A catchment scale evaluation of the SIBERIA and CAESAR landscape evolution models

摘要

Landscape evolution models provide a way to determine erosion rates and landscape stability over times scales from tens to thousands of years. The SIBERIA and CAESAR landscape evolution models both have the capability to simulate catchment-wide erosion and deposition over these time scales. They are both cellular, operate over a digital elevation model of the landscape, and represent fluvial and slope processes. However, they were initially developed to solve research questions at different time and space scales and subsequently the perspective, detail and process representation vary considerably between the models. Notably, CAESAR simulates individual events with a greater emphasis on fluvial processes whereas SIBERIA averages erosion rates across annual time scales. This paper describes how both models are applied to Tin Camp Creek, Northern Territory, Australia, where soil erosion rates have been closely monitored over the last 10 years. Results simulating 10 000 years of erosion are similar, yet also pick up subtle differences that indicate the relative strengths and weaknesses of the two models. The results from both the SIBERIA and CAESAR models compare well with independent field data determined for the site over different time scales. Representative hillslope cross-sections are very similar between the models. Geomorphologically there was little difference between the modelled catchments after 1000 years but significant differences were revealed at longer simulation times. Importantly, both models show that they are sensitive to input parameters and that hydrology and erosion parameter derivation has long-term implications for sediment transport prediction. Therefore selection of input parameters is critical. This study also provides a good example of how different models may be better suited to different applications or research questions.