Prediction of End Splitting in Log of Eucalyptus Nitens (H. Deane & Maiden) Maident through Regression Models Using Longitudinal Residual Strain, Physical and Dendrometric Properties

Abstract

The objective of this research was to develop regression models, with the purpose of estimating the growth stresses and end splitting, in 12-yr-old Eucalyptus nitens trees located in the province of Valdivia, Chile. A total of 40 trees were samples. Their longitudinal residual strain (LRS) was measured using the CIRAD-Forêt method and compared with the Council of Scientific and Industrial (CSIR) index and with a new index proposed in this investigation, called the splitting index. Physical and dendrometric properties of each selected tree were measured to identify possible correlations between these variables. These predictors were analyzed statistically, allowing the development of regression models, which showed a coefficient of determination (r2) of 0.783 and 0.770 to predict the LRS and CSIR indexes, respectively. Finally, a relatively robust regression model was developed to predict end splitting in logs using LRS and physical and dendrometric parameters.