Narrowband red-edge indices effectively monitor species-level mangrove health in multisensor time series

Abstract Spatiotemporal changes in mangrove health within urbanized estuaries are indicators of coastal resilience under increasing anthropogenic and climatic pressures. This study analyzes species-level mangrove health dynamics in Benoa Bay, Bali, an area experiencing significant hydrological alteration due to reclamation and infrastructure development. We used a multiscale, multisensor remote sensing approach (PlanetScope SuperDove, Sentinel-2 MSI, and Landsat 8/9 TIRS,) spanning the period from 2018–2025, to separate the spectral-physiological responses of six dominant mangrove species, namely, Sonneratia alba Sm., Rhizophora apiculata Blume, Rhizophora mucronata Poir., Bruguiera gymnorhiza (L.) Lam, Ceriops tagal (Perr.) C.B. Robinson, and Avicennia spp. We used a random forest classification model, which achieved a validation accuracy of 0.82, to stratify the study area. We subsequently analyzed a suite of 10 vegetation indices (VIs), prioritizing red-edge indices (IRECI, SeLI, and NDCI) alongside traditional broadband indices (NDVI and EVI) and thermal metrics (Tc), to distinguish structural biomass signals from physiological indicators of chlorophyll content and stress. We validated our statistics using one-way analysis of variance, Tukey’s Honestly Significant Difference post hoc analysis, and spatial autocorrelation metrics. Our results indicate that red-edge-based indices, particularly the inverted red-edge chlorophyll index and Sentinel-2 LAI Green Index (SeLI), significantly outperform traditional indices in discriminating species-specific health status, revealing gradual degradation in stands where the broadband NDVI indicates stability. We observed different health trajectories, while C. tagal exhibited a positive trend in physiological vigor, S. alba, the pioneer species most susceptible to hydrodynamic alteration, displayed signs of stagnation and decline in specific zones, which were correlated with sedimentation-induced pneumatophore stress. Cluster analysis (K-means) and Moran’s I revealed highly clustered spatial degradation patterns linked to proximity to reclamation infrastructure. Our results show that the red-edge region detects early physiological stress in mangrove ecosystems before structural canopy loss occurs and provides spatial data for the management of Ngurah Rai Forest Park.