# Developing compatibility biomass model based on UAV LiDAR data of Chinese fir (Cunninghamia lanceolata) in Southern China

## TL;DR

Researchers developed a biomass model for Chinese fir trees using UAV LiDAR data to improve biomass estimation accuracy across growth stages in southern China.

## Contribution

A Power Function model with age-based dummy variables and a SUR model were introduced to enhance biomass estimation accuracy.

## Key findings

- The Power Function model showed best performance for stem and bark biomass estimation.
- Introducing age groups increased R² by 2.6% on average, with bark and branch biomass seeing the largest gains.
- The SUR model improved consistency between individual and total biomass estimates despite slightly lower individual component accuracy.

## Abstract

Chinese fir (Cunninghamia lanceolata) is a key native tree species in southern China. Accurate estimation of above-ground biomass and its distribution is essential for the sustainable use of Chinese fir forests. UAV-based high-density point clouds and high-resolution spectral data provide critical remote sensing for detailed 3D tree structure analysis. This study aimed to explore the aboveground biomass allocation characteristics across the different growth stages of Chinese fir and to develop accurate biomass models. Measurements of 20,836 Chinese fir trees were used for the purpose. Through the comparative analysis of four basic models, the Power Function model was identified as the optimal one, particularly excelling in fitting the accuracy for stem and bark biomass. To further enhance the model’s fitting performance, age groups were introduced into the dummy model, categorizing the Chinese fir forests into the five distinct growth stages. Results showed age groups used as dummy variables led to an average increase in R² by 2.6%. The fitting accuracy for bark and branch biomass saw the most significant improvements, with increases in R² by 4.2% and 3.1%. To address the inconsistency between the sum of individual biomass components and total biomass, we employed a seemingly unrelated regression (SUR) model. Even though fitting accuracy for individual tree components decreased by an average of 2.5%, from a practical perspective SUR model would be more suitable for understanding the interrelationships between different components. These findings offer robust support for accurately estimating the aboveground biomass in Chinese fir forests across different growth stages.