Estimation of aboveground biomass of savanna trees using quantitative structure models and close-range photogrammetry

Abstract

In efforts to mitigate climate change and optimize resource management, the demand for accurate aboveground biomass (AGB) estimates has significantly increased. Traditional AGB estimation methods rely on allometric models, which have inherent limitations. Recent advancements in remote sensing technologies present new opportunities for obtaining precise and nondestructive AGB data. This study evaluated the accuracy of AGB estimates derived from close-range photogrammetry (CRP), comparing it with destructive sampling and allometric equations. Thirty trees from five Sudanian savanna species, spanning six diameter classes, were photographed with a handheld camera. Images were processed to reconstruct 3D models of the trees, from which tree volume was calculated using quantitative structure models (QSM) and converted to AGB with species-specific wood density. Agreement between reference and estimated AGB was assessed using coefficient of variation of root mean square error (RMSE%), mean absolute bias (MAB) and concordance correlation coefficient (CCC). CRP-derived AGB closely matched with reference data (RMSE% = 23.4%, CCC = 0.98, MAB = 241 kg) and outperformed pantropical (RMSE% = 81.6%, CCC = 0.62, MAB = 694 kg) and regional (RMSE% = 74.3%, CCC = 0.70, MAB = 640 kg) allometric models. Accuracy varied by tree size, with CRP performing best for trees with DBH ≥ 30 cm. These results demonstrate CRP's effectiveness in AGB estimation for Sudanian savanna trees and its potential for timely, accurate, and scalable assessments across diverse ecosystems.