Abstract Following a large earthquake, the sparsely instrumented Southwest Pacific fails to consistently meet the UNESCO-IOC targets for operational tsunami forecasting. We present a method to improve regional distance tsunami early warning for such sparsely instrumented regions to meet 10-minute and 45-minute operational targets by utilizing novel application of seismic array processing and a hybrid approach combining array methods, W-phase inversion, and tsunami wave analysis. Our approach implements a procedure to directly delineate earthquake rupture extent in near-real time, bypassing conventional magnitude-to-rupture scaling and enabling high-resolution source estimation. We have tested these new ideas on the 19 May 2023 Loyalty Islands earthquake ( $$M_w$$ 7.7) using data from a broadband seismic monitoring array in New Zealand. The estimated seismic source correlates well with post-event finite-fault inversion, and simulated tsunami signals match deep-ocean and coastal tide-gauge observations. Notably, synthetic tsunami waveforms show over 70% coherence with observed coastal tsunami signals at key coastal sites. A simulated real-time application of these methods, compared to existing early warning systems, demonstrates that integrating array processing significantly enhances forecast accuracy and timeliness and can effectively be used to achieve forecasting targets.

Abstract Many extreme hazardous events do not occur in isolation, but rather interact with other events, as they occur simultaneously, cascadingly, or consecutively. Hazard interrelations are critical components in hazard assessments in today’s complex and interconnected world. One event may directly or indirectly affect other hazard events through various types of spatiotemporal connections within the geophysical environment. These interrelations can greatly amplify or create new emerging impacts, often in ways that are underestimated by traditional methods. Especially considering the large volume of work done within the area of multi-hazards, there is still a lack of applying these ideas in practical disaster risk management. Drawing on the state-of-the-art conceptual models and empirical insights, as well as iterative feedback from practitioners, we propose a practical classification aid for these multi-hazard interrelations. Guided by the division into type, spatial and temporal assessment of interrelations, we propose a TST (Type–Space–Time) classification showing examples for all the different connection possibilities. We present 36 combinations of hazard interrelations and provide examples of historical multi-hazard events for each of them, including a practical framework that supports preparatory measures, response strategies, and planning decisions. It aims to provide a practical approach that aids in defining methods for assessing the frequency and intensity of multi-hazard events, the dynamic exposure and vulnerability, and combining these in multi-hazard loss assessment for specific multi-hazard scenarios, as well as in risk assessment, including the probability of such scenarios occurring. We also provide conceptual models on how to assess the impact of multi-hazards for various combinations of hazard interrelations.