Using remote sensing to quantify sediment budget components in a large tropical river - Mitchell River, Gulf of Carpentaria

Abstract

The recent drought in southern Australia and the increasing realisation that this may be the norm as a result of global warming has led to increased interest in developing the land and water resource of the tropical savannah in northern Australia. Intensive land and water resource development in tropical savannahs in other parts of the world has been shown to lead to significant land degradation, and there are signs this has already occurred with the comparatively moderate levels of historic grazing and agriculture in northern Australia. It is critical that we understand contemporary and historical landscape dynamics before land use is intensified in this region. As a way of beginning to quantify key sediment sources and to start to understand potential changes in relative contributions associated with land use and other drivers, we have begun to construct a sediment budget for the Mitchell River, a 71 360 km2 catchment draining into the Gulf of Carpentaria. We present some preliminary results from two key components of the sediment budget: alluvial gully erosion and channel erosion. A remote sensing analysis has identified around 130 km2 of active alluvial gullies within the Mitchell megafan, which has an estimated active front length of around 5560 km. Preliminary measurements suggest the average rate of scarp retreat to be 0.34 m per year, with scarp heights ranging between 0.3 and 8 m. Using a spatially distributed method we calculated an annual catchment sediment yield from alluvial gullying of >5 Mt/year. Estimates of minimum river channel bed turnover were also determined using a remote sensing approach from Landsat archival data. A measure of minimum annual bed turnover of 15 Mm3/year was derived from analysis of bed area change between 1988 and 2005 (i.e. from either pools to sand bars and vice versa, or vegetated island to pools and vice versa). When analysed at a tributary scale, these data show a pattern of net sediment scour from the Walsh and Palmer rivers, and net accumulation within all other mainstem channels in the period between 1988 and 2005. These data support the anecdotal evidence that there has been a trend towards sediment accumulation within the lower Mitchell River in recent times, with a resulting net reduction in total pool area of 0.6 km2. While preliminary, these data highlight the value of remote sensing for constructing sediment budgets in large dynamic tropical rivers.