Spatial Cross‐Correlation Models for Next‐Generation Amplitude and Cumulative Intensity Measures

Spatial correlation of ground motion intensity measures (IMs) plays a critical role in seismic hazard and risk assessment, particularly for understanding how IMs vary across sites during an earthquake event. While conventional IMs such as , and have been widely studied, there is a growing interest in so‐called next‐generation IMs, such as and , due to their increased accuracy in characterising seismic vulnerability. However, to date, no spatial correlation models exist for these next‐generation IMs, nor for their cross‐spatial correlations with traditional IMs. This study proposes such a model for within‐event residuals among five IM types, including both conventional and next‐generation ones. Principal component analysis (PCA) and geo‐statistical methods were used to develop the model based on 8484 ground motion records from the NGA‐West2 and ESM databases. Two modelling strategies were explored: a global PCA formulation that captures the joint spatial behaviour of all IMs and a pairwise‐PCA approach. The pairwise strategy improves the representation of semivariograms and cross‐semivariograms for specific IM combinations. However, it is primarily intended for pairwise analysis, while the global model provides a consistent framework for joint simulation of multiple IMs. The proposed model shows good agreement with the existing spatial correlation models for traditional IMs. The study found that clustering datasets by moment magnitude, , did not significantly impact spatial correlations. In contrast, clustering based on site conditions, specifically (i.e., soft soil vs. hard rock), notably influenced the correlation structure and was hence considered. Furthermore, illustrative applications indicate that neglecting cross‐IM correlation can lead to underestimation of joint exceedance probabilities. Given the current lack of spatial correlation models for next‐generation IMs, this study fills a critical gap by providing tools that support their integration into more accurate regional seismic hazard and risk modelling frameworks.