Forest Stand Types Classification Using Tree-Based Algorithms and SPOT-HRG Data

Forest types mapping, is one of the most necessary elements in the forest management and silviculture treatments. Traditional methods such as field surveys are almost time-consuming and cost-intensive. Improvements in remote sensing data sources and classification –estimation methods are preparing new opportunities for obtaining more accurate forest biophysical attributes maps. This research compares performance of three non-parametric and tree-based algorithms i.e. the Classification and Regression Tree (CART), Boosting Regression Tree (BRT) and Random Forest (RF) for general forest type mapping using semi high resolution of SPOT-HRG data. Using a systematic random sampling design in a small area of the Hyrcanian forests, tree and shrubs species were registered in 150 sample plots. The general forest types of plots were named based on frequency of dominant species methods. After geometric and atmospheric corrections of SOPT-HRG data, suitable image processing transformations were applied on main bands to produce general vegetation indices and principal components. A wall-to-wall forest type classification of processed bands was done using three nonparametric algorithms. The forest type maps were assessed using unused test plots. Results shows that RF algorithm compared to CART and BRT algorithms with overall accuracy of 70% and kappa coefficient of 0.63 could better classify the forest stand types, while the CART method had the lowest accuracy with overall accuracy of 60% and kappa coefficient of 0.51. A performance result of the BRT classifier shows that their result is slightly similar to RF classifier.