4 
Remarks on quality issues and data processing 
To distinguish between data errors and thermal fronts, particularly in the Western North 
Atlantic where eddies and meanders are frequent, CMS "Köln Atlantic" has been equip 
ped with an automatic sea surface temperature (SST) thermometer which has proved to 
be very helpful. It is mounted inside the ship’s hull in the engine room and functions as 
a contact thermometer. The SST as measured by this thermometer is noted in the XBT 
drop tables when recorded by the operator. A brief description can be found in Sy and 
Ulrich (1990). SST for CMS "Monte Rosa" was acquired from a thermometer at the 
vessel’s seawater intake. 
The depth fall rate error was the focus over the past few years of a study by the 1GOSS 
Task Team on Quality Control of Automated Systems (TT/QCAS). Test procedures for 
the evaluation of the depth fall rate were developed and XBT versus CTD comparisons 
were carried out under controlled conditions in different ocean regimes. Recommen 
dations for a new depth formula and for the revision of XBT data in data bases were 
developed (Hanawa et al., 1994). Following the recommendations of competent WOCE, 
TOGA and IGOSS bodies, no depth fall rate corrections have been applied to the data 
presented here. No correction of the bowing problem (Bailey et al., 1989) has been 
applied either. 
On transect 148 extensive tests have been carried out to evaluate the performance of 
XBTs used on fast-running vessels. XBTs of various types were launched from the 
vessel’s stern as supplementary tests besides the regular programme. A special 20 knots 
version of the T-5 probe (now available as "Fast Deep" XBT) was developed by Sippican, 
Inc. for CMS "Köln Atlantic" and was first used on this transect. The data from 
westbound transect 148 are presented here, but detailed test results will be published 
separately. 
The acquisition of new mainframe hardware and software by the BSH in September 1991 
necessitated the development of new XBT data processing software. See Fig. 3 for an 
overview of data processing. It is a scheme for the bulk of the data. In this context, the