Modeling forest stand attributes using Landsat ETM+ and QuickBird satellite images in western Turkey

Abstract

This study assessed the suitability of Landsat ETM+ and QuickBird digital number values and various vegetation indices for predicting some structural parameters of forests in western Turkey. The empirical relationships between the structural parameters such as stand volume, basal area, tree density and quadratic mean diameter, and Landsat ETM+ and QuickBird satellite images were estimated using stepwise multiple regression analysis. Results indicated weak relationships between forest structural parameters and Landsat ETM+ images. The adjusted R² values of the regression analysis using the spectral digital number values for stand volume, basal area, tree density and quadratic mean diameter were found to be 0.37, 0.32, 0.44 and 0.25, respectively. Based on the vegetation indices, the adjusted R2 values of the regression analysis were attained as 0.36, 0.34, 0.28 and 0.17, respectively. However, the results demonstrated moderate relationships between the forest structural parameters and the QuickBird satellite image. The adjusted R² values from the regression analysis using the digital number values for stand volume, basal area, tree density and quadratic mean diameter were found as 0.57, 0.45, 0.29 and 0.30, respectively. Depending on the vegetation indices, the adjusted R² values from the regression analysis were obtained as 0.54, 0.41, 0.41 and 0.44, respectively. When the results from Landsat ETM+ and QuickBird satellite images are compared with each other, it could be stated that the QuickBird satellite images provide better representation of structural parameters of forests.