Multivariate rule-based seismicity map of Iran: a data-driven model

Abstract

The seismic hazard map or delineation of regions with high earthquake hazard is important to plan risk mitigation strategies. Identifying areas of high seismic hazard can lead city planners to enforce better construction standards and predict areas vulnerable to slope instability. Conventional seismic hazard maps are based on limited factors like ground acceleration, ground velocity, etc. This paper presents a new class of data-driven multivariate rule–based model to create online as well as offline interactive seismic hazard map that is flexible and readily automated. A multivariate rule-based seismicity map (MRBSM) is defined as the map of regions with a future high hazard of earthquakes. The classification and regression tree method is used to extract rules that predict regions with high hazard of earthquakes with mb ≥ 4.5 in Iran. The rules generated for our MRBSM of Iran are based on a large number of geological and geophysical parameters. The MRBSM indicates that the province of Bandar Abbas, a major population center in the South of Iran has a high hazard of earthquakes with mb ≥ 4.5. In addition, our method allows identification of the most important parameters associated with earthquakes. Our analysis shows that the isostatic anomaly has the strongest correlation with earthquakes while magnetic intensity, regional Bouger anomaly, Bouger anomaly, and gravity anomaly also correlate well. Despite widespread application of a- and b-values of the Gutenberg-Richter formula, these parameters do not correlate well with earthquake hazards in the area.