308 resolution of Landsat TM, ETM+, and OLI sensors and represents a widely accepted standard for long-term environmental monitoring. All datasets were resampled to this resolution to preserve temporal consistency and ensure reliable interannual comparison. Higher-resolution Sentinel-2 imagery (10–20 m) was downscaled to 30 m, thereby aggregating multiple measured pixels into a single representative value, reducing noise and mimicking the observation scale of historical sensors. Landsat MSS imagery, with a native instantaneous field of view of approximately 60 m and commonly distributed at 80 m pixel size, was upscaled to 30 m in order to ensure spatial alignment across the multi-sensor time series, without implying increased spatial detail. Importantly, downscaling high-resolution data is scientifically preferable to upscaling older coarse-resolution imagery, as aggregation preserves real observations, whereas interpolation artificially introduces spatial detail that was never measured. This distinction is particularly critical for tailings facilities, which exhibit strong spatial heterogeneity, sharp boundaries, and dynamic interfaces between water, dry tailings, and vegetation. Upscaling coarse-resolution data would fragment mixed pixels into artificially sharpened patterns, leading to false precision and misleading change detection. By contrast, downscaling preserves consistent mixed-pixel behavior across all sensors, enabling robust interpretation of true spatiotemporal dynamics and supporting reliable multi-decadal change analysis of tailings facility evolution. Figure 3 – Resampling (Upscaling and Downscaling) Spectral analysis represents a core methodology in remote sensing, focusing on the interpretation of interactions between electromagnetic radiation and Earth surface materials to derive information about their physical and environmental properties. Variations in surface reflectance across different spectral wavelengths provide a quantitative basis for identifying, characterizing, and tracking surface features and their temporal evolution. Within this analytical context, spectral indices constitute a powerful means of extracting thematic information from multispectral satellite imagery. These indices are formulated through mathematical relationships between reflectance values recorded in selected spectral bands and are specifically designed to highlight particular surface characteristics. Spectral indices are widely applied in land cover mapping, vegetation and water monitoring, change detection, mineral and resource assessment, and environmental impact analysis. By evaluating the intensity and relative behavior of spectral
RkJQdWJsaXNoZXIy MTM0Mzk2