GUIDE WAVE ANALYSIS AND FORECASTING - WMO
GUIDE WAVE ANALYSIS AND FORECASTING - WMO
GUIDE WAVE ANALYSIS AND FORECASTING - WMO
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
114<br />
quality control and re-analysis of historical information<br />
necessary to produce a high-quality database (particularly<br />
of the winds needed to drive the hindcast wave<br />
model).<br />
One approach, which has been adopted in order to<br />
minimize the costs associated with continuous hindcasts,<br />
is to archive the analysis portion of operational wave<br />
analysis and forecast programmes. This is a very costeffective<br />
means of producing a continuous database,<br />
being a by-product of an existing operational programme.<br />
The disadvantages are that the operational time<br />
constraints mean that not all available data are included<br />
in the analysis, that time-consuming techniques such as<br />
kinematic analysis cannot be performed, that the use of<br />
backward as well as forward continuity in the development<br />
of weather patterns is not possible, and that it will<br />
take N years of operation to produce an N-year database.<br />
There is also a danger that such archives may suggest<br />
“climate change” that is not real but is a result of<br />
changes in the characteristics of the models used.<br />
Nevertheless, this approach represents a viable way to<br />
develop a continuous database of wave information,<br />
albeit of lesser quality. Attention also needs to be given<br />
to the temporal homogeneity of the model data, especially<br />
when different procedures to estimate ocean winds<br />
are used at different times.<br />
9.6.2.2 Storm hindcasts<br />
In order to perform the extremal analysis necessary for<br />
establishing design criteria for offshore operations, it is<br />
necessary to have wave information for a period of at<br />
least 20 years. In fact, recent experience suggests that<br />
even 20 years may not be sufficient to produce stable<br />
estimates of long return period wave heights. The cost<br />
associated with producing a continuous hindcast for<br />
more than 20 years is prohibitive in most instances. As a<br />
result, an approach has been adopted in many countries<br />
whereby a selection of the top-ranked wave-producing<br />
storms over a period of 30 or more years is hindcast,<br />
with the wave heights analysed using peak-overthreshold<br />
techniques (see Section 9.4.1.2).<br />
This approach has several advantages, not least of<br />
which is that the cost is a fraction of that for a continuous<br />
hindcast. Storms hindcast are typically of about five<br />
days duration, so for a sample of 50 storms in a period of<br />
30 years, for example, the total hindcast period would be<br />
250 days. Another major advantage is that the storms<br />
can be hindcast in considerably more detail, with full reanalysis<br />
of each storm, including kinematic analysis, and<br />
forward and backward continuity. All available data can<br />
be used in the analysis, including those data abstracted<br />
from ships’ logs as well as that originally obtained from<br />
the Global Telecommunication System (GTS). There is<br />
no doubt that this approach produces the highest quality<br />
data of any hindcast procedure for the storm periods<br />
selected.<br />
On the negative side, the database produced, while<br />
suitable for extremal analysis, is inappropriate for any<br />
<strong>GUIDE</strong> TO <strong>WAVE</strong> <strong>ANALYSIS</strong> <strong>AND</strong> <strong>FORECASTING</strong><br />
analysis which requires a time series of data at one or<br />
more locations (such as persistence analysis). It also<br />
cannot give any information on frequency distributions<br />
of waves, since only the extreme conditions are analysed.<br />
One additional drawback is that one cannot<br />
remove lingering doubts that the storms selected for the<br />
detailed hindcasts are the most severe wave-producing<br />
storms, since these must usually be selected in the<br />
absence of direct wave measurements, by proxy criteria<br />
such as the time history of pressure gradients over wavegenerating<br />
areas.<br />
9.6.2.3 Hybrid hindcasts<br />
Hybrid hindcasts are being used increasingly to try to<br />
combine the best features of the continuous and storm<br />
hindcasts. They start from a continuous hindcast produced<br />
as described above, and are then augmented by<br />
hindcasts of the most severe storms and of periods where<br />
the verification against measurements shows that the<br />
continuous hindcast has produced significant errors. The<br />
periods from the storm hindcast then replace the archive<br />
in the continuous hindcast. This approach has been used<br />
effectively in the Gulf of Mexico and off the west coast<br />
of Canada. It is relatively cost effective, and continued<br />
correction to the continuous hindcast can be made<br />
depending on available resources. The database<br />
produced is suitable for all types of statistical analysis,<br />
including extremal analysis if the hindcast period is<br />
sufficiently long.<br />
9.6.3 Hindcast procedure<br />
The following paragraphs describe a wave hindcast<br />
storm procedure. Most aspects of the hindcast procedure,<br />
other than storm selection, are the same whether the<br />
continuous or storm approach is selected. If the hindcast<br />
is produced using the analysis cycle of an operational<br />
wave forecast programme, most of the decisions on<br />
model domain, input wind fields, etc. will have been<br />
made; the only remaining decision will involve the<br />
archiving process. A detailed description of the steps<br />
involved in a storm hindcast are given in the <strong>WMO</strong><br />
Guide to the applications of marine climatology (<strong>WMO</strong>,<br />
1994(b)).<br />
The application of the hindcast method includes the<br />
following main steps:<br />
(1) Given the point or area where hindcast wave data<br />
are required, decide the area necessary to be<br />
included to represent the wave conditions in the<br />
region of interest (to catch distantly generated<br />
swells, etc.).<br />
(2) Select the time span for the hindcast; previous<br />
experience with the historical marine meteorological<br />
databases supports selection of storms from<br />
about the past 30 years. The database for earlier<br />
periods is much less extensive and wind fields may<br />
not be specified as accurately. Therefore, the historical<br />
period which should generally considered<br />
extends from about the mid-1950s to the present.