Search

Computer based landscape evolution models can provide insight into both erosion rates and processes (i.e. sheetwash, rill, gully erosion). One important data requirement of these models is long term, high quality, high-temporal resolution rainfall data (given that the physical nature of the erosion process is strongly related to rainfall). However, in many cases such data is limited - data is often short, incomplete or not of a sufficient temporal resolution. Therefore, the aim of this study was to test the sensitivity of modelled erosion rates to small changes in rainfall input. To achieve this we firstly assess the existing rainfall data from an established weather station and secondly, stochastically generate rainfall time series based on the longest and most reliable rainfall data. We then test the sensitivity of different rainfall sequences on sediment output using a well-tested landscape evolution and sediment transport model (CAESAR-Lisflood) over a simulated period of 100 y on a proposed rehabilitated mine landform. It was found that each rainfall scenario produces a unique pattern of erosion (i.e. the location and extent of the gullies is variable). Further, each rainfall scenario produces a unique pattern of sediment output that suggests non-linear processes. Importantly, this is the first time stochastically generated rainfall has been employed in landform evolution modeling and provides a statistical approach to quantify sediment transport and landform evolution. The method demonstrates a risk based approach and allows rainfall, runoff and sediment transport studies to be conducted in data poor environments. The findings clearly demonstrate that rainfall variability can greatly affect sediment transport and form of erosion as well as landscape evolution. This information is of particular importance for the design and testing of rehabilitated landscape systems such as post-mining landscapes. [ABSTRACT FROM AUTHOR]

An important part of planning for the rehabilitation of a mine site is the design of stable final landforms for waste rock dumps or spoil piles. Whilst able to be assessed over the short-term (years to decades), the longer term behaviour (centuries to millennia) of such landscapes is not within any meaningful human time frame of observation. Predictive numerical models, therefore, form an important tool with which current landscape behaviour and longer term trajectory can be assessed. However, an important issue associated with the use of models is the ability to assess the reliability and accuracy of the model. Here the SIBERIA and CAESAR-Lisflood Landscape Evolution Models (LEMs) are used to simulate and assess the geomorphic stability of a conceptual rehabilitated landform of the Ranger uranium mine in the Northern Territory, Australia, for a simulated period of up to 1000 years. Utilising both models in this study enabled an independent assessment of likely landscape processes and evolution as well as each model. Results show that SIBERIA and CAESAR-Lisflood produce erosion rates and patterns that are broadly similar. At millennial time scales, short-term processes such as gullying appear to be the dominant erosion features in the proposed landforms and may produce substantial erosion in terms of size and amount of hillslope material eroded and transported downslope. Vegetation was found to have a major effect on the erosion potential of the landform surface. Overall both models produce very similar results. [ABSTRACT FROM AUTHOR]