Species-specific modeling of tree diameter at breast height using tree height and relative density with implications for remote sensing-based forest inventory

Abstract

Accurate estimation of tree diameter at breast height (DBH) is essential for forest monitoring, biomass modeling, and carbon accounting. While DBH is traditionally measured in the field, this approach is labor-intensive and costly, especially at large scales. In contrast, tree height can now be efficiently obtained from remote sensing platforms such as airborne LiDAR and photogrammetry, creating opportunities to estimate DBH indirectly. To address this, we developed a species-specific nonlinear framework to predict DBH from tree height and stand-level relative density (RD) in the mixed temperate forests of New Brunswick, Canada. Our analysis used 1807 trees from 653 permanent sample plots (1985–2014), representing six dominant species: Abies balsamea, Acer rubrum, Acer saccharum, Picea mariana, Picea rubens, and Picea glauca. Allometric (height-only) models explained part of DBH variation, with R² ranging from 0.15 to 0.35 (broadleaves) and 0.41–0.74 (conifers), but predictive accuracy was notably low for Acer rubrum and Acer saccharum. Incorporating RD as a competition index substantially improved model performance, with R² increasing to 0.85–0.89 (broadleaves) and 0.72–0.88 (conifers). Prediction errors (RMSE and MAE) consistently decreased, with broadleaves showing the greatest improvement compared to conifers, reflecting their stronger sensitivity to stand density. These findings demonstrate that combining tree height with RD provides reliable estimates of DBH across diverse species. The framework bridges ground-based inventory with remote sensing applications, offering a scalable approach for biomass estimation, stand density analysis, and sustainable forest management in temperate mixed-species forests. • Developed species-specific nonlinear models to predict DBH from tree height and RD. • Used 1807 trees from 653 permanent sample plots across New Brunswick (1985–2014). • Including RD as a competition index improved model accuracy (R² up to 0.89). • Broadleaves exhibited greater DBH response to stand density compared to conifers. • Framework supports scalable DBH estimation in mixed temperate forests.

Plant phenotyping relevance

樹高と相対密度から樹木DBHという明示的な植物形質を推定する種別非線形モデルを開発・評価しており、形質推定手法が研究の中心である。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.