ABSTRACT
Landslides are rather common in Himalayan country Nepal. Almost every monsoon the country suffers significant losses and fatalities due to floods and landslides. A landslide, which occurred on May 24,2015, had some specific characteristics and aftermath. The landslide occurred during dry period, and in effect, there was no trigger other than the fact that it occurred a month after the ‘great’ earthquake, which was very destructive. Furthermore, the landslide formed a natural dam in Kali Gandaki River (the length was approximately 700 m and the height was about 30m), and formed a lake of about 2 km length and more than 30 m deep. Particularly, downstream towns Galeshwor Bazar and Beni Bazar were in threat of inundation in case the breach occurred. The town is located in alluvial deposits, and very vulnerable to extreme flows, propagating from upstream. The landslide dam breached relatively quickly (about 15 hours later), so fortunately there was no fatality due to the propagating flood flow.
In this study, an attempt has been made to reproduce first the flood propagation using 2D Delft-Flexible Mesh (developed at Deltares). Furthermore, a sediment transport and morphological study has been carried out using Delft3D morphological model. The Delft3D model, used herein, is depth-averaged with uniform and non-uniform sediment transport and morphology. The model incorporates all kinds of innovative, recently developed aspects, amongst which domain decomposition, consideration of floodplains including wet and dry processes, sediment transport over non-erodible layers and functionality for sediment management.
The inflow hydrograph was deduced from the BREACH − a model that computes the flow hydrograph, induced by the breach of an earthen dam (in our case a natural landslide dam). The model can be used to compute the breach erosion and sediment transport as well. Since data are sparse in these remote areas, landslide geometry was estimated from available information and Landsat images. Available photos of the landslide allowed gross estimation of geotechnical parameters. Reservoir surface area as a function of stage was determined using SRTM topography data.
The hydrograph, resulting from the BREACH model, was used as upstream condition to simulate the downstream flow propagation. Topography, used in each model, had been processed with hydrological smoothing and incorporated a 1 m deep and 45 m wide channel carved into the thalweg. Hydrologic smoothing and carving of the topography is found to be important, since simulations with a raw topography appear to be causing hindcast flood waves with too little depth, too diffusive, and altered wave propagation speed. High-frequency results were extracted at the locations of the Beni Bridge and Maldhunga Bridge, for comparison between models and with videos recorded by residents. Figure 1 shows the model domain, used for the numerical computations. The computational model domain. https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781315623207/4fbc492d-6678-4a12-aaf6-5c2b8ea38e5f/content/fig106_1.tif"/>
2D morphological model will be used to replicate the sediment transport and downstream morphological changes in a relative manner. Also, some synthetic cases will be simulated in order to check some hypotheses and assess their applicability to address such issues and demonstrate the usability of computational models as a part of nonstructural measures in disaster prevention and management.
The work is still under progress.
