Retrieving the Leaf area Index of Individual Trees and Stands using Single-Scan Data From a Terrestrial Laser Scanner

Abstract

Terrestrial laser scanning (TLS) has been proven to be a promising tool for estimating the leaf area index (LAI). However, the TLS-based LAI accuracy has been influenced by various factors. Therefore, it has been the research hotspots to improve the accuracy of TLS-based LAI in the current study. In this study, we proposed algorithms about the retrieval of gap fraction, clumping index $\Omega$ and leaf inclination respectively to improve the TLS-based LAI accuracy, and the final accuracy of improved TLS-based LAI was validated by field measurement data in individual and stand level respectively. In individual level, ground validation data was collected by destructively sampling 35 trees and measuring all of their leaves. While the ground validation data in stand level was collected by Tracing Radiation and Architecture of Canopies (TRAC) and hemispherical photography (HP). Results showed that taking clumping index $\Omega$ into account could improve the TLS-based LAI accuracy with a decrease of 0.32 in RMSE. Conversely, different projection coefficient models showed little effect on the accuracy of LAI estimation. Final results showed that TLS-based LAI estimations based on the improved algorithms were strongly correlated with the destructively sampled LAI measurements in individual level $(\pmb{R}^{\mathbf{2}}=\mathbf{0.78},\pmb{RMSE}=\mathbf{0.44})$ . Moreover, in stand level, TLS-based LAI estimations were in relatively good agreements with the ground truth collected by $\mathbf{HP}(\pmb{R}^{\mathbf{2}}=\mathbf{0.90},\pmb{RMSE}=\mathbf{0.56})$ and in poor agreements with the ground truth collected by TRAC $(\pmb{R}^{\mathbf{2}}= \mathbf{0.31},\pmb{RMSE}=\mathbf{0.83})$ .