Monitoring Ecological Restoration Dynamics in the Jharia Coalfield Through Remote Sensing and Geospatial Analysis

Abstract

The Jharia Coalfield (JCF), India’s most environmentally stressed coking-coal basin, has endured prolonged ecological degradation driven by open-cast mining, persistent coal-seam fires, and progressive vegetation loss. This study presents a novel multi-decadal assessment of ecological restoration dynamics across JCF from 2005 to 2025 using an integrated geospatial framework that combines Tasseled Cap Transformation (TCT), Spectral Mixture Analysis (SMA), Continuous Change Detection and Classification (CCDC), Normalised Burn Ratio (NBR), hotspot analysis (Getis–Ord Gi*), and a Random Forest (RF)-derived Ecological Restoration Score (ERS). Multi-temporal Landsat TM/ETM+/OLI and Sentinel-2 MSI data were processed within Google Earth Engine (GEE) at five-year intervals to derive TCT components (Brightness, Greenness, Wetness), sub-pixel fractional cover maps, and statistically significant landscape change events. LULC maps were generated using supervised RF classification with overall accuracies ranging from 87% to 91%. Hotspot analysis delineated persistent coal-fire risk clusters concentrated in the central and eastern sectors. Proximity-based buffer analysis revealed that substantial settlements lie within 500–1000 m of identified fire hotspots. ERS mapping confirms that 52.4% of the coalfield demonstrates moderate-to-high restoration probability, while 19.8% remains critically degraded. Findings highlight that active mining, persistent coal-seam fires, and inadequate post-reclamation management are the primary constraints on sustained ecological recovery and demonstrate the superiority of machine learning ensemble methods over conventional index-based approaches for mine-land restoration assessment.