Tectonic and climatic controls on terrace preservation in a large Himalayan river catchment

Terraces are important archives of past environmental conditions, recording variations in both climate and tectonics on thousand-to-million-year timescales. For example, fill terraces in large mountain ranges may contain repeated deposits from large scale debris flow or landslide events, or may deform in response to gradients in tectonic uplift rate. However, our knowledge of the geographical extent of Himalayan terraces is currently very limited. This hinders our understanding of the spatial and temporal patterns of extreme hazard events such as large landslides, major earthquakes, and glacial lake outburst floods. Our limited understanding of Himalayan terraces may be a consequence of low preservation potential due to erosional processes within a rapidly uplifting mountain range. Even a comprehensive assessment of terraces in an area may not provide a complete archive of depositional processes as terraces can be destroyed or modified. Alternatively, terraces may be present, but difficult to recognise in the field or to manually identify from aerial photographs or satellite imagery. An automatic method for identifying river floodplains and terraces has recently been developed (Clubb et al, 2017). Using this method, we identify terraces at a catchment scale for the first time within the Gandaki catchment of central western Nepal using the 12m TanDEM-X digital elevation model. We explore the spatial pattern of terraces along the long profiles of each major river within the catchment by calculating the total terrace area adjacent to the channel. We then attempt to link terrace preservation to tectonic drivers by analysing the relationship between terrace exposures and channel steepness, knickpoints and major structural boundaries along the river profile. Coupling an analysis of spatial patterns in terrace preservation with the shape of terrace profiles downstream compared with the modern channel allows for investigation into whether terrace preservation is controlled by long term tectonic forces or stochastic high magnitude flooding events. We find that terrace preservation within the Gandaki catchment is largely tectonically controlled, with terraces mostly preserved directly upstream of major tectonic structures such as the Main Frontal Thrust, the Main Boundary Thrust and within the Thakkhola-Mustang Graben. However, we link a pattern of preserved terraces directly south of the Main Central Thrust to the stochastic occurrence of high magnitude debris flow events. These highly elevated terraces decrease in downstream elevation compared with the modern channel, are unconstrained by downstream tectonic structures and source from the steep topography of the High Himalaya. Our work demonstrates the potential of automated terrace extraction techniques for understanding controls on sediment storage and dynamics across actively uplifting mountain ranges.