Landslide susceptibility and monsoon preparedness in Nepal: An engineering perspective
By Kshitij Dahal, Rocky Talchabhadel, and Bhesh Raj Thapa
Just within about 20 days of June this year, 20 people died with 24 people missing, and 193 families affected by landslides and floods across the country. Landslides are major disasters in Nepal and now landslide dam outburst flood (LDOF) is also challenging. Every monsoon, we embrace a chain of slope failures. They often cause cascading effects such as landslide dam formation, the collapse of the dam, flash flood, and sediment washout and aggregation. For example, the Seti river flood killed more than 70 people in 2012 while the Kali Gandaki river buried 27 buildings in the Baseri landslide in 2015.
The recent Melamchi case on June 15, 2021, is also a landslide-dam outburst flood (LDOF) scenario. Such a case highlights the dire need for preparedness across Nepal. Though the precipitation extremity from the lens of hydrometeorology is not extraordinarily large in these particular weeks, a compound interaction with sediment hazards turns these landslides into a devastating disaster.
Therefore, we need to be prepared for a proper early warning system that encompasses the probability of sediment and landmass movement. It is high time to analyse cascading multi-hazards. We find rainfall extremes are the promoter or carrier of such multi-hazards. However, the assessment of the rainfall threshold for multi-hazards in the Himalayan region is still challenging.
Rainfall-triggered landslides, if predicted ahead of time, could provide crucial information for monsoon preparedness. Here, we quickly predict the probability of rainfall-triggered landslides happening in Nepal. We use a “machine learning” model – where the machine learns from the landslides that happened in various locations and tries to find a common pattern – and using that common pattern, the model tries to predict the probability of future landslide occurrence in these locations.