Investigation of the impact of different elevation data on horizontal positional accuracy in orthoimage production: The case of Göktürk-1 satellite

With the rapid advancement of technology, the ease of data acquisition and processing has significantly increased the strategic importance of remote sensing technologies. Through Turkey's growing investments in space technologies, satellite projects such as Göktürk-2, Göktürk-1, and Göktürk 2B have enabled the acquisition of high-resolution data from the Earth's surface, providing substantial advantages for both military and civilian applications. The capability of satellite imagery to collect data over wide areas while reducing the need for conventional ground-based methods makes this technology both economically and operationally efficient. In this context, the effective utilization of remote sensing technologies and their transformation into high value-added products has become increasingly important. A comprehensive and comparative study is required to evaluate how orthorectified images derived from new-generation satellites such as Göktürk-1 are affected by the use of different elevation data and production scenarios, particularly in regions with complex topography. In this thesis, the horizontal positional accuracies of orthorectified images generated using different elevation datasets were evaluated and compared by employing tri-stereo high-resolution imagery obtained from the Göktürk-1 satellite. Within the study, Digital Terrain Elevation Data (DTED-2), Advanced Land Observing Satellite (ALOS), TanDEM-X High Resolution Elevation Data Exchange (TREx), and Digital Elevation Models (DEMs) derived from Göktürk-1 stereo pairs were used. Two production scenarios were implemented: one without Ground Control Points (GCPs) and one incorporating GCPs. The study area covers the districts of Kepez, Konyaaltı, and Döşemealtı in the Antalya province, which exhibit high topographic variability. As reference data, 20 control points with a horizontal accuracy of 0.5 m and a vertical accuracy of 1 m—derived from balanced stereo aerial photographs provided by the General Directorate of Mapping—were used. The accuracy of the produced orthorectified images was assessed by comparing them with these reference points, and horizontal positional accuracy statistics were calculated for each elevation dataset. Additionally, the vertical positional accuracy of the elevation datasets used in the study was analyzed, and these results were utilized to support the interpretation of horizontal accuracy performance in orthorectification. Accordingly, the Göktürk-1 L4B DEM produced using one GCP achieved the best vertical accuracy with a value of ±1.62 m (LE90), followed by TREx (±3.30 m), Göktürk-1 L4A DEM produced without GCPs (±3.31 m), ALOS (±8.15 m), and DTED-2 (±8.32 m). In the horizontal positional accuracy analysis of the Göktürk-1 orthorectified images produced without GCPs (L3A), the image generated using Göktürk-1 L4B DEM achieved the highest accuracy with a CE90 value of 8.33 m, followed by TREx (8.43 m), ALOS (8.68 m), and DTED-2 (8.69 m). For the orthorectified images produced with GCPs (L3B), the highest accuracy was again obtained using the Göktürk-1 L4B DEM (CE90 = 1.89 m), followed by TREx (1.96 m), ALOS (2.04 m), and DTED-2 (2.31 m). The results revealed that low-resolution elevation datasets (DTED-2 and ALOS) tend to cause accuracy degradation, particularly in mountainous and highly variable topographic areas, while high resolution DEMs (TREx and Göktürk-1 L4B) provided more consistent results. Furthermore, the use of GCPs significantly improved positional accuracy for all datasets, with the improvement being more pronounced for lower-resolution DEMs. Based on the findings of this study, it is recommended that high-resolution Digital Elevation Models be used in combination with geodetically measured Ground Control Points, especially in areas with rugged topography. This approach contributes to improving both horizontal and vertical positional accuracy and ensures the production of more reliable orthorectified images under complex terrain conditions.