A hybrid method of debris flow velocity estimation based on empirical equation

Abstract

Debris flow is a multi-phase liquid flow with high bulk density and complex components. The velocity of a debris flow is one key part of debris flow dynamics, which is essential to know in order to formulate a plan for early hazard warnings and mitigation. The normal methods of calculating debris flow velocity have their limits due to the difficulty in precisely determining parameters, resulting in unsatisfactory predictions. In this article, we propose a new hybrid prediction model to estimate the debris flow velocity, which is based on the empirical equation of flow velocity calculation combined with a least squares support vector machine (LSSVM) and Particle swarm optimization (PSO). The new method can improve the performance of velocity empirical equations. The result has shown that the new hybrid model is better than the former method and it exhibits high potential for becoming a useful tool for identifying debris flow velocity.