Exploring the Potential of Sentinel-1 C-band InSAR Coherence for Above-Ground Biomass Estimation in Dutch Forests

By Leonora Lesi

Abstract
Accurate mapping of forest above-ground biomass (AGB) is essential for climate change monitoring, carbon accounting, and sustainable forest management. This study evaluates the potential of Sentinel-1 C-band repeat-pass interferometric coherence to estimate AGB across diverse forest types in the Netherlands. Given the known challenges of using coherence data in temperate environments, we assessed how spatial and seasonal aggregation strategies can improve the coherence-AGB relationship. AGB reference data were obtained from the National Forest Inventory (NFI-7) and complemented by canopy height information from the AHN4 LiDAR dataset. Using Sentinel-1 coherence imagery from 2016 to 2021, we applied a non-linear exponential model to capture the saturating relationship between coherence and increasing biomass. First, we analysed the data by forest type and used the coefficient of variation (CV) in canopy height to represent spatial heterogeneity. Next, we evaluated the effects of spatial aggregation using multiple window sizes and explored both selective and seasonal multi-date combinations to assess the impact of temporal averaging. Results show that forest type significantly influences model performance, with evergreen coniferous stands giving the best fit. While only moderate spatial aggregation improved the coherence-AGB relationship, seasonal averaging confirmed that winter and autumn acquisitions led to better model fits, when reduced vegetation dynamics and atmospheric conditions minimized temporal decorrelation. Although the explanatory power of coherence alone remains limited in these conditions, the findings show the value of both spatial and temporal aggregation in improving C-band InSAR coherence sensitivity to AGB (maximum R² = 0.47).