Classifying Complex Mountainous Forests with L-Band SAR and Landsat Data Integration: A Comparison among Different Machine Learning Methods in the Hyrcanian Forest

Integrating L-band SAR and Landsat data significantly improves the classification accuracy of forest stand age classes in complex mountainous regions, such as the Hyrcanian Forest. The study found that combining these datasets achieves an overall accuracy (OA) of 86%, a notable improvement over using either dataset alone or traditional classifiers like maximum likelihood classification (MLC). Among machine learning methods, support vector machines (SVM) and random forest (RF) outperformed neural networks (NN), with SVM and RF showing the highest robustness and accuracy. Terrain correction was crucial, as non-corrected data failed to differentiate forest classes effectively (OA = 65%). The research highlights the potential of multisource remote sensing for precise forest classification in challenging terrains.

Attarchi, S., & Gloaguen, R. (2014). Classifying Complex Mountainous Forests with L-Band SAR and Landsat Data Integration: A Comparison among Different Machine Learning Methods in the Hyrcanian Forest. Remote Sensing, 6(5), 3624–3647. https://doi.org/10.3390/rs6053624