Landslide Debris-Flow Prediction using Ensemble and Non-Ensemble Machine-Learning Methods

Landslides and associated soil movements (debris-flow) are the common natural calamities in the hilly regions. In particular, Tangni in Uttrakhand state between Pipalkoti and Joshimath has experienced a number of landslides in the recent past. Prior research has used certain machine-learning (ML) algorithms to predict landslides. However, a comparison of ensemble and non-ensemble ML algorithms for debrisflow predictions has not been undertaken. In this paper, we use ensemble and non-ensemble machine-learning (ML) algorithms to predict debris-flow at the Tangni landslide. Non-ensemble algorithms (Sequential Minimal Optimization (SMO), and Autoregression) and ensemble algorithms (Random Forest, Bagging, Stacking, and Voting) involving the non-ensemble algorithms were used to predict weekly debris-flow at Tangni between 2013 and 2014. Result revealed that the ensemble algorithms (Bagging, Stacking, and Random Forest) performed better compared to non-ensemble algorithms. We highlight the implications of predicting debris- flow ahead of time in landslide-prone areas in the world