The European Database

The EUROPE database is in most aspects similar to the worldwide database, it is derived from visual observations of wind and wave heights reported by merchant ships of the Voluntary Observing Fleet. The data has been quality-controlled and quality-enhanced by the well-validated NMIMET analysis method.

The EUROPE database provides ocean wind and wave climate data for 31 sea areas on and around the northern European continental shelf. The sea areas stretch from the Bay of Biscay in the south up to the Norwegian Sea in the north. Whilst these areas of ocean are covered by the standard worldwide database, the EUROPE database, with its much smaller sea areas, provides greater geographical resolution. [Map 2]


Map 2 - The 31 Sea Areas of the European Database

The choice of the size of the sea areas has inevitably been a compromise between geographical resolution and statistical reliability of the data. Very small sea areas would provide good geographical resolution, but the number of wind and wave observations in each area would be small, and the statistical reliability of the climate low.

As far as possible the sea areas have been chosen to be reasonably homogeneous in wind and wave climate terms.

Data for each of the 31 sea areas is split up into 5 seasonal classes (the 4 seasons of the year, plus the whole year) and 9 directional classes (the 8 points of the compass, plus all directions). Thus the database was compiled from 1395 datasets.

Where there is insufficient data to arrive at a reliable wave climate in a particular dataset, or where the quality control checks in the NMIMET analysis have identified questionable results, the wave data is omitted from the database.

The user will find some datasets contain a specific warning message about an aspect of the data which appears to BMT to be questionable or unreliable. These casts are often the result of excessive scatter or 'wild points' in the input data, but we have nevertheless decided to include the data in the database because other aspects of the data appear to be valuable.

The NMIMET analysis used to generate the EUROPE database is the same as that used to product the GLOBAL database with the exception that the method of obtaining the modelling coefficients has been changed somewhat. Continuous refinement of the NMIMET analysis method has led to some improvements in the way these coefficients, which relate the mean wave height and the standard deviation of the wave height distribution to the wind speed distribution are obtained. See Appendix_A3 for definition of these coefficients.

Note that Area 31 (The Baltic) was included in the GLOBAL database but has been re-analysed with these minor changes to be consistent with the remainder of the EUROPE database, and the user may therefore notice some slight differences when compared with Area 5 in the worldwide: database.

In undertaking revision of the analysis, primary importance was attached to validation. In developing the original GLOBAL database confidence in reliability was established by extensive comparisons with measured data in widely separated areas of the world. The region covered by the EUROPE database is particularly well endowed with instrumental data especially in areas where there has been intensive oil and gas exploration and production activity. It was natural therefore to compare the climate statistics, and particularly the extreme return period wave heights with these instrumental measurements. Unfortunately, whilst much instrumental data exists, and some has been published, there are even greater quantities that have yet to be published in the open literature.

However, one of the key references for extreme wave heights is the UK Department of Energy Guidance Notes [49], which provides contours of 50 year return period significant wave height compiled using data from 41 instrumental stations coupled with visual observations of wind speed. One of the supporting documents to the guidance notes, [50] gives more detail of the instrumental measurements used to generate the wave height contours around the UK continental shelf.

Figure 29 - Comparison of 50 Year (3 hour) Storm Significant Wave Heights with contours from reference [49]. Figure 30 - Comparison of 50 Year Storm Significant Wave Heights with Instrumental Values from reference [50].

The 50-year significant wave heights predicted by the EUROPE database (annual, all Directions) have been compared with the Reference [49] contours in Figure 29. This comparison demonstrates good agreement in some areas, and much poorer agreement in others. The crosses indicate maximum and minimum values of the range of 50 year wave contours in [49] spanned by each sea area, whilst the line indicates the values given for the particular sea area in the EUROPE database. Examination of these results indicated a general trend for agreement to be best in the deeper water and more exposed locations.

A further comparison with the individual instrumental stations listed in [50], and the appropriate EUROPE sea area in Figure 30 shows the ratio of the extreme wave height values plotted as a function of water depth and demonstrates a clear trend with water depth. (Instrumental stations were rejected for this analysis if they were less than 30 miles from the shore as it was considered that they would be affected too much by coastal topography to be comparable with the open ocean visual observations.) It can he seen that the 50 year wave height values agree quite well for water depths greater than 200m, but in shallower depths there is a growing tendency for the EUROPE database to over-predict the value.

When these discrepancies were first discovered a rigorous checking procedure was instigated to look for errors in the original raw data and the NMIMET analysis procedure used to process it. However, no fault could be found and, at the time of writing, the differences remain unexplained.

It is possible, however, to advance some potential hypotheses.

  1. A bias towards more severe visual observations in shallower waters due to exclusion from the NMIMET analysis of all observations not reporting both a sea and a swell wave height and a greater tendency for swell to be unreported in milder conditions (instead of being correctly reported as zero swell height).

  2. Tendency for ships' wave height observations to be over-estimates in shallower water due to greater steepness of the waves.

  3. Tendency for the ship observations to be biased towards the deeper water in a given sea area, whilst many of the instrumental data sets in shallow water are very near to the coastline and sheltered from at least some directions. It may be relevant to note here the experience reported in [50] gained from the use of wind speed data to assist the compilation of the extreme wave height contours in [49]. This showed a clear trend for increasing over-prediction with decreasing water depth leading to the introduction of an empirical correction formula.

Each of the above has some plausibility, but none appears likely to provide he whole answer. It could of course be a combination of these and other effects

The estimation of extreme wave heights is an area fraught with numerous difficulties and users are always urged to compare their extreme wave estimates with other data, if it is available. In the case of such EUROPE estimates in water depths less than 200m, users could decide to adjust the result for the water depth effects apparent in Figure 30.