Estimation of Carbon Storage Based on AAV Multispectral Photogrammetry: An Offshore Islands Study in Pearl River Delta

Abstract

Aboveground biomass (AGB) of forests is a crucial metric for assessing ecosystem carbon storage and cycling. The geographical complexity of offshore islands, especially their perennial cloud cover, poses a challenge to traditional AGB remote sensing, so unmanned aerial vehicle (UAV) based remote sensing technology is particularly important. However, UAV approaches face limitations from ground interference, structural parameter errors, and allometric equation discrepancies. This study develops a framework integrating UAV photogrammetry and the Mask R-CNN to identify tree species, quantify forest structure, and estimate AGB in a representative offshore island to infer the impact of tree species differences, dominated by allometric equations and structural parameter errors, dominated by canopy occlusion on the estimated AGB of island forests. Compared to the AGB results calculated by species identification of individual trees, multispectral sensors effectively identified four dominant tree species and land cover, as general allometric equations resulted in significant errors ranging from − 68 % to + 36 %. The optimized canopy height model (CHM) approach revealed a 5.6 % underestimation of tree height and 10.0 % error in diameter at breast height (DBH) due to canopy occlusion. This study provides new insights into AGB estimation and forest species identification methodologies, which are important for studying carbon management activities in similar ecosystem types.

Plant phenotyping relevance

UAVマルチスペクトル写真測量とMask R-CNNを用いて樹種、樹木構造、樹高・DBH・地上部バイオマスを推定する手法を開発し、誤差も評価しており、植物表現型取得が中心である。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.