Until recently the existing world climatological databases basically extended back to 1854. Examples are the Jones land database(1) and the Comprehensive Ocean Atmosphere Data Set COADS(2,3), recently renamed ICOADS(4). Observations prior to 1854 are generally considered to be of poorer quality. One reason is that, going back further in time, there is an increasing deficiency of instrumental observations of temperature and surface air pressure; another reason is the lack of standardization of the observation practices in the pre-1854 era.
Although the data in the latest version of ICOADS (Release 2.1)5) now extend back to the early 1800’s by adding data from ship logbooks from the US Maury collection6), no serious attempts have been undertaken so far to homogenize the pre and post 1854 data. The main reason is that no standardized methods exist to convert archaic wind and weather terminology from old logbooks to present day units. There is, however, increasing consensus that daily wind reports made aboard ships represent an enormous potential for climate reconstruction of the pre-1850 era, enabling e.g. the study of low frequency climate variability. In 2000 the three year EU project CLIWOC (Climatological Database for the World's Oceans 1750 1854) started the labour-intensive task of keying the contents of pre-1854 ship logbooks from voyages over the open oceans, to make them available for the scientific community.
CLIWOC was launched to compile a world’s oceans meteorological database (back to 1750) by digitising and interpreting data from logbooks of pre 1854 oceanic voyages by English, Spanish, Dutch and French ships. Countries involved were Spain, UK, Argentina and The Netherlands. Also a ‘dictionary’ was developed that translates the archaic wind force terms from the four languages into Beaufort equivalents7). This dictionary is included in the database as a dynamic module, to facilitate updating. Apart from producing the database for the world’s oceans between 1750 and 1854, the principal objective of CLIWOC was to draw attention to the scientific potential of the climatic data in the ships’ logbooks, and to prove the value of the daily weather observations made aboard ships.
The ship’s logbook is a legal document that contains detailed descriptions of what took place on board. It was also used to document new shipping routes and to assist in safe navigation. By 1750 the use of logbooks, often in a prescribed and tabular format, was standard practice on most ships. This resulted in a heritage of many logbooks in which detailed meteorological observations are systematically recorded.
Unravelling ancient time and position information
To be useful for climatic research the date of the observation must be known, as well as the position with sufficient accuracy, say, within 200 km. To unravel this information several problems had to be overcome.
In the period 1750-1752 the date on board some of the British vessels referred to the old style Julian calendar instead of the currently used Gregorian calendar. Fortunately in most cases the weekdays were recorded as well, which allows for an unambiguous correction.
A second problem was the use of variable prime (or zero) meridians. A range of prime meridians was used by different nations. Apart from some well-known historic zero meridians, e.g. Tenerife, Cadiz and Paris, it was also common to denote the longitude with respect to the land that was sighted most recently. Hence, the prime meridian could shift several times during a voyage (Figures 1 and 2). In the project 646 different prime meridians were recognized. In about half of these cases, the meridians were documented explicitly in the logbooks and corrections were trivial.
However, a straightforward reconstruction of the zero-meridian’s position was frequently hampered by the fact that the geographical names often differed from present-day names. At times it was necessary to study old atlases and charts. Even more problematic are the many transitions in prime meridian that are not explicitly mentioned in the logbooks. These often become apparent only after plotting the ships’ positions. Fortunately, it turned out that the latitude of the break in the ship’s longitude usually provides enough information to deduce the location of the landmark upon which the new prime meridian was based.
Quantitative interpretation of the observation was the next step. This applies in particular to wind speed, being the bulk of the pre-1854 meteorological reports and almost exclusively the contents of the reports before 1800.
Wind speed recordings were usually given in descriptive terms. Some of them seem to bear a relationship with the Beaufort scale, but this scale was only proposed in the beginning of the 19th century and adopted by most seafaring nations in the 1850’s. Since logbooks from four different countries were used, the terminology problem was even four-fold. Hence, the quantitative interpretation of the descriptive expressions of the wind force in m/s – or at least the Beaufort scale – was a major challenge.
The basis of the wind force terminologies varied markedly between the countries. The English, French and Spanish expressed the wind force in ordinary terminology (e.g. strong, or light breeze). The Dutch, however, related the wind force terminology to the amount of sail used. It is obvious that this terminology bears a direct relationship with the ship’s operations and the wind force (e.g. ‘dubbelgereefde marszeilskoelte’ or ‘double reefed topsail wind’).
The CLIWOC database contains 1,606 Dutch, 1,043 English, 1,019 Spanish and 253 French wind force terms, a total of 3,911 different expressions. Fortunately only 927 of them made up 98% of all reports. It was decided to use flexible lookup-tables for the conversion, enabling changes on second thought. We compared wind force reports from ships of one or more nations during the same voyage. Analogous comparisons were carried out with observations from different ship types and sizes8). Although differences remained, in particular between the different countries9), it seems that the current translation tables7)are of sufficient quality to allow for meaningful analyses of the pre-1854 climate and its variability10).
The first public version9) of the CLIWOC database, Version 1.5, was released in early 2004. It can be accessed through a dedicated CLIWOC website (see below). An important development is its simultaneous availability through the ICOADS website in a standardized data format. Release version 1.5 is available in IMMA11) format, which is being developed under JCOMM. Release 1.5 contains 280,280 records. The database includes information on date, time and position of each report, followed by the key climatic information for wind direction and wind force. If available, instrumental observations are included as well. This is generally limited to the more recent observations. Though not of interest for climatic studies, additional information (conditions on board, reports on deaths and diseases, sightings of animals, etc.) was keyed as well.
The database is formally documented by the project’s final report, submitted in 2004 to the EU12). The database is also extensively described in articles in a Special Issue of Climatic Change on CLIWOC. Figure 3 shows that spatial coverage of the CLIWOC data and Figure 4 the annual number of keyed observations per country.