Calanchi badlands reconstructions and long-term change detection analysis from historical aerial and UAS image processing

Abstract

In Italy, highly dissected badlands may be found widespread (over the entire Apennine region) and are locally described as “calanchi” (singular: “calanco”). Accelerated geomorphic processes led to the rapid evolution of these erosional landunits with intense drawbacks on agriculture, infrastructure and society. Historical aerial photographs (HAP) of such badland areas exist worldwide. To reconstruct the past surface and related topographic changes offers high potential for the assessment of long-term topographic changes of many interesting landscapes. Due to technical and conceptual demands of traditional photogrammetry those photographs have been used comparable little in long-term geomorphology. Processing them with user-friendly SfM-MVS photogrammetry would provide a new perspective on the assessment of long-term erosion rates on the catchment and slope scale and the related badland evolution. Therefore, this multi-temporal approach focusses on the appliance of SfM-MVS photogrammetry in order to reconstruct the past landscape and to calculate the related long-term surface changes of a badland located in the upper Val d’Orcia (Tuscany, Italy). The analysis is based on a set of high-resolution UAV images (2016) and two sets of historical aerial images of varying quality and scale (1976 and 1994). Digital elevation models (DEMs), shaded relief raster, several morphometric parameters and DEMs of difference (DoD) (resolution: 1 m) were used to obtain overall terrain changes and to calculate sediment balances and annual surface change rates on catchment and slope scale. Change detection analysis revealed a mean surface lowering of 0.01 m*a–1 (1976–2016) on the catchment scale and 0.03 m*a–1 (1976–1994) to 0.06 m*a–1 (1994–2016) on the slope scale. Our results further show a reduced activity of fluvial erosion processes from the mid 90s’ on while mass movements dominating the topographic changes in the second half of the 40 years observation period. Erosion processes are enhanced near the tear-off edges of the lateral valleys. Deposition mainly occurs in the main valley floor which can be classified as a temporary sediment sink/store. The geomorphic development of the catchment indicates a self-stabilizing behaviour towards an equilibrium state. A reduction of the mean slope gradient in the main valley from 25° in 1976 to 14° in 2016 confirms the self-stabilizing theory. The study also revealed the potential and limitations of historical aerial images for long-term geomorphic change detection and showed that SfM-MVS algorithms operate less reliable when applied to only a few images (n=3) of a singular flightline. Nevertheless, the good agreement of our results with the long term rates obtained by other authors with different methods/techniques and different time scales proof the great potential of historical aerial photographs for 3D surface reconstructions (SfM-MVS) of semi-arid badlands when ceratin quality parameters, mainly high image overlaps and entropy, are fulfilled.