Ekstremværditeori og stormflodsrisiko i Esbjerg: Statistisk modellering og økonomisk vurdering

Abstract

This thesis applies Extreme Value Theory (EVT) to assess storm surge risk in Esbjerg, Denmark, with a particular focus on the Peaks Over Threshold (POT) method and the Generalized Pareto Distribution (GPD). Using daily maximum sea level data from 2006–2024, a threshold of 165 cm DVR90 was selected based on diagnostic tools, and GPD parameters were estimated using Maximum Likelihood Estimation(MLE) after declustering dependent events. A stationary model estimates the 100-year return level at approximately 499 cm, but a seasonally stratified analysis reveals substantial variability: the 100-year return level reaches 606 cm in winter and only 414 cm in summer. This seasonal divergence highlights the inadequacy of stationary models in dynamic climate contexts and demonstrates the value of non-stationary modeling. To quantify economic impacts, the estimated seasonal return levels are linked toa flood damage function calibrated to local conditions. It results in the Expected Annual Damage (EAD) that is estimated at DKK 8.47 million for the winter season alone. Assuming a 75% damage reduction from adaptation measures, the discounted net present value (NPV) of avoided losses over 100 years exceeds DKK 196 million, indicating strong economic justification for investment. The findings underscore the importance of incorporating temporal variability and local risk modeling in climate adaptation planning. By combining EVT with economic assessment, the study provides a robust decision-support framework for managing coastal flood risk in a changing climate.