Assessment of ground-motion prediction equations using observations and information theory: application to the Northeastern Tibetan Plateau

Abstract Ground-motion prediction equations (GMPEs) are fundamental to seismic hazard assessment, yet their applicability is difficult to evaluate in tectonically complex regions where instrumental observations are sparse or absent, such as the northeastern Tibetan Plateau. We evaluate representative GMPEs using two approaches: residual-based error metrics and an information-theory-based evaluation. The two recent earthquakes, the 2022 Menyuan and 2023 Jishishan events, provide an opportunity to examine model behavior under observed conditions. The information-theory-based approach is first applied to these two earthquakes to verify feasibility and to compare with residual-based outcomes, and is then extended to two large historical earthquakes in the region, the 1920 Haiyuan earthquake and the 1927 Gulang earthquake, to assess stability under extrapolative, data-limited conditions. Across these applications, the information-theory-based evaluation yields consistent rankings and stable weights, whereas the residual analysis captures event-specific variability. Together, the two approaches provide complementary insight into model applicability and offer a practical basis for ground-motion assessment and extrapolation in tectonically complex and observationally sparse regions.