Complex fjord topography (bathymetry and coastline) may differentiate significantly wave conditions not only compared with the offshore ones but in the vicinity of few tens of meters. In the present work, possible inhomogeneities of wave conditions are investigated in a hypothetical bridge crossing in the area of Sulafjorden, central Norway. More specifically, wave conditions at ten positions across the bridge crossing have been derived by means of numerical modelling. The analysis has been carried out by transferring o shore wave conditions to the nearshore area by successive applications of the well–known third–generation wave model SWAN. As input, a very detailed bathymetry of the area, and time series of wind and wave parameters, derived from ERA5 database, have been used. At the target points, long–term time series of directional wave spectra have been used as input for the assessment of the inhomogeneity hypothesis. Various statistical features have been examined including, among others, the seasonal variability, the probability structure, the directionality, the correlation structure, and the long–term wave spectra.
The study of wave conditions in an area is essential for a number of nearshore applications, such as coastal structures, marine transport, fish farming and renewable energy. Last years, the Norwegian Public Roads Administration is planning a number of large fjord crossings with different type of bridges, and the knowledge of local wave conditions is instrumental in both the design and the operational phase.
The most reliable source of information for the local wave conditions should be long–term in situ measurements of wave parameters. However, measurement campaigns are expensive, time consuming (since they are performed in real–time), and refer to some specific points. There is no possibility to cover large areas with buoy instruments.
On the other hand, numerical models can provide us with equally good datasets of wave parameters with some extra advantages: good spatial coverage, no gaps, reduced cost (in comparison with the measurements), easy way to update datasets (e.g., via reanalysis). Third generation spectral models (TheWAMDI Group, 1988; Tolman, 1991; Booij et al., 1999) are based on a statistical representation of waves using two–dimensional (frequency–direction) wave spectra, and they are also known as phaseaveraged models.
