An exploration of the use of machine learning to predict lateral spreading

The recent availability of large amounts high-quality data from post-disaster field reconnaissance enables an exploration the use machine learning (ML) approaches to predict earthquake-induced damage. 2011 Christchurch earthquake in New Zealand caused widespread liquefaction and lateral spreading, development ML models spreading was enabled by high-resolution for displacements, ground shaking, surface subsurface features. A dataset more than 7300 spread observations a single event geologic setting were used develop classification using Random Forest approach binary problem identify occurrence multiclass amount displacement. best developed this study accurately patterns with overall accuracy 80% 70% displacement models. These show that peak acceleration, distance river, elevation, groundwater table contribute most predictions dataset, inclusion cone penetration test (CPT) features improves only prediction largest class (>1.0 m). Further research is needed are generalizable other events settings.