineralogy and geochemistry analyses are underway to effectively validate mineral endmembers and interpretations. This study is part of an ongoing research within the M4Mining project consortium. The findings establish a foundation for future research efforts and highlight areas of development of methodology for real-time HSI UAS-based monitoring.