A Bayesian Hierarchical Spatio-Temporal Model for Extreme Sea-Level Prediction in Ireland.
Niamh Cahill, Maynooth University.
26th of March 2026
Abstract:
Rising sea levels increase the risk of flooding, coastal erosion, and extreme sea-level events. Coastal communities in Ireland are particularly vulnerable due to a combination of long, varied shorelines, low-lying urban areas, and exposure to both Atlantic storm systems and surges propagating from the Irish Sea. Accurate risk assessment depends on understanding the drivers of extreme sea levels, especially storm surges. A Bayesian hierarchical spatio-temporal model is developed to estimate extreme sea-level surges at both gauged and ungauged locations, drawing on tide-gauge records from Ireland and the west coast of Great Britain in the Global Extreme Sea Level Analysis (GESLA) database. Data from Great Britain are incorporated to compensate for the relatively short record lengths at most Irish tide gauges. Annual maxima of sea-level surges are modelled using the Generalised Extreme Value (GEV) distribution, incorporating both spatial and temporal dependencies. A barrier model captures complex spatial correlations along irregular coastlines.
Model evaluation shows that combining spatial and temporal components improves predictive skill. This is particularly valuable for Ireland, where short records limit site-specific analysis; the model’s ability to share information across locations enhances estimates for data-sparse areas. The analysis reveals key patterns in extreme sea-level variability and detects an upward trend in surge annual maxima. By explicitly integrating spatio-temporal dependencies, the framework offers a flexible, data-driven approach to representing extreme sea-level behaviour, supporting risk management and coastal planning in Ireland and similar coastal settings.