Improving carbon estimation in restored tropical forests using LiDAR

Forest restoration projects are gaining increasing attention as nature-based solutions, due to their potential to sequester carbon over time while simultaneously supporting biodiversity recovery, water regulation, and social benefits.

Carbon stocks are commonly estimated from forest inventories based on tree-level measurements of species identity, diameter, and height combined with allometric equations. While accurate at plot scale (~1 ha), this method is difficult to apply over large areas (>100 ha), relying on extrapolation and leading to uncertainties in landscape-scale carbon estimates.

LiDAR enables rapid coverage of large areas by generating high-resolution three-dimensional representations of forest structure, particularly when using unmanned aerial platforms with high point density. Although LiDAR-based models for estimating forest aboveground carbon are well established, most have been developed for mature or degraded forests in Amazonia. Consequently, models specifically calibrated for young restored forests and different restoration techniques are needed to improve accuracy and ensure the integrity and credibility of carbon estimates. Here, we develop a carbon modelling equation using LiDAR metrics for the specific context of restored forests.

We compared carbon estimates for 150 restored forest plots (including natural regeneration and planted) across the Atlantic Forest, Brazil, comparing aboveground biomass estimated from field inventories and allometric equations, with estimates from Airborne LiDAR data acquired by unmanned aerial vehicles. The LiDAR data was used to derive mean canopy height, which served as the primary structural metric for modelling the relationship between LiDAR measurements and field-based aboveground biomass estimates.

Our restored-forest LiDAR model explained 78% of biomass variability (R²cv = 0.78; RMSEcv = 1.67±1.19 KgC/m²) and estimated 52% higher carbon stocks at 10 m mean canopy height than the existing Amazonian-based model (Longo et al. 2016).

The improved performance of our restored-forest LiDAR model enables scalable and repeatable monitoring of carbon stocks across large areas, supporting decision-makers, project developers, and investors with more reliable and transparent estimates of climate mitigation benefits. These advances contribute to strengthening carbon accounting frameworks and the integrity of nature-based climate solutions.