Spatiotemporal Analysis of Vegetation Health and Moisture Dynamics in Rehabilitated Mining Quarries Using Satellite Imagery
Emmanouil Α. Varouchakis, Kostas Komnitsas, M. Galetakis
Abstract
Abstract This study evaluates the effectiveness of Normalized Difference Vegetation Index (NDVI) and Normalized Difference Moisture Index (NDMI) for monitoring vegetation health and soil moisture dynamics in rehabilitated mining quarries. The research focuses on developing an integrated framework of processing and analyzing these indices using remote sensing data and employing machine learning techniques to address challenges such as missing or defected pixels. Sentinel-2 Level-2 A imagery, known for its high spatial and temporal resolution, was employed to calculate these indices. To address potentially missing or affected pixels in such works a Self-Organizing Map (SOM) machine learning technique was utilized for data interpolation, ensuring accuracy and continuity. Cross-validation techniques were applied to validate the performance of SOM, yielding low Mean Absolute Error (MAE) values for NDVI and NDMI, indicating minimal bias in the interpolated data. Results demonstrate significant improvements in vegetation health and moisture retention in rehabilitated areas, though inconsistencies remain compared to natural sites. The combined use of NDVI and NDMI provided complementary insights, highlighting areas requiring further intervention. This methodology offers a solid framework for monitoring rehabilitation progress and guiding future restoration efforts, not only in rehabilitated mining quarries but also in other rehabilitated urban and industrial sites.