Gully erosion is a severe land degradation process, primarily impacting land resources on-site and water resources off-site. When active in a catchment, it can be the dominant driver of soil loss, causing significant environmental and socio-economic consequences. However, other soil erosion mechanisms remain at the forefront of research, which contributed to our inability to assess gully erosion on a catchment to regional scale. The current capability to model gully erosion on larger geographic extents remains limited due to the complexity of interactions of control factors and various sub-processes driving gully expansion. In this study, an approach to apply local case studies to inform on regional gully severity is introduced to address modelling shortcomings, and an initial scaled framework is provided, which could be implemented for future regional scale investigations and monitoring. South Africa has a long history of erosion problems and has been considered an area with high gully incidence. The “hotspot” perception, coupled with the diverse climatic and geo-environmental attributes exhibited in South Africa, motivated the use as the focal region for this study. Local case study sites were used to extract physiographic properties and gully severity to produce a susceptibility map for South Africa. Additional local sites were selected across the E-W climate gradient of South Africa to assess gully severity and to isolate climate and land use controls of gully erosion to provide clues on how environmental change may influence future gully erosion. The findings from the susceptibility map, which used secondary data from the literature, converged with the findings from primary data derived from sites located across the climate gradient of South Africa. Gully erosion severity increases eastwards towards the Grassland biome, in which gullying is most severe. Here, gully erosion resulted in soil losses of up to 17 t ha-1 y-1, which exceeds the baseline limit (27 times more) and is almost twice the sustainable limit calculated for South Africawhen the upper thresholds for both these limits are used. Perceptions from landowners/ -users/ and -managers mostly align with gully concerns from the field sites, showing that their appraisals are concurrent with local gully severity. Remediation efforts are ongoing at several sites; however, measures focus on gully headcuts and do not consider vegetation establishment. Vegetation is considered critical, especially for long-term success rates of mitigation, and could be a reason for the lack of successful mitigation. The poor success rate is also disconcerting, as climate change will likely exacerbate gully erosion in South Africa. Although climate change is predicted to increase gully erosion due to larger storm magnitudes, the data presented here indicates that rainfall intensity is likely to play a secondary role in exacerbating gully erosion. Rainfall variability may be the principal driver of gully erosion. If climate change increases the frequency of El Niño Southern Oscillation events, gully erosion severity may increase and even reactivate previously stabilised gullies due to more intense rainfalls after periodic droughts. Continuous assessment and monitoring of gully extents are crucial to assessing where gullies are of concern and whether there is a change in severity. Manually digitising gullies or solely relying on fieldwork will not sufficiently address a need for monitoring via temporal data. Semi-automated detection strategies which are scaleable and transferrable would enable the extraction of gully dimensions unbiasedly and would allow to quantitively assess gully expansion (or contraction) by subtracting polygon- or raster-based output. A semi-automated approach that uses gully morphology to extract gully dimensions is developed and tested with datasets from South Africa, Namibia, Spain, and Australia. Initial assessment shows positive results, accurately predicting > 75.4% of the gullied area when scaling between small gullies (planimetric area of 1619 m2) to large gullies (planimetric area of 70246 m2). Regarding transferability to benchmark areas where other land uses were practised and where different spatial resolution data were used as input, the variance between 1.4% and 14.8% was determined, with producer accuracies above 84.5% and 70.6%. The semi-automated method has some shortcomings, with the requirement for manually digitising gully headcuts being the most pertinent. As a framework, regional assessments and monitoring should implement a scaled approach. The initial step should produce a susceptibility map using key variables associated with gullying. Following that, more computationally intensive detection strategies could be implemented, constrained to areas of most concern defined by susceptibility. Lastly, representative field sites can be identified from the detected gullies, where primary data can be retrieved to quantify gully processes, severity, and implications. Continued work is required to refine this framework, for example, refining semi-automated approaches to increase accuracy and increasing localised field sites in different geo-environments to improve trend analysis and better our understanding of how various controls interact to steer gully evolution. Lastly, this new information should yield data that can be used to build and calibrate models; such gully evolution modelling currently needs to be improved and is pivotal to further our understanding of how gully networks will react to climate and land-use changes.
Date of Award | Mar 2024 |
---|
Original language | English |
---|
Awarding Institution | - Coventry University
- Stellenbosch University
|
---|
Supervisor | Marco Van De Wiel (Supervisor), Jana Fried (Supervisor) & Willem P. de Clercq (Supervisor) |
---|
- gully erosion
- soil loss
- susceptibility
- automated detection
- climate change
- environmental change
- El Niño
- South Africa
Gully dynamics evolution under environment change pressures
Olivier, G. (Author). Mar 2024
Student thesis: Doctoral Thesis › Doctor of Philosophy