                       CTD CALIBRATION: CRUISE POSEIDON 31                 
                                                                           
(1) Instrumentation and Data Acquisition                                   
Data  were  collected  using  a   Neil  Brown  Instrument  Systems  MK  III
Conductivity, Temperature and Depth (CTD) probe.  The data were archived on
magnetic  tape  using  a Nova computer which also calculated  salinity  and
density (sigma t).  Plots of profiles and yo-yo sections were also drawn by
the  computer but those presented here have been subjected to an  extensive
correction  and  editing  scheme  as the parameters  derived  in  real-time
suffered from instrumental effects. The sampling program also recorded data
from the shipboard meteorological instruments once every two minutes.      
                                                                           
The  NBIS CTD delivers a data cycle every 32 ms,  and the sampling  program
stored  a  cycle generally every 96 ms.  A program change on  03  September
(246) enabled every second data cycle to be stored, and this version of the
program  was  used during the Second Box.  Data cycles were lost  when  the
computer  was  occupied  writing data blocks to magnetic tape  (0.288s  for
blocks  to  MT0,  0.416s for MT1) and when  the  meteorological  data  were
sampled (0.192s when MT0 was on-line, 0.288s for MT1).                     
                                                                           
A  new  conductivity  cell was fitted at the start of the  cruise,  on  the
manufacturer's  recommendation,  and a new thermistor was installed  on  26
August  (238)  so the temperature data from the First Box were  taken  with
only the platinum resistance thermometer.                                  
                                                                           
The CTD was lowered with a nominal rate of 0.5 m/s,  which gave a  sampling
interval  of  about 5 cm or 3 cm depending on the program speed.  Only  the
descending parts of the profiles (including yo-yos) have been processed.   
                                                                           
Throughout the periods of data acquisition,  navigation fixes were taken at
half-hourly  intervals and at the start of individual stations of  the  box
surveys.  Generally the fixes were made using the Loran-C system but,  when
reception  was  poor,  combinations of good Loran- C  and  Decca  Navigator
channels  were found to produce consistent readings.  Radar  ranging  using
moorings  and  other  ships  were frequently made,  especially  during  the
multiship  experiments but,  while they endorse the positions derived  from
the  routine  navigation procedure,  they have not been used to derive  the
navigation information used in this report.                                
                                                                           
                                                                           
(2) Calibration and Accuracy                                               
The  precision  of  the NBIS CTD is ultimately limited  by  the  digitising
interval  in  each  of  the three channels.  These limits  are  0.0005K  in
temperature,  0.001  ms/cm in conductivity and 0.05 dbar in  pressure.  The
electronics  noise is stated to be smaller than these values (Brown  1974).
Comparison between derived salinity and water bottle samples has shown that
a precision of +/- 0.0024 ppt over a range of salinity values is  possible.
In  practice,  these  limits are rarely reached and the accuracies  of  the
measurements are determined by other factors,  such as the accuracy of  the
calibration  procedure,  calibration drift,  contamination of  the  sensors
(especially  the conductivity cell) and,  in the presence of gradients,  by
the speed of response of the sensors.                                      
                                                                           
The CTD used from F.S.  Poseidon during JASIN was loaned by the  University
of Liverpool, UK, and had been calibrated at the Institute of Oceanographic
Sciences (Wormley), UK. The temperature channel was found to be accurate to
+/- 0.001K.  This pre-cruise temperature calibration,  being an equilibrium
calibration,  is  of  the platinum resistance thermometer  and  is  equally
applicable  to  temperature data gathered with or  without  the  additional
thermistor.  The post cruise calibration in Kiel showed a small  correction
of less than 0.005K in the temperature range met in the JASIN area. As this
is not larger than the quoted accuracy of the reference thermometers  which
were  available  at  the time,  the pre-cruise temperature  calibration  is
believed  to have been applicable throughout JASIN.  This is in  accordance
with  the manufacturer's claim for the stability of the temperature channel
of  +/- 0.001K per month.  Comparison with sea-surface  temperatures  taken
with  a  bucket  thermometer  at the start and  end  of  selected  profiles
revealed no appreciable bias.                                              
                                                                           
The  conductivity cell and its matched interface card were replaced at  the
start of the cruise. The value of the cell factor given by the manufacturer
is  0.99993  +/-  0.00001.  Comparison  with  sea-surface  salinity  values
determined  with a Guildline `Autosal' thermostatic salinometer  gave  cell
factors  in the range 0.99948 to 1.00048.  The variation with time was  not
systematic.  In processing the data,  a constant cell factor has been  used
for data from profiles forming a specific part of the cruise (e.g. for each
of the Boxes, or for groups of yo-yo stations), which are shown in Table 5.
The  cell  factors  all  lie within about +/- 0.0005  of  unity,  and  this
deviation  produces changes in the derived salinity of about +/- 0.020  ppt
in  the range encountered in JASIN.  Comparison with the surface salinities
gives a scatter in the corrected derived salinities of 0.0073 ppt (standard
deviation of 100 samples).                                                 
                                                                           
In  addition  to  this limit on the accuracy of  the  salinity  value,  the
`spiking'  effect  in the regions of rapidly changing temperature  gradient
produces a further local degradation of the salinity data.  This effect has
largely  been removed in the data processing by a time constant  correction
to  the  temperature  signal  and a  salinity  spike  rejection  criterion.
However,  it  is  likely  that some of the small  closed  contours  in  the
salinity sections are caused by bad data which were not correctable.  Thus,
the  accuracy  of the derived salinity could be estimated at +/- 0.010  ppt
with  possible  local  degradation to say +/- 0.030 ppt  in  parts  of  the
thermocline.                                                               
                                                                           
The  temperature coefficient of the strain gauge pressure cell is  believed
to  be  one  of  the  main sources of error in  the  pressure  signal.  The
coefficient  is  given  as 0.3 dbar/K.  The  characteristics  of  the  heat
transfer  from  the water to the cell are not known,  but even if the  full
range  of  the  temperature change reached the cell,  an  error  of  a  few
decibars  at most would have been caused.  As only data from the descending
profiles  have been used,  the relative accuracy in the pressure signal  is
believed  to  be  better  than  +/- 1 dbar.  The  effect  of  the  pressure
measurement error on the derived salinity is negligible: a 1 dbar error  in
pressure would introduce a salinity error of less than 0.0005 ppt.         
                                                                           
The effects on the derived density (sigma t) of a measurement error of  +/-
0.005K  in  temperature and an error of +/- 0.010 ppt in salinity  are  +/-
0.001 and +/- 0.008 respectively.  Thus, in the region of a badly corrected
salinity  spike  a  sigma t error of say 0.025 might  be  experienced,  but
elsewhere the accuracy of the derived sigma t values is about +/- 0.010.   
                                                                           
A selection of T-S curves drawn from the processed data were compared  with
similar  diagrams of data from other ships when the stations were close by.
Comparison  with  the  processed data of ATLANTIS  II,  DISCOVERY,  METEOR,
PLANET  and TYDEMAN,  failed to reveal any discrepancies which could not be
attributed to oceanic variability.                                         