Abstract
Motivated by recent work on Markov extremal models, we develop a nonstationary extension and use it to characterize the time evolution of extreme sea state significant wave height (HS) and storm direction in the vicinity of the storm peak sea state. The approach first requires transformation of HS from a physical to a standard Laplace scale achieved using a nonstationary directional marginal extreme value model. The evolution of Laplace-scale HS is subsequently characterized using a Markov extremal model and that of the rate of change of storm direction described by an autoregressive model, the evolution variance of which is HS-dependent. Simulations on the physical scale under the estimated model give realistic realizations of storm trajectories consistent with historical data for storm trajectories at a northern North Sea location.
Original language | English |
---|---|
Article number | e2541 |
Pages (from-to) | 1-14 |
Number of pages | 14 |
Journal | Environmetrics |
Volume | 30 (2019) |
Issue number | 1 |
DOIs | |
Publication status | Published - 2018 |
Keywords
- extreme
- Markov extremal model
- nonstationary
- significant wave height
- storm trajectory