Sediment output from a post-mining catchment – Centennial impacts using stochastically generated rainfall

Abstract

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 100y 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.