Modelling multi-decadal sediment delivery to rivers by debris flows and lahars with SedCas

Debris flows and lahars convey large quantities of sediment through fluvial systems in mountainous and volcanic regions. Constraining decadal to centennial patterns of sediment transport by these potentially destructive flows is crucial for understanding their drivers and subsequently modelling the evolution of downstream hazard and channel morphology with time. Relatively few modelling frameworks have been designed to capture sediment transport dynamics at these timescales. Existing models instead tend to either 1) simulate the runout of individual debris flow events or 2) forecast landscape evolution over longer millennial timescales. Our work seeks to address this research gap by developing SedCas. SedCas is a spatially lumped sediment cascade model developed to simulate decadal patterns of sediment transport by debris flows from the Illgraben, an Alpine catchment in Switzerland, into the Rhône River. Its relatively simple structure is computationally inexpensive and has enabled its use in forecasting debris flow hazard and sediment yield from the Illgraben over the 21^st century in response to a range of climate change scenarios. Here, we present the first application and adaptation of the SedCas model framework to non-alpine catchments. Firstly, we simulate sediment transport by lahars in a catchment on the island of Montserrat which has been disturbed episodically by explosive volcanism between 1995 - present. In this model iteration, SedCas_Volcano, we account for variations in vegetation cover induced by eruptive events, in addition to water and sediment supply. The model results capture the first-order patterns (aggregate magnitude-frequency) of the largest observed lahars, and the timing and relative order of magnitude of fluctuations in sediment yield. Seasonal and interannual variations in lahar activity are not fully captured, however. We attribute these shortfalls to limitations of available data and the model not accounting for important dynamic hydrological processes that alter runoff generation on evolving volcanic deposits. These limitations in turn provide avenues of further research and development. Secondly, we present preliminary experiments to simulate bedload sediment delivery as input into a new global flood model that accounts for evolving channel geometry.