Acid Mine Drainage Impoundments in the Nkangala District South Africa

Hypothesis: Temporally stable mine-site impoundments (pit lakes, tailings, return, and seepage ponds) accumulate acid mine drainage (AMD) with higher rare-earth element (REE) concentrations than natural waters which are contaminated by AMD. These can be systematically mapped, screened, and quantified to prioritise recovery and remediation. What the data are: A spatial inventory of AMD impoundments across the Nkangala District, South Africa. The core product is a polygon layer of 2,886 AMD-containing waterbodies, with a detailed attribute table including geometry, morphometrics, monthly Sentinel-2 reflectance statistics (2024), and screening variables for AMD/REE potential. Key fields: Geometry; area (area_m2, area_km2); coordinates; reflectance statistics (mean, min, max, stdDev, selected indices); estimated depth and volume (m³, L) from DEM-based shoreline proxy; water type label (pit lake, tailings, seepage, return pond); monthly reflectance band stats (Jan–Dec 2024) for B1–B12 and B8A. How it was built: Mining boundaries were refined in QGIS by reconciling multiple datasets. In Google Earth Engine, Sentinel-2 L2A imagery (cloud/shadow masked) was processed; the AWEI index was selected after testing and applied with a threshold > −0.35. Features ≥ 4 pixels (~400 m²) and present in ≥ 50 % of 2024 monthly composites were retained. Polygons were vectorised, some were visually classified, and linked to morphometric data derived from Copernicus GLO30 DEM shoreline buffers. Temporal medians informed conservative depth proxies; volume = area × depth. Monthly reflectance values were computed per polygon across B1–B12 and B8A (± 7 days). 2023 field sampling inside vs outside mining boundaries confirmed higher sulfate, lower pH, and measurable REEs within mine sites compared to bodies sampled within a 10km radius of a mine. What the data show: AWEI combined with temporal persistence filters delineated 2,886 AMD impoundments across the study area. DEM benchmarking reproduced approximately 78–80 % of known capacities, supporting the reliability of the volumetric estimates. Indicative ∑REE concentrations derived from reflectance-based comparisons suggest an average of 419 mg/L in these AMD waters, with lows of 89 mg/L. How to interpret: Best used as a screening layer to: map AMD extent, target high-priority ponds for sampling, treatment, or REE recovery, and support techno-economic scoping. Volumes are lower-bound estimates (based on 30 m DEM); concentrations are indicative, not assay results. Users should pair this layer with site visits, water chemistry, and high-resolution elevation data. Format: GeoJSON (EPSG:4326) Software: QGIS, Google Earth Engine. Limitations: DEM resolution, empirical thresholds, sparse paired reflectance–chemistry samples, and no flow or residence-time data. Volumes are conservative; REE values are screening outputs.