Wave spectra often exhibit several peaks due to the coexistence of wind sea generated by local wind conditions and swells originating from distant weather systems. For an accurate description of sea states, those spectra can be partitioned and the partitions can be interpreted physically as representing independent wave systems. Furthermore, partitions can be associated in time to track the evolutions of wave systems originating from the same meteorological event (storm, depression, etc).
This paper proposes a new method to identify temporal sequences of wave systems parameters, consistent with respect to the meteorological events that are the sources of the phenomena. This method, as many conventional partitioning ones, is based on the watershed algorithm which is however directly applied in our case to the whole time-history of wave spectra. Using appropriate criteria, the identified events are classified into swell or wind sea events. Field data from West Africa are used to illustrate the method and the results are compared with those of conventional software. Hindcast data from the Iroise Sea are also used to validate the proposed method and the results show a good identification of wave systems events with a good correlation between wind sea events and wind characteristics.