Enhancing timber traceability through combining multi-element analysis and satellite-based forest disturbance products

By Maaike Uijttenboogaard

Abstract
Illegal logging undermines efforts to sustainably manage and conserve forests, with widespread environmental, economic and social impacts. Multi-element analysis (MEA), a forensic tool that looks at the chemical composition of wood, has shown promise for independently tracing the geographic origin of timber. However, its practical application is constrained by limited reference data and poor spatial resolution. This study had two main objectives to enhance MEA for origin assignment of timber in Central Africa: (1) generating continuous maps of trace element concentrations using spatial interpolation techniques, and (2) integrating satellite-based forest disturbance products, such as logging roads and RADD alerts, into an origin assignment model. Existing multi-element data from 22 concession sites across the Congo Basin were used as input for quantile regression forests (QRF) and kriging with external drift (KED). QRF performed slightly better, though both methods showed high uncertainty. Random forest (RF) classification models combining MEA with satellite-derived predictors outperformed the baseline model, improving assignment success and spatial precision. However, the results varied considerably across runs and model performance depended on the parameters used to calculate the number of RADD alerts. While the multi-source RF models show potential for improving timber tracing, further research is needed. This includes improving understanding of the relationships between covariates and trace elements for spatial interpolation, validating classification models and exploring how spatial interpolation outputs can be effectively combined with origin assignment models. Together, these developments are essential for the practical application of timber tracing tools and for supporting legal enforcement in the timber trade.