Erosion and weathering forensics of a catastrophic glacial lake outburst flood in Nepal

Glacier Lake Outburst Floods GLOFs are well known for their destructive powers and far-reaching consequences. These events are typically very difficult to predict, have a short duration, and often destroy installations in- and close to the river, making observational insights into GLOFs very rare. This limits our process understanding of such events. Here, we will present new observational data of the 2016 Bhotekoshi-Sunkoshi GLOF in Nepal. This data provides new insights on the erosion and weathering processes linked to violent floods and their aftermath.

In the night of the 5^th of July 2016, a GLOF roared down the Bhotekoshi-Sunkoshi, originating from China, and causing severe damage along the river for roughly 50 km. This flood was recorded by an array of seismometers, allowing us to reconstruct the hydrodynamics of this event. Furthermore, 3 hydrological monitoring and sampling stations along the river captured the event. These stations cover the entire river length from Barabise, located in the affected upstream reach, to Kurkot situated halfway down, to Chatara, located at the outlet to the Gangetic foreland. For all 3 stations, we present daily resolved river discharge, suspended sediment flux and major element geochemistry for a month before and after the event. The dataset is complemented by mineralogical analysis of the suspended sediments and trace element analysis of the dissolved load.

We observed a clear enhanced suspended sediment signal that is carried through the entire river system. Sediment concentrations dilute from >30g/l at Barabise to ~ 7g/l at Chatara. At Barabise, sediment concentration decays in an exponential manner over roughly 14 days, mimicking the seismically deduced bedload activity. At the same time, we observed a clear peak in total dissolved chemistry of 50-70 mg/l above background. However, not all major elements contribute equally to this peak, with K⁺ and HCO₃^- being the major ions of this signal. Surprisingly, F^- also has a well sustained peak, tripling the background concentration from ~0.07 to 0.21mg/l and mimicking the suspended load. This F^- peak is present all the way down to Chatara, despite an almost 2 order of magnitude increase in discharge. We identified muscovite as potential source where F^- can replace OH^- in the crystal lattice. This is supported by the mineralogical and trace element analysis, e.g. Rb⁺ replacing K⁺ in muscovite. We propose that the very high energy sediment transport during the flood, together with a complete reorganization of the river bed, caused violent abrasion, leading to mechano-chemical dissolution of weakly bound F^- from the fresh muscovite surfaces. In order to test our hypothesis, we were able to reproduce the signal in a circular flume.

This very fast evolution of F^-, an element generally little considered in weathering processes, highlights the role of mechanical rock grinding and crushing in chemical dissolution processes in rivers. Furthermore, we show for the first time the application of F^- as a tracer for catastrophic floods that can project well beyond the actual flood affected area and provides valuable insights on the chemical processes of such extreme events.