Evolution of farm terraces in the upper Sacos River catchment (Alacant) and the effects on soil conservation

Abandoning farm terraces can lead to the collapse of its walls, triggering high soil losses, with the consequent deterioration of the ecosystem services that these terraces generate in Mediterranean mid-mountainous areas. For the management of these areas, a very useful tool is the quantification and detection of processes promoting the degradation of these structures.

In this work, the terraces of the upper area of the Sacos River, (0.44 km²; Tàrbena, Alacant, Spain) were characterized and the material volume lost in the breaks is quantified. For this purpose, SfM-MVS (Structure from Motion, Multiview Stereo) photogrammetry was used to create a high-resolution DEM (0.01 m) of each break. For each one, around 45-50 photographs were taken with the phone camera to create the DEM that matched the surface and height of the collapse, obtaining a 3D model with the Metashape Pro v1.7 software allowing to adjust the surface to calculate the losses. This procedure was applied in 9 of the 82 breaks, representing 10% of them at the whole study area. With this DEM and the ArcGIS software, an interpolation of the terrace heights was carried out without considering the collapse to subsequently obtain the volume lost, from the difference of both rasters. The characterisation of the walls was done by a semi-automatic process that allowed the estimation of the walls’ height and the total length (in metres) within the study area, obtaining densities and relating this to the breaks. To explain the position of the collapses within the catchment, ergo the convergence pattern of surface runoff, a Connectivity Index (CI) was applied on a 1 m resolution DEM obtained from LiDAR data (https://pnoa.ign.es/web/portal/pnoa-lidar/presentacion).

The results showed an estimated volume of runoff loss, ranging from 0.02 m³ to 0.34 m³ depending on the walls’ height. A loss of 7.28 t ha^-1 yr^-1 was estimated for all the studied sub-catchments, higher than an erosion rate of 1 t ha^-1 yr^-1, considered to be the "sustainable" limit that can occur in a system in the form of sediment loss, indicating the severity of the losses due to falling walls. In addition, the normalised CI allowed the establishment of a threshold at ca. 40% of the same CI, above which, the 96% of the breaks were located. The type of terraces is another determining factor for the location of the breaks, since 68% of them were check-dam terraces, although this type only occupies 32% of the total surface.

The estimated volume per wall collapse in the study area is 15.78 m³, estimated from the regression line of the breaks analysed in detail, obtained from the height of each wall and the material mobilised in each collapse. This is extrapolated to all the breaks detected, considering their height for this extrapolation. This amount of eroded material indicates the need for establishing urgent conservation strategies of these soils, due to their collapse is promoting a sediment cascade process in which concentrated erosion initiate the generation of new watercourses at these hillslopes, no longer functioning as terraces.