This study investigates the hydro-morphological processes driving the channel shift of the Karnali River in Western Nepal after the 2009 monsoon season. Exceptionally high discharge during this year caused significant morphological changes to the bifurcation system, leading to a drastic discharge reduction in the branch that flows into and provides habitat for the Bardiya National Park. The park is home to various endangered species, such as the Royal Bengal Tiger, and the change in flow distribution is impacting ecosystem dynamics and increasing human-wildlife conflict. An extensive field campaign, combined with satellite imagery analysis of planform evolution,  provided the foundation for a conceptual model. Two hypotheses are tested: (1) bend sorting and flood-induced aggradation in the Geruwa branch, and (2) erosion-driven opening of the Kauriala branch. These hypotheses are tested using a schematized version of the bifurcation system in a depth-averaged 2D hydro-morphological numerical model (Delft3D). 

Model results suggest a self-reinforcing feedback loop initiated by increased sediment load during peak flows in the Geruwa branch, leading to aggradation and reduced flow capacity. This process promoted further sedimentation and the eventual partial closure of the branch. The model highlights the critical role of mixed-sized sediment interactions, particularly the processes that influence the local sediment supply of finer grain sizes, such as hiding and exposure. Additionally, the processes that influence the direction of sediment particles, namely secondary flow and transverse bed slope gravity pull, significantly impact the discharge distribution. The size, magnitude, and direction of particles towards a particular branch can lead to exceeding its transport capacity and subsequent aggradation, resulting in the dominance of the other branch. Furthermore, the model reveals limited sensitivity to slope changes in the Kauriala branch. Although limited by the lack of calibration and validation data, the results from this schematized model offer valuable insights into the drivers behind the real-world system and may inform future studies and potential interventions.