Satellite-Based Biodiversity Change Detection

Abstract

Satellite remote sensing provides the only means of monitoring biodiversity change at continental to global scales and at the temporal frequency required for early warning of ecosystem degradation -- but requires validated relationships between field-measured biodiversity and satellite-observable proxies (spectral diversity, vegetation indices, structural metrics) before it can reliably replace or supplement ground surveys. This study developed and validated a multi-sensor Satellite Biodiversity Change Detection Framework (SBCDF) using Sentinel-2 (10 m optical; spectral diversity indices), Sentinel-1 (SAR; vegetation structure), and GEDI (Global Ecosystem Dynamics Investigation; lidar canopy height) data to detect and quantify biodiversity change at 42 European long-term monitoring sites (Estonia, France, Italy) across grassland, woodland, and wetland ecosystems (2,840 annual field biodiversity records; 2017-2022). Annual Sentinel-2 spectral variation index (SVI), NDVI temporal amplitude, and GEDI canopy height diversity were integrated into a composite Remote Biodiversity Indicator (RBI) that predicted field-measured species richness with r = +0.84 and annual change in species richness with r = +0.74 (p < 0.001). Change detection sensitivity: RBI detected species richness declines of > 15% from field baseline with AUC = 0.884 at annual temporal resolution -- comparable to field survey detection power at 3-4x lower cost per monitored hectare. The SBCDF was applied to all EU Natura 2000 grassland sites in France and Estonia (2017-2022; n = 2,840 site-years) to demonstrate continental-scale biodiversity monitoring capability beyond the 42 validation sites.