Tracking the temporal dynamics of insect defoliation by high-resolution radar satellite data

Quantifying tree defoliation by insects over large areas is a major challenge in forest management, but it is essential in ecosystem assessments of disturbance and resistance against herbivory. However, the trajectory from leaf-flush to insect defoliation to refoliation in broadleaf trees is highly variable. Its tracking requires high temporal- and spatial-resolution data, particularly in fragmented forests. In a unique replicated field experiment manipulating gypsy moth Lymantria dispar densities in mixed-oak forests, we examined the utility of publicly accessible satellite-borne radar (Sentinel-1) to track the fine-scale temporal trajectory of defoliation. The ratio of backscatter intensity between two polarizations from radar data of the growing season constituted a canopy development index (CDI) and a normalized CDI (NCDI), which were validated by optical (Sentinel-2) and terrestrial laser scanning (TLS) data as well by intensive caterpillar sampling from canopy fogging. The CDI and NCDI strongly correlated with optical and TLS data (Spearman's rho = 0.79 and 0.84, respectively). The Delta NCDIDefoliation(A-C) significantly explained caterpillar abundance (R-2 = 0.52). The NCDI at critical timesteps and Delta NCDI related to defoliation and refoliation well discriminated between heavily and lightly defoliated forests. We demonstrate that the high spatial and temporal resolution and the cloud independence of Sentinel-1 radar potentially enable spatially unrestricted measurements of the highly dynamic canopy herbivory. This can help monitor insect pests, improve the prediction of outbreaks and facilitate the monitoring of forest disturbance, one of the high priority Essential Biodiversity Variables, in the near future.

Automated summary

This study investigated the utility of satellite-borne radar (Sentinel-1) in tracking the fine-scale temporal trajectory of tree defoliation by insects. The radar data's canopy development index and normalized CDI were validated and found to strongly correlate with optical and TLS data, effectively explaining caterpillar abundance and discriminating between heavily and lightly defoliated forests. This high spatial and temporal resolution of Sentinel-1 radar potentially enables unrestricted measurements of dynamic canopy herbivory, providing valuable tools for monitoring insect pests and predicting outbreaks in forest ecosystems.