Evidence Hub

A Log-Gaussian Cox process framework for zone-free probabilistic seismic hazard assessment in Sumatra

Aluicius IE, Susyanto N, Suryanto W, Ertiningsih D, Adi-Kusumo F

2026

Abstract

Abstract Conventional probabilistic seismic hazard analysis divides a region into area source zones with uniform background rates and assumes that events occur independently in time. Both assumptions are difficult to justify in subduction settings, where the background rate follows smoothly varying tectonic loading and large earthquakes generate clustered aftershock sequences. We propose a zone-free alternative in which the spatial source term of the hazard integral is a fitted Log-Gaussian Cox process. Its log-intensity is modelled from two physically interpretable covariates, the inter-seismic coupling on the plate interface and the distance to the dominant crustal fault, plus a latent Gaussian random field. The triggered component is described separately by a self-exciting Epidemic-Type Aftershock Sequence (ETAS) model with an Omori–Utsu temporal kernel. Mathematically, the contribution is the composition of a covariate latent-Gaussian point-process intensity with the Cornell hazard functional, together with likelihood-based model selection and prospective-style forecast validation. We apply the framework to Sumatra, Indonesia, using a working catalogue of 5,356 events (M ≥ 4.5, 1998–2024). The covariate Cox process is preferred over a Neyman–Scott cluster baseline by the Akaike Information Criterion and the inhomogeneous L-function, and a temporal hold-out shows that the fitted spatial pattern is preserved on unseen data. The two covariates reproduce the observed hazard…

Read the paper

Related studies

Go deeper in Evidence Hub

Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.

Open this study Sign up free