DATA USED
GLOSWAC is based on model wave spectra from the ERA-Interim reanalysis project (see Dee et al., 2011), from the ECMWF (European Centre for Medium-Range Weather Forecasts). ERA-Interim contains global data on a spatial resolution grid of about 100km. The time coverage is from 1979 to 2015. The wave model used is WAM.
WAVE SPECTRAL STATISTICS
Wave spectral statistics are based on statistics of wave partitions. These partitions can be characterized by any of their integral parameters. For spectral characterization we use the empirical distribution of the peak positions (frequency and direction) of the wave systems. This indicator is one of the most revealing in terms of spectral wave features (see Portilla et al., 2015 for details).
WAVE PARTITIONS
A wave partition is defined as a cluster of energy bins in the spectral domain that are self-consistent and correspond physically to a packet of energy generated by a single meteorological event. A wave partition is detected in the spectrum using the so called mountaineer scheme. In this scheme, starting from any point of the spectral domain we move following the steepest ascent path until a local peak is reached. The collection of all points associated to the same local peak constitutes a partition. A further post processing step is usually necessary to assess the significance of each partition. This is achieved by means of image processing tools (see Portilla et al., 2009).
GLOSWAC PRODUCTS
GLOSWAC products are distributed on-line in the form of jpg figures to prioritize portability. A map tool is provided for the user to select the location of interest. The following basic indicators give a comprehensive account of the local wave and wind climate.
- Probability density distribution of spectral partitions (PDS): is the joint probability of the peak position (frequency and direction) of all partitions in the series. The colorbar indicates the number of occurrences at each spectral grid point.
- Statistically defined wave systems: similarly as in the wave spectrum, we can identify clusters in the PDS. In this case we associate these clusters to long-term (statistically defined) wave systems. The partitioning algorithm again allows us to determine their spectral domains. They are indicated with a contour line and a corresponding number.
- Significant wave height distributions per wave system: each long-term wave system has its own spectral time series and therefore its own series of any integral parameter. Hs distributions for each system are given in the compact form of box-whiskers plots. In these plots the red line indicates the median value, the box limits correspond to the 25th and 75th percentiles, the whisker distance is equal to 1.5 times the interquartile range, and the red crosses point to extreme values.
- Hs monthly variability: in order to obtain an idea of the seasonal variability of each wave system, the monthly mean Hs values are represented in the form of a bar plot.
- Crossing seas probability: for many applications it is useful to know not only which are the wave systems present and their magnitude, but also how often do they occur simultaneously. This information is provided in the form of a square graphical matrix (the x and y axes correspond to the crossing pair). The lower part of the diagonal indicates the joint probability of occurrence of a particular pair in percent. The upper part gives the same information in color form. The threshold value used is 1.0 m (this value is arbitrary).
- Wave age distribution: the wave age parameter is an indicator of the wind-sea or swell stage of a particular wave system. It is found as the ratio between the wave phase speed and the vector component of the wind in the wave direction. The larger the wave age, the bigger and faster the wave, the less influence it can have by the wind. Wave age is provided in the form of box-whiskers plots. Typically the value of 1.3 is used to differentiate between wind-sea and swell. However such a value is not physically meaningful.
- Wind rose: the directional distribution of the wind in different scales is given in the form of a typical wind rose.
- Wind seasonal variability: the variability of wind speed is given for the seasons DJF – MAM – JJA – SON.
THE ECMWF GAUSSIAN GRID
The WAM wave model of ECMWF is not implemented on a regular grid. Its grid is Gaussian (on a sphere). The grid contains 40698 points from which 27948 correspond to the sea. In the map tool provided, the user can select any geographical point. However, GLOSWAC will search the nearest grid point to display results (no interpolation is carried out). The coordinates of the ECMWF selected grid point are indicated in the figures.
SPECTRAL RESOLUTION
The spectral resolution of the WAM-ECMWF model is 30 frequencies, from XX to YY Hz in geometric steps of 1.1, and 24 directions, from 7.5 to 352.5 degrees.