Each year, aerial imagery is captured for one third of Austria’s territory to produce up-to-date orthophotos. The cross-track and along-track overlap of these images enables the computation of high-resolution, three-dimensional surface data through image matching. From this data, key spatial information such as vegetation structures, vegetation heights, terrain changes, and building heights can be derived. Previously, data processing required approximately one year, the suitability of the derived elevation models for agricultural applications had not yet been assessed, and no concept for the long-term preservation of the original aerial imagery existed.
The project therefore aimed to accelerate the generation of Digital Surface Models (DSM) using a GPU-based cluster server architecture. Additionally, the results were to be optimised for applications in agriculture and forestry, and further elevation-based products – including 3D meshes, true orthophotos, and point clouds – were generated on a test basis and evaluated for their usability.
Through the deployment of GPU-supported server architecture and scalable software, processing times were significantly reduced and large-scale data could be processed efficiently. Targeted parameter optimisation enabled the creation of high-quality DSMs within an automated workflow without manual post-processing. Point clouds, meshes, and true orthophotos were successfully computed, with a focus on open-source solutions. Qualitative limitations exist primarily in shadow areas and in high mountain terrain. The products are suitable for interdisciplinary use in agriculture, spatial planning, disaster management, and forestry. Their particular strength lies in the combination of high spatial resolution with spectral and elevation information.
In addition, a nationwide long-term archiving concept was developed to sustainably preserve the approximately 4 petabytes of original aerial imagery from the Federal-State Orthophoto Cooperation. The concept is based on the 3-2-1-1-0 backup rule and employs LTO tape format with the platform-independent LTFS file system. By February 2026, ten flight years had been incorporated into the library. This marks the first nationwide long-term archiving concept for uncompressed aerial imagery in Austria, providing a solid foundation for future analyses and research.
Schedule
Project start: 01/2021
Project end: 12/2025
