-5- 
Evaluation of the depth fall rate error 
The accuracy of Sippican’s empirical XBT depth fall rate equation has been questioned 
in the literature over the last 10 years (e.g. Heinmiller et al., 1983). Thus, to obtain 
more insight into the data quality some laboratory and seegoing tests were carried out 
to evaluate the depth fall rate for "Deep Blue" probes. There is no other way than to 
compare XBT data against CTD data, and usually differences observed in simultaneous 
temperature readings are interpreted as errors in the depth fall rate. However, the 
depth fall rate error derived is a function of possibly two error sources; i.e. of the depth 
fall rate equation by which the time elapsed is converted into depth, and the tempera 
ture measured by the probe’s thermistor. In addition, the specific XBT unit used may 
also be a source of error. 
Consequently, the first step in our test was the calibration of those XBT thermistors 
which were to be used later for the in-situ comparison. The laboratory test should permit 
the elimination of one of the possibly two error sources. For both laboratory and in- 
situ test phases the one and same ship’s unit was used which is an identical type to that 
onboard CMS "Köln Atlantic". 
The laboratory calibration was carried out in the following way: the controller (Bathy 
Systems SA-810) was calibrated according to the manual’s guidelines. As further 
instrumentation a HP85B computer with original Bathy Systems software, a brand new 
Sippican Handlauncher (LM-3A), and two insulated water baths one of which was 
equipped with a calibrated PtlOO thermometer for the reference temperature and with 
a stirring-propeller to guarantee a homogeneous temperature distribution in the bath. 
As a first step, the probes were pre-cooled close to the reference temperature in the first 
water bath. The launch was simulated in the second water bath, and the reading of the 
PT100 temperature and the depth at the display were taken about one minute after the 
launch began. During the launch, the probe was moved up and down to guarantee a 
better water exchange through the hole of the zinc nose where the thermistor is located. 
The probe adaption, in particular that of the zinc nose with a great thermal mass, was 
necessary in order to allow stable readings. However, the up and down movements 
sometimes caused noise in the XBT record, possibly due to connecting tolerances of the