An integrated approach to seismic risk assessment using random forest and hierarchical analysis: Pisco, Peru

As Peru is subject to large seismic movements owing to its geographic condition, determining seismic risk levels is a priority task for designing appropriate management plans. These actions become especially relevant when analyzing Pisco, a Peruvian city which has been heavily affected by various seismic events through the years. Hence, this project aims at estimating the associated seismic risk level and its previous requirements, such as hazard and vulnerability. To this end, a hybrid approach of machine learning (i.e., Random Forest) and hierarchical analysis (i.e., the Saaty matrix) was used. Risk levels were calculated through a double-entry table that establishes the relation between hazard and vulnerability levels. Results suggest that the city of Pisco exhibits both medium (lower city areas) and high (higher city areas) hazard levels in similar proportion. In addition, the coast area is considered a very-high hazard zone. Regarding vulnerability, the central area of the city exhibits a medium vulnerability level, whereas the periphery denotes high and very-high vulnerability levels. The interrelation of these components results in overall high-risk levels, with very-high levels in some central areas of the city. Finally, the results from this research study are expected to be useful for the authorities in charge of fostering specific activities in each sector and, simultaneously, as a motivator for future studies within this field.