摘要：River valleys have been observed on Titan at all latitudes by the Cassini‐Huygens mission. Just like water on Earth, liquid methane carves into the substrate to form a complex network of rivers, particularly stunning in the images acquired near the equator by the Huygens probe. To better understand the processes at work that form these landscapes, one needs an accurate digital terrain model (DTM) of this region. The first and to date the only existing DTM of the Huygens landing site was produced by the U.S. Geological Survey (USGS) from high‐resolution images acquired by the DISR (Descent Imager/Spectral Radiometer) cameras on board the Huygens probe and using the SOCET SET photogrammetric software. However, this DTM displays inconsistencies, primarily due to nonoptimal viewing geometries and to the poor quality of the original data, unsuitable for photogrammetric reconstruction. We investigate a new approach, benefiting from a recent reprocessing of the DISR images correcting both the radiometric and geometric distortions. For the DTM reconstruction, we use MicMac, a photogrammetry software based on automatic open‐source shape‐from‐motion algorithms. To overcome challenges such as data quality and image complexity (unusual geometric configuration), we developed a specific pipeline that we detailed and documented in this article. In particular, we take advantage of geomorphic considerations to assess ambiguity on the internal calibration and the global orientation of the stereo model. Besides the novelty in this approach, the resulting DTM obtained offers the best spatial sampling of Titan's surface available and a significant improvement over the previous results.
