Accuracy of high-resolution photogrammetric measurements of gullies with contrasting morphology

Field techniques allow accurate direct measurements on gully geometry, even in three-dimensional (3D) coordinates. Accuracy and detail reproduction is limited mainly by experimental setup and density of measurements, and less so by the precision of the measuring equipment. In contrast, remote-sensing techniques permit the coverage of large study areas with a minimum of time and effort. However, the indirect measurements from imagery are known to depend on factors like image resolution, quality of ground control, vegetation cover and image evaluation technique, which strongly influence the measurement accuracy. The objective of the present study was to investigate to what extent the accuracy of 3D gully measurement using photogrammetric techniques depends on gully morphology. At a study site in the Bardenas Reales (Navarre, Spain), field measurements of cross-sections were taken for five gullies with contrasting morphology and dimensions and used as reference data for analysing the errors associated with a corresponding dataset obtained using small-format aerial photogrammetry whose pixel size on the ground is 16 mm. Results show that volumetric gully measurements by means of photogrammetric techniques are strongly affected by the gully morphology; in particular by its width/depth (W/D) ratio, because of the increasing sun-shadowing and sight-shadowing effects associated with narrower gullies. Only wide, shallow gullies are little affected by this problem. For gullies of an intermediate typology (W/D between 0·5 and 2·5), the accuracy of photogrammetric measurements will much depend on the time of day and the period of the year when the photographs are taken, and narrow/deep gullies (W/D