TUBITAK Project, 2020 - 2021
Very high-resolution (VHR) stereo satellite images provide detailed spatial and spectral information for detecting and monitoring artificial or natural objects. This information can also be used for object-based change detection. This project aims to reveal the possibilities of using three-dimensional position information generated from very high resolution stereo satellite images in modeling Gordion tumuli and the potential of these data to determine the physical changes that may occur in tumuli. In this context, first of all, point cloud data were generated from WorldView-3 and KOMPSAT-3 stereo satellite images. The accuracy of these datasets has been tested using 164 control points obtained by land surveys. On the other hand, digital surface models produced from point cloud data were used for object-based modeling of tumuli. The accuracy of these models was tested by comparing them with reference tumulus models produced from unmanned aerial vehicle images. Finally, the potential of the produced tumulus models to determine the physical properties of Gordion tumuli was tested.
As a result of the vertical accuracy evaluation of the point cloud data, 0.16m and 0.50m root mean square error (RMSE) values were obtained for WorldView-3 and KOMPSAT-3 data, respectively. In addition, the comparison of tumulus models produced from stereo satellite images with reference tumulus models yielded 0.38m RMSE values for WorldView-3 and 0.70m for KOMPSAT-3. As a result of the comparison of the physical properties of the tumuli, it was determined that there is over 90% correlation between the tumulus models produced from satellite data and the reference tumulus models in terms of slope and elevation values. All these results show that tumulus models produced from stereo satellite data can represent Gordion tumuli quite sensitively. In addition, these results show that the sub-meter level physical changes that may occur in Gordion tumuli can be detected by very high-resolution stereo satellite images.