Detection of windthrow in mountainous regions with different remote sensing data and classification methods

After a disastrous storm event, quick and reliable information on the extent of forest damage is required. This study evaluated different remote sensing data and methods to detect windthrown forests in mountainous regions as an alternative to the manual analysis of aerial images or terrestrial methods. To this end, both optical satellite sensors (Landsat-7, Spot-4 and Ikonos) and synthetic aperture radar (SAR) data at various frequencies (X-, L-, P- and C-band) were evaluated, and classifications of the windthrown forests were performed. This study was designed to state the advantages and disadvantages of the investigated data and methods. Classification results were compared with aerial images which were interpreted manually on a stereoscopic base. The study showed that the manual interpretation of Ikonos data revealed the most accurate results, followed by an automatic classification of Spot-4 data. Except for ERS-1/2 data, which are too inaccurate in mountainous regions, and SAR P-band data, all sensors and methods investigated have different advantages, so the choice of a specific sensor and method will depend on the question being answered.