Home > Research > Publications & Outputs > Spatial and seasonal variability of metocean de...

Electronic data

  • AnkEA19OMAE

    Accepted author manuscript, 2.95 MB, PDF document


Text available via DOI:

View graph of relations

Spatial and seasonal variability of metocean design criteria in the Southern South China Sea from covariate extreme value analysis

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paperpeer-review



This paper describes spatial and seasonal variability of metocean design criteria in the southern South China Sea. Non-stationary extreme value analysis was performed using the CEVA approach (Covariate Extreme Value Analysis,[1]) for a 59-year long SEAFINE hindcast of winds and waves, estimating metocean design criteria up to 10,000-year return period. Wind design criteria are mostly driven by large-scale monsoonal events; at higher return periods infrequent cyclonic events have strong influence on the tail of the extreme value distribution but confined to a limited geographical area. The CEVA analysis of waves showed much less dependence on the tropical cyclone events; the spatial metocean design criteria were smoother, mostly influenced by the monsoonal wind strength, fetch and local bathymetry. Return value estimates illustrate the strong seasonality of metocean design criteria, with boreal winter (December-February, Northeasterly monsoon) contributing most to the extremes, while April and May are the mildest months. Estimates for the ratio of 10,000/100-year return values are also presented, both for winds and waves. There is empirical evidence that the range of “typical” values of generalised Pareto shape parameter observed for Hs is different to that observed for wind speed. For this reason, an upper bound of +0.2 for generalised Pareto shape was specified for wind speed analysis, compared to 0.0 for Hs. In some cases, increase of upper bound for waves to 0.1 is justified, leading to slightly more conservative Hs values. We confirmed that the upper end point constraint was not too influential on the distributions of generalised Pareto shape parameter estimated. Nevertheless, it is apparent that specification of bounds for generalised Pareto shape is a critical, but problematic choice in metocean applications.