Allometry-based estimation of forest aboveground biomass combining LiDAR canopy height attributes and optical spectral indexes

Accurate estimates of forest aboveground biomass (AGB) are essential for global carbon cycle studies and have widely relied on approaches using spectral and structural information of forest canopies extracted from various remote sensing datasets. However, combining the advantages of active and passive data sources to improve estimation accuracy remains challenging. Here, we proposed a new approach for forest AGB modeling based on allometric relationships and using the form of power-law to integrate structural and spectral information. Over 60 km2 of drone light detection and ranging (LiDAR) data and 1,370 field plot measurements, covering the four major forest types of China (coniferous forest, sub-tropical broadleaf forest, coniferous and broadleaf-leaved mixed forest, and tropical broadleaf forest), were collected together with Sentinel-2 images to evaluate the proposed approach. The results show that the most universally useful structural and spectral metrics are the average values of canopy height and spectral index rather than their maximum values. Compared with structural attributes used alone, combining structural and spectral information can improve the estimation accuracy of AGB, increasing R2 by about 10% and reducing the root mean square error by about 22%; the accuracy of the proposed approach can yield a R2 of 0.7 in different forests types. The proposed approach performs the best in coniferous forest, followed by sub-tropical broadleaf forest, coniferous and broadleaf-leaved mixed forest, and then tropical broadleaf forest. Furthermore, the simple linear regression used in the proposed method is less sensitive to sample size and outperforms statistically multivariate machine learning-based regression models such as stepwise multiple regression, artificial neural networks, and Random Forest. The proposed approach may provide an alternative solution to map large-scale forest biomass using space-borne LiDAR and optical images with high accuracy.