River dynamics in the Himalayan foreland basin

Rivers sourced in the Himalayan mountains support more than 10% of the global population, where the majority of these people live downstream of the mountain front on the alluvial Indo-Gangetic Plain. Many of these rivers however, are also the source of devastating floods. The tendency of these rivers to flood is directly related to their large-scale morphology. In general, rivers that drain the east Indo-Gangetic Plain have channels that are perched at a higher elevation relative to their floodplain, leading to more frequent channel avulsion and flooding. In contrast, those further west have channels that are incised into the floodplain and are historically less prone to flooding. Understanding the controls on these contrasting river forms is fundamental to determining the sensitivity of these systems to projected climate change and the growing water resource demands across the Plain. This thesis examines controls on river morphology across the central portion of the Indo-Gangetic Plain drained by the Ganga River (the Ganga Plain). Specifically, the relative roles of basin subsidence, sediment grain size and sediment flux have been explored in the context of large-scale alluvial river morphology over a range of timescales. Furthermore, this thesis has developed and tested techniques that can be utilised to help quantify these variables at catchment-wide scales. This analysis has been achieved through combining new sediment grain size, pebble lithology and cosmogenic radionuclide data with quantitative topographic and sedimentological analysis of the Ganga Plain. In the first part of this thesis, I examine the contrast in channel morphology between the east and west Ganga Plain. Using topographic analysis, basin subsidence rates and sediment grain size data, I propose that higher subsidence rates in the east Ganga Plain are responsible for a deeper basin, with perched low-gradient rivers systems that are relatively insensitive to climatically driven changes in base-level. In contrast, lower basin subsidence rates in the west are associated with a shallower basin with entrenched river systems that are capable of recording climatically induced lowering of river base-level during the Holocene. Through an analysis of fan geometry, sediment grain size and lithology, I then demonstrate that gravel flux from rivers draining the central Himalaya with contributing areas spanning three orders of magnitude is approximately constant. I show that the abrasion of gravel during fluvial transport can explain this observation, where gravel sourced from more than 100 km upstream is converted into sand by the time it reaches the Plain. I attribute the over-representation of quartzitic pebble lithologies in the Plain (relative to the proportion of the upstream catchment area likely to contribute quartzite pebbles) to the selective abrasion of weaker lithologies during transport in the mountainous catchment. This process places an upper limit on the amount of coarse sediment exported into the Indo-Gangetic Plain. Finally, I consider the use of cosmogenic 10Be derived erosion rates as a method to generate sediment flux estimates over timescales of 102-104 years. Cosmogenic radionuclide samples from modern channel and independently dated Holocene terrace and flood deposits in the Ganga River reveal a degree of natural variability in 10Be concentrations close to the mountain front. This is explored using a numerical analysis of processes which are likely to drive variability in catchment-averaged 10Be concentrations. I propose that the observed variability is explained by the nature of stochastic inputs of sediment (e.g. the dominant erosional process, surface production rates, depth of landsliding, degree of mixing), and secondly, by the evacuation timescales of individual sediment deposits which buffers their impact on catchment-averaged concentrations. In landscapes dominated by high topographic relief, spatially variable climate and multiple geomorphic process domains, the use of 10Be concentrations to generate sediment flux estimates may not be truly representative. The analysis presented here suggests that comparable mean catchment-averaged 10Be concentrations can be derived through different erosional processes. For a given 10Be concentration, volumetric sediment flux estimates may therefore differ