Compilation of digital elevation model for Turkey in 3-arc-second resolution by using SRTM data supported with local elevation data

In February 2000, the Space Shuttle Endeavour of the “Shuttle Radar Topography Mission” (SRTM) launched by NASA collected elevation data by scanning the earth landmasses between ~60º north and south parallels. After this 11 days mission, the collected data were processed, and a Digital Elevation Model (DEM) at 1-arc-second – and also 3-arc-second – resolution was created. SRTM DEM contains data voids because of several disturbing effects. The proportion of data voids in Turkey (0.17% for second corrected version) is approximately same as the global proportion (0.15%). The data voids are the most important obstacles for efficient use of the data. DEMs with high resolution are important data sources for all disciplines in geosciences. Since collecting such data and creating DEM are both time consuming and expensive processes, many scientists have been studied on filling these voids in a reliable way. The aim of this paper is to present the results of a project, which aims to validate the SRTM within Turkish territory, and to create a DEM based on the SRTM data covering the same area without data voids. The validation of SRTM was performed using following ground truth data sets: topographic maps (25K maps) and DGPS tracks. The executed statistical analysis for accuracy assessment of SRTM within Turkish territory gives the results consistent with those of global analyses performed by several researchers. The voids are visually classified as the clustering and the scattered voids. The clustering voids were filled with a local data source. For this purpose the 1:25000 scaled national topographic map set (25K maps) was used. The scattered voids were filled with interpolation. As a result of the project, which was completed in September 2008, a complete DEM of Turkey at 3-arc-second resolution has been created and published via Internet for the use of researchers and professionals in geosciences.