World Wide Wave Statistics (WWWS)
From Ocean Wave Climate
The basic WorldWaves database consists of wave model time series for 9,665 positions calibrated against Topex and Jason data. In addition to the data for these 9,665 locations, Fugro OCEANOR holds uncalibrated data in some 26,000 positions. These latter grid points can be calibrated on demand for clients. We may then use additional satellites other than Topex and Jason. This can then give even higher quality data in coastal but deep water regions with large spatial gradients in wave conditions. This benefit is less in the open ocean areas which we are most concerned with in what we call our WorldWideWaveStatistics (WWWS) database.
Those of us who have been around for some years will remember the old voluminous Global Wave Statistics (GWS) book compiled by Neil Hogben and associates from, primarily, visual observations from the voluntary observing fleet. WWWS is a modern-day equivalent based on the highest quality WorldWaves data and providing ready-to-go wave statistics for 150 sea areas globally in a similar format to GWS. It was originally developed on the initiative of a leading Japanese shipping company. It will primarlily be of interest to other shipping companies and classification societies requiring modern reliable global wave data.
The provision of statistics for all grid points globally would be prohibitively expensive and voluminous and discussion with the shipping industry lead to the world oceans being partitioned into 150 sub-areas for the compilation of the WWWS statistics. The partition is based on the original areas from the traditional Global Wave Statistics as many shipping companies had used this in the past and a similar partitioning fitted well to existing in-house software. Some of the original areas exhibiting a large degree of variability have, however, been split into smaller areas to give better resolution. In addition, some areas not included in the original version, due to lack of observations, have been added. The final partitioning is shown in Figure 1.
The simplest way of selecting what WorldWaves data to use for each area would have been to use all the existing grid points within each area. However, this is not the best approach, because the WorldWaves data positions are not evenly distributed over the world oceans (although the original ECMWF model data did have a regular 0.5º grid everywhere). Closest to the coast, the full resolution of 0.5 had been retained by us. Further offshore the resolution had been reduced to 1.5, and in the open ocean far from coastlines the WorldWaves resolution is 3. There were two main reasons for this choice of variable resolution:
1. The demand for data close to the coast is in general larger than for data in the open ocean.
2. The variability and spatial gradients are much larger close to the coast.
If we had used all the calibrated data within an area containing coast lines, the estimated sea states would then have been biased towards conditions representative of the seas nearer to the coast, and not necessarily representative of the spatial average over the area (as seen, for example, by an ocean-going vessel). Since the wave heights in general tend to be lower nearer the coast, such a procedure would also tend to underestimate the wave heights estimated for the area (in areas containing coastlines). We have therefore chosen a different approach, where we first selected all the data points lying on either a 1.5 or a 3 grid, depending on the size of the area. As this may give a too low density in some of the smaller areas, we have then supplemented the first selection with available positions to give a fairly uniform and representative coverage in each area. The final selection of points is shown for the whole world in Figure 3 (The number of points shown is 3868). A close-up view for North America is shown in Figure 4. Note that positions lying exactly on the border between areas are used in the statistics for both the neighbouring areas.
In order that the computed statistics should be representative of an area, the wave climate should also be as uniform as possible within the area. As a measure of variability of wave climate within each area we have computed a normalized standard deviation of wave height for each area. The standard deviation is based on the annual mean wave height in each of the data points within an area. That is, for one area there are as many values as the number of data points in the area. An area mean wave height is estimated as the average of the mean values in each point. The normalized standard deviation is computed as the standard deviation of the annual means divided by the area mean value. This latter statistic has then been used to decide whether a region needs to be further split with the aim to end up with a minimum of reasonably homogeneous regions covering the global oceans.
The statistics are based on the standard 10-year WorldWaves time series for the period 1997 – 2006, with a temporal resolution of 6 hours. The following wave parameters and criteria are used:
- Hm0 Significant wave height
- Mdir Mean wave direction (incoming), clockwise from north
- Tm-10 Mean period Tm-10, or energy period
- Month Used to determine season
The statistics provided consist of bivariate tables between Hm0 and Tm-10. The resolution of wave height is 1m, from 0 to 17m. Values above 17 m are grouped into one class. The resolution of wave period is 1s, from 3s to 16s. Values below 3s and values above 16s are considered as two additional classes.
The statistics are provided both as annual statistics (all data) and as seasonal statistics. Four 3-month seasons have been defined as follows (with the same definition used for all areas):
- March to May
- June to August
- September to November
- December to February
The statistics are further divided according to wave direction, i.e. both as omnidirectional statistics (all directions) and statistics in eight 45-degree sectors. The sectors are defined as centred on the following directions:
- North West
- South West
- South East
- North East
The frequencies in the bivariate tables are given with a resolution of parts per thousand (ppt) with two decimals. The data are delivered as text files -- one for each of the 150 areas or, alternatively, a web-based geographic interface is provided for users to access the statistics on-line.