Acid Mine Drainage (AMD) is a substantial source of water pollution, especially in regions with active mining operations. This phenomenon occurs when sulfide minerals in mined rock come into contact with air and water, resulting in the formation of sulfuric acid. The acidic runoff can dissolve Potentially Toxic Elements and other harmful substances from the surrounding rock, contaminating nearby water sources like rivers and streams. To tackle AMD effectively, mining operations must adopt measures such as proper handling and treatment of extracted materials, as well as implementing robust water management systems. Utilizing imaging spectroscopy offers a practical approach for mine assessment and characterization, serving as a valuable alternative to traditional chemical analyses. It focuses on identifying minerals that indicate subaerial oxidation of pyrite ("hot spots") and the resulting formation of sulfate minerals (such as jarosite) and their oxidation products (such as oxy-hydroxides and oxides).

Our cutting-edge approach is dedicated to enhancing the development of robust monitoring systems for Acid Mine Drainage (AMD) through the utilization of Machine Learning techniques applied to multi-temporal optical multispectral, and hyperspectral data. Specifically, our research delves into the utilization of hyperspectral data (PIKA L) obtained at various time points via Unmanned Aerial Vehicles (UAVs) and multi-temporal datasets from the Sentinel-2 satellite. This investigation has been conducted at the Lítov post-mine dump in the western region of the Sokolov lignite basin, Czech Republic. The Lítov dump stands out due to its highly acidic substrates, sparse vegetation, and the presence of a unique semi-desert environment. The accuracy of our Machine Learning classifications has been validated using ground truth data (mineralogy resolved by X-ray diffraction), demonstrating that the Radial Basis Function Support Vector Machine (RBF SVM) and Random Forest (RF) models surpass other ML approaches in performance when effectively identifying the main AMD hotspots and providing good separation between classes. More specifically, when using both hyperspectral Pika L and Sentinel-2 data RBF SVM and RF classifiers excel at detecting the AMD discharge, as well as mixtures between different mineral classes indicating the increasing pH e.g. oxy-hydroxides and oxides.

Future work will focus on testing ML techniques on extended multi-temporal data and evaluating the transferability of the model to other geographical locations (e.g., Greece).

The increasing demand for metals essential for net-zero technologies has led to as surge of intense mining activities, resulting in substantial waste production. Effectively managing this waste is crucial for minimizing environmental impacts and reducing greenhouse gas (GHG) emissions. This study investigates the potential for carbon sequestration in mine waste through mineral carbonation, focusing on silicate-rich waste from ultramafic ore deposits in Northern Europe.

Mine Environment Management Ltd. in collaboration with Geochemic Ltd., and Cardiff University, are developing innovative methods to measure CO₂ sequestration and emissions from mine waste. Based on previous work (e.g., Shiimi et al., 2023; Schoen et al., 2023), this project brings new insights to waste characterisation and properties, using closed-system experiments coupled with open-system equipment testing to simulate future field trials. Waste characterisation involves detailed mineralogical, elemental, geotechnical, and geochemical analyses, while closed-system experiments utilize bespoke sealed cells and Xylem-WTW Oxitop® devices to monitor CO₂ and O₂ flux.

Key findings from the closed-system sealed cell experiments revealed dynamic changes in CO₂ and O₂ concentrations in ultramafic waste rock over a two-year monitoring period. Initially, O₂ concentration decreased from atmospheric 21% to a negligible concentration, reflecting sulfide oxidation. To date, O₂ levels remain negligible, indicating a depletion of O₂ within the closed system. Conversely, CO₂ levels increased from 399 mg/L to 14,983 mg/L due to carbonate dissolution, before steadily declining as carbonation processes became dominant within the closed system. These results highlight the potential for enhanced weathering and effective CO₂ sequestration over time.

To upscale monitoring to mine site field conditions, an open barrel (OB) equipment test has been developed to examine the gas flux and geochemical behaviour of mine tailings under partially controlled, open-atmospheric conditions. This system integrates sensors to monitor CO₂ and O₂ levels, temperature, pH, and moisture content, yielding key data for real-world applications. The field trials aim to inform carbon sequestration strategies and contribute to the EU-funded C-SINK project by establishing Monitoring, Reporting, and Verification (MRV) protocols for future carbon dioxide removal (CDR) efforts. This study demonstrates that carbonation in mine waste has the potential to reduce CO₂ emissions, supporting the development of scalable carbon management strategies in the mining industry.

Mining environments involve complex hydro-bio-geochemical systems. Reactive transport modeling (RTM) is essential to rigorously describe these processes. Yet, process-based RTM is computationally intensive and limited in practical applications. To mitigate such challenges, this paper provides a novel deep learning-based surrogate accelerator, hidden-reactive-transport-neural-network (HRTNet), to simulate pyrite oxidation, a process of key importance for acid mine drainage. HRTNet relies on a flexible two-network architecture integrating chemical and physical equations. The model can effectively capture the desired spatio-temporal dynamics in a considerably reduced computation time (almost eight-fold). Additionally, HRTNet shows a good generalization capability covering a wide range of conditions beyond the training datasets.