Advancing trait mapping of Congo basin secondary forests using multispectral and hyperspectral satellite imagery

The forests of Central Africa remain understudied compared to those of the Amazon and Southeast Asia. The unique dominance of smallholders shifting cultivation as the driver of disturbance in the region also increases uncertainty about the future of the forest. This activity is poorly monitored and is associated with cycles of forest regeneration and deforestation, creating a complex mosaic of primary and secondary forests and agricultural fields. As slash-and-burn practices are expected to intensify with population growth in the region, understanding forest regeneration and the dynamics of secondary forests is essential, particularly for accurate carbon accounting.

Extensive studies on secondary forest morphological and physiological traits have been conducted using field inventory from logging concession and study plots throughout the basin. While detailed, these studies are limited spatially due to the constraint of field inventory. On the other hand, studies using satellite remote sensing are spatially extensive but lack the detailed view given by field measurements or are limited by an insufficient spatial resolution for the mosaic landscape of secondary forest.

In this study we evaluate the potential of high-resolution multispectral (Sentinel-2) and hyperspectral (EnMAP) for quantifying key morphological and physiological traits of secondary forests along successional stages in the Congo Basin. Using fields measurement in secondary forests plot and advanced Sentinel-2 processing algorithm, we explore the capability of satellite-based spectral data to capture and predict forest characteristics typically assessed through ground-based inventories. By correlating the spectral signatures from EnMAP with specific vegetation properties such as leaf area index and foliar biochemical properties we identify spectral indicators of structural and physiological development. The spectral analysis focuses on identifying the drivers of the shifts in reflectance patterns as the forest matures, linking spectral characteristics to ecological changes along the successional trajectory.

Preliminary results reveal a notable decrease in forest reflectance with age across the entire spectrum, indicating a darkening trend in the canopy as secondary forests mature. Forests aged 50-80 years show spectral signatures similar to those of primary forests, despite still exhibiting differences in structural traits and above-ground biomass. The retrieval of leaf properties like SLA and LNC is made possible with robust Sentinel-2 image processing.

This study highlights the potential of combining high-resolution multispectral and hyperspectral data to provide spatially extensive and detailed information on the structural and functional dynamics of secondary forests. Improving the ability to monitor secondary forest characteristics and their recovery trajectories in regions where shifting cultivation is prevalent.