Abstract:

Thanks to the development of remote sensing technologies, different analysis methods, and software, it is possible to classify objects with high accuracy over satellite images. In this study, classification successes of land cover types and other class types were investigated by using a high resolution GeoEye-1 satellite image taken on September 3, 2011, of the 12X12 km study area belonging to Kastamonu Regional Directorate of Forestry, Central Operations Directorate. While the overall accuracy value was 90.59%, the kappa value was found to be 0.872. After the high-accuracy classified image obtained by using the object-based classification method with eCognition software, the stand map with 1/25000 scale raster data was converted into vector data with NetCAD v5.2 software and prepared as a base for comparisons. In the continuation of the application, the GeoEye-1 satellite image was vectorized manually and compared both visually and numerically with the classification results and the vector data of the stand map. As a result of the study, it was concluded that the stand map used in forestry activities is not sufficient for the precise classification of objects, since it shows both large-scale and low spatial resolution and only shows class types with large areas such as forest cover types. It has also been shown that the classified GeoEye-1 satellite image obtained with high visual and numerical accuracy can be easily used instead of stand maps in the field of forestry.

Keywords: