Variable selection based on entropic criterion and its application to the debris-flow triggering

TL;DR

This paper introduces the minimum entropy analysis (MEA), a new quantitative method for variable selection in complex geophysical systems, demonstrated on debris-flow data to identify key triggering factors.

Contribution

The paper presents a novel entropy-based variable selection method (MEA) and applies it to debris-flow data, successfully identifying relevant physical variables affecting debris-flow triggering.

Findings

Identified hydrological form factor, landslide numbers, and areas as key variables.

Validated MEA as effective for geophysical data analysis.

Demonstrated application on debris-flow data from Taiwan.

Abstract

We propose a new data analyzing scheme, the method of minimum entropy analysis (MEA), in this paper. New MEA provides a quantitative criterion to select relevant variables for modeling the physical system interested. Such method can be easily extended to various geophysical/geological data analysis, where many relevant or irrelevant available measurements may obscure the understanding of the highly complicated physical system like the triggering of debris-flows. After demonstrating and testing the MEA method, we apply this method to a dataset of debris-flow occurrences in Taiwan and successfully find out three relevant variables, i.e. the hydrological form factor, numbers and areas of landslides, to the triggering of observed debris-flow events due to the 1996 Typhoon Herb.