Development of Uav Spectral Models for Estimating Leaf Water Content at Canopy Level for Eucalyptus Globulus and Pinus Radiata in Central Chile

Abstract

The estimation of leaf water content (LWC) is essential for understanding the physiological status of forest species and assessing fire hazards.This study explores the potential of unmanned aerial vehicle UAV-acquired multispectral imagery for LWC estimation in Pinus radiata and Eucalyptus globulus, two common species in mediterranean forest plantations. Leaf samples were collected from two plots in the Valparaíso region of Chile, and the actual leaf water content was measured using gravimetric methods. To develop and train regression models, both classical and machine learning techniques were applied, including Linear Regression, Mixed Linear Regression, Random Forest (RF), Support Vector Regression (SVR), and k-Nearest Neighbors (k-NN). These models were trained to predict LWC based on the spectral reflectance data obtained from the UAV imagery.Results from both LWC and spectral models highlight distinct physiological traits of each species, with Pinus radiata exhibiting more conservative water strategies and higher water retention compared to Eucalyptus globulus. The best-performing model for Eucalyptus globulus was a mixed linear regression, achieving an R2 of 0.69, an RMSE of 9.38, and a bias of -0.86. For Pinus radiata, the best model was a simple linear regression with an R2 of 0.33, an RMSE of 10.29, and a bias of 2.35. The Eucalyptus globulus model showed more consistent predictions of leaf water content and effectively mapped LWC.

Keywords: Leaf Water Content, Vegetation Assessment, Remote sensing, Drone Photogrammetry, Machine learning