Combining spectral and textural remote sensing variables using random forests: predicting the age of Pinus patula forests in KwaZulu-Natal, South Africa

Abstract

In this study we examined the utility of a statistical technique, random forests, to combine spectral and texture variables to accurately predict the age of Pinus patula stands. Using the QuickBird panchromatic band (0.6 m), five texture variables were calculated using 12 moving window sizes. The spectral variables used in this study consisted of the QuickBird visible and near infrared (NIR) bands (2.4 m). Using random forests, various methods of combining the spectral and texture variables were evaluated. The best model was based on random forests with a backward variable selection process which selected only five variables (NIR, green, variance with a 3 × 3 window, red and blue) of the original 64 variables and obtained the best predictive accuracies (R 2 = 0.68).