Assessing small-scale degradation patterns in a heterogeneous semi-arid landscape using aerial imagery : a case study in the Sneeuberg rangelands, South Africa

Land degradation in drylands is recognised as a global environmental problem that affects ecosystem services negatively. It is mostly prominent in the form of soil degradation, which is defined as a decline in soil quality, caused by soil erosion, contamination, acidification, or salinisation amongst other processes. Despite the vast area covered by drylands and its crucial impact on the environment and human wellbeing, only limited information about the spatial extent of degradation is available. Drylands are characterised by a sparse vegetation cover that is often arranged in a regular or irregular heterogeneous pattern of vegetated and bare soil patches. This is also reflected in heterogeneous chemical, structural, and textural soil properties. Conventional landscape mapping is very time- and labour intense, and this highly dynamic spatial heterogeneity is not depictable on soil maps. Also emerging technologies, as satellite imagery, are not sufficient to detect the small-scale patterns in these heterogeneous landscapes yet. Hence, detailed spatial data on soil types or degradation features are missing. However, it is crucial to evaluate the relevance of processes affecting soil quality, soil redistribution, and biogeochemical cycling in semi-arid landscapes. To address these knowledge gaps, the following three questions frame this research:
1. Is land use-induced patchiness of soil and vegetation a prevalent feature in the Karoo landscape?
2. How can remote sensing products represent the heterogeneity of soil degradation in drylands accurately?
3. Which information from remote sensing products is needed to improve conventional field mapping in heterogeneous landscapes?
A small catchment area in the Karoo rangelands of South Africa was chosen as a representative of a semi-arid ecosystem. The widespread erosional features of various shapes and degrees, such as silted-up reservoirs, incised sediment, badland and gully formation, make this area an ideal study site, where soil redistribution through erosion and deposition has happened and is still on-going. The objectives of the thesis are the documentation of small-scale heterogeneity of a semi-arid rangeland to assess the relevance of human-induced degradation and the development of a method for mapping and quantifying degradation. Soil samples were collected throughout the study site, from areas of various degrees of soil and vegetation degradation, to identify differences in soil properties. The results confirm the patchiness of the soil properties. Soil property patterns do not coincide with the vegetation cover but show a high degree of variability within the same landscape units, especially soil nutrients and total organic carbon. Apart from soil degradation by erosion, sediments trapped behind dams can be identified as a depositional substrate and show strong potential for formation of anthropogenic soils new to rangelands. The derived GIS analysis based on high-resolution unmanned aerial vehicle (UAV) imagery and data products generated with them proved that this data can contribute to identifying soil degradation with similar accuracy to conventional field mapping. Additionally, UAV derived data was evaluated regarding its suitability to assess badlands as sediment sources. The developed workflow enabled to quantify badlands and obtain best volume estimations for deeply incised badland systems that ideally would have a low amount of vegetation cover.
The lack of globally consistent information on the extent of dryland degradation is a major source of uncertainty in assessing their current role within the global carbon cycle and projecting drylands’ future change. The results of this study demonstrate that UAVs provide a valuable tool for low-cost and rapid assessment of soil degradation, particularly in heterogeneous landscapes where manual field sampling is very time consuming and limited to subjective assessments by the surveyor. Substantial improvements in degradation mapping can be achieved, if multispectral UAV imagery is combined with topographic information. Collecting local high-resolution data is needed for the validation and the upscaling to regional or global biogeochemical models. It improves the understanding of the relevance of spatial and temporal dynamics in heterogeneous landscapes under global change.