                  CTD CALIBRATION: DISCOVERY CRUISE 132                    
                                                                           
1) Data Collection                                                         
Two Neil Brown CTD profilers were used. These were referred to as `shallow'
and  `deep'  CTDs  as  they  used  1500  and  6000  dbar  pressure  sensors
respectively.  The  shallow  CTD was only used in lowered mode  on  Station
10620  reported  here.  The CTD casts (Table 1 and Fig 1) fall  into  three
groups):                                                                   
                                                                           
- Casts 10621, 24,  25,  48 and 49 were primarily calibration casts,  taken
before  or after SeaSoar runs and used to cross-calibrate the shallow  CTD.
This  was necessary because bottle samples and thermometer readings  cannot
be taken during SeaSoar towing.                                            
- Casts 10626-38 and 10650 comprise a north-south section along 20deg W +/-
1deg from 47.5deg N to 40.5deg N.                                          
-  Casts 10639-47 comprise an east-west section along 41.5 - 42deg  N  from
20deg W to 12deg W.                                                        
                                                                           
                                                                           
2) Calibration                                                             
CTD data were logged in the first instance onto a PDP11/34 computer,  After
sampling (CTDSAMP) and averaging to 1-second raw values (CTDAVE),  the data
were calibrated (CTDCAL) using approximate calibration constants (Table  2)
and archived to magnetic tape.  The values output by CTDCAL are subscripted
CAL  in  Table  2.  Every fifth calibrated value was listed so  that  final
calibration constants could be determined as follows.                      
                                                                           
Down  casts  were  generally completed without  stopping.  All  calibration
samples  were  taken  with  a General  Oceanics  Rosette  Sampler  stopping
typically  at  200  m intervals during the upcasts.  Some  bottles  carried
reversing  thermometers.  The  number of calibration values  for  salinity,
temperature  and  oxygen on each cast are shown in Table  1.  Pressure  was
generally  checked with a single unprotected/protected thermometer pair  at
the deepest calibration depth.                                             
                                                                           
For the shallow CTD, only one calibration cast (10620) was available.      
                                                                           
The  changes  made to the calibrated values (CAL) to give  final  corrected
(CORR) values are shown in Table 2,  and discussed individually below.  The
statistics of the CORRected values are given in Table 3.                   
                                                                           
a) Pressure                                                                
The  additive  constant C2 is determined from the deck offset,  or  surface
value. It was altered from 12 dbar (C1*C2) to 5 dbar after cast 10621.     
                                                                           
The default calibration for C1 (0.1 dbar/count,  6553 dbar full scale)  was
used  throughout.  If reversing thermometers were used as standard,  the 26
sample  values obtained would suggest that CTD pressures should  have  been
corrected by -2.6+/-2.5 dbar (Table 3 and Fig 2). No correction was made.  
                                                                           
The  shallow  CTD had a deck offset of just over 1 dbar,  and  the  default
calibration (0.025 dbar/count, 1638 dbar full scale) appeared to be correct
(Table 3).                                                                 
                                                                           
b) Temperature                                                             
The  laboratory  calibration of the deep CTD  (TCAL  deg  C=0.035+0.0005005
T(RAW)  was used throughout the cruise.  The raw CTD values were speeded up
with a 0.25 second time constant to match the conductivity time constant.  
                                                                           
Analysis  of 32 reversing thermometer calibrations suggested that  the  CTD
was reading too low by 16.9+/-8.4 mK (Table 2, C7; Table 3 and Fig 3). Both
CTD  and  reversing thermometer laboratory calibrations had  been  referred
back  to absolute standards,  but it was thought at the time that  the  CTD
calibrations  might be different if the entire instrument could be immersed
in the temperature bath, rather than just the sensor assembly. Accordingly,
the CTD temperatures were corrected by adding 17 mK.  (The same  correction
was,  somewhat arbitrarily,  made to the shallow CTD cast 10620,  as  three
reversing thermometers had given offsets of 15, 20 and 8 mK).              
                                                                           
c) Conductivity/Salinity                                                   
The  conductivity ratio (C5,  Table 2) for the first deep CTD cast  (10621)
was  taken  as  the best value from the previous cruise  (1.0003).  Onboard
analysis  of  the 6 rosette samples (with a  Guildline  bench  salinometer)
suggested  an improved value of 1.0002,  which was inadvertently entered as
1.002  for casts 10624,  25 and 26.  An improved value of 1.00013 was  then
calculated,  and  used for the rest of the cruise,  except that  the  value
1.002  was again inadvertently entered off an old listing for  casts  10648
and 49.                                                                    
                                                                           
The mean salinity offset for casts 10624,  25,  26,  48 and 49 (all with  a
conductivity  ratio of 1.002) was 0.076 psu,  the offset for cast 10621 was
0.008 psu, and for all remaining casts was zero (Table 2).  Thus the errors
in conductivity ratio were corrected by applying constant salinity offsets,
after which Table 3 gives a standard deviation of less than 0.003 psu  over
129 samples (ignoring 5 obvious outliers).                                 
                                                                           
A  final calibration check is provided by comparing the theta/S curves  for
the two deep casts (10648 and 49) with the line found by Saunders (1984) to
fit  all deep casts north east of the Azores.  By comparison with Saunders'
curve, our salinities are 0.002-0.004 psu too high.                        
                                                                           
For the shallow CTD,  a final salinity correction of 0.035 psu was  applied
(Table 2) based on 7 samples (Table 3).                                    
                                                                           
d) Oxygen                                                                  
The  first cast with the deep CTD (10621) suggested that  a  multiplicative
adjustment  to C6,  reducing it by 0.885 from 0.00165 to 0.00146,  would be
sufficient to correct the oxygen values to the calibrated samples.  For the
shallow CTD the corrected value was found to be 0.00148.  The value 0.00146
should, therefore,  have been applied to all casts with the deep CTD.  At a
late  stage  of  the processing it was found that  the  value  0.00148  had
inadvertently been applied to casts 10639-47. Since the corresponding error
in  oxygen is less than 0.07 ml/l the data have not been further corrected.
From Table 3 it can be seen that the corrected oxygen values are  unbiased,
with  a  standard  deviation of less than 0.1 ml/l over  111  samples,  and
nearly all values lie within 0.2 ml/l of the true value.                   
                                                                           
We note that:                                                              
(1) Down and up casts differ by up to 0.4 ml/l.                            
(2) Although the CTD was always soaked in the water for five minutes before
starting the down cast,  there is still a marked hysteresis in the top  400
dbar  on  cast 10621.  This was the first cast of the cruise made with  the
deep  CTD,  and  shows  how  long  the  Beckman  oxygen  sensor  takes   to
acclimatise.  (3)  Drift and jumps in the oxygen values can  frequently  be
seen  at  each  calibration level where hauling  ceased.  (4)  The  largest
persistent  differences  between down and up casts occur between  1000  and
2000  dbar on the five casts which went deeper than 1200 dbar  (10621,  25,
39, 48 and 49).                                                            
                                                                           
In  summary,  the down casts may be absolutely in error  by  0.3-0.4  ml/l.
Relative errors between casts at similar depths should be smaller.         
                                                                           
3) Editing                                                                 
Two primary causes of noise were identified:                               
                                                                           
i)  A battery pack had been installed inside the CTD to keep it powered  up
while  the  multisampler  was  fired at calibration  points.  Without  this
modification, the oxygen sensor can take five minutes or more to recover on
each occasion.  Unfortunately,  the batteries themselves appear to go  open
circuit internally when subjected to considerable vibration.  This occurred
particularly  in  the top 150 dbar,  and at other depths when  cable  strum
built up,  which depended on the position of the cable laying guide  across
the  main  drum.  Irrecoverably  noisy data resulted which has  had  to  be
deleted, giving rise to the many gaps in the profiles,  extending across up
to 50 dbar.                                                                
                                                                           
ii)  A subtle computer fault caused the most significant bit of each  eight
bit  bype to be occasionally dropped as it was read in.  This affected  the
conductivity  sensor in particular,  causing errors of 128 in a single  raw
value.  After averaging over one second, or about 16 samples,  the error is
reduced  to an integer multiple of 8 raw conductivity units,  equivalent to
about 0.008 psu in salinity. The error could be spotted as quantised spikes
in  the salinity profiles,  most obvious in the deep casts near the end  of
the cruise.                                                                
                                                                           
To remove the computer caused small quantised spikes, a number of the casts
were listed in full and scanned,  identifying individual data cycles to  be
deleted.                                                                   
                                                                           
                                                                           
Table 1 - CTD Station List                                                 
                                                                           
Station   Day   Down    Lat.      Long.     Maximum      Calibration       
No.       No.   Time    N         W         pressure       Levels          
                (GMT)                       (dbar)     Sal   Oxy   Temp    
                                                                           
10620     29    2214    48 18.9   12 52.6    604       9     9     3       
10621     30    1809    47 03.9   14 08.5   2011       8     9     3       
10624     33    1226    45 46.9   15 58.9   1000       9     8     2       
10625     35    1334    46 49.0   18 08.1   2016       9     9     2       
10626     35    2115    47 30.8   18 59.2   1007       9     9     2       
10627     36    0213    46 55.7   19 08.0   1005       4     0     1       
10628     36    0657    46 19.9   19 15.2   1004       4     0     1       
10629     36    1130    45 45.5   19 23.0   1201       4     4     1       
10630     36    1552    45 26.8   19 24.3   1208       4     4     1       
10631     36    1847    46 10.2   19 30.5   1210       3     3     1       
10632     37    0010    44 36.7   19 38.9   1004       4     0     1       
10633     37    0520    44 00.0   19 46.3   1210       4     0     1       
10634     37    0943    43 24.8   19 53.5   1006       4     4     1       
10635     37    1409    42 50.4   20 02.5   1017       4     0     1       
10636     37    1837    42 15.6   20 09.1   1013       4     4     1       
10637     37    2241    41 40.8   20 16.0   1005       4     4     1       
10638     38    1845    40 29.7   20 29.6   1005       3     4     1       
10639     39    1213    41 16.5   19 41.0   1514       4     4     1       
10640     40    1907    41 21.2   18 51.5   1213       6     6     1       
10641     41    0226    41 27.1   18 06.6   1200       6     0     1       
10642     41    0931    41 31.2   17 18.7   1209       6     6     1       
10643     41    1614    41 34.8   16 31.8   1208       6     6     1       
10644     41    2255    41 39.9   15 43.6   1210       6     6     1       
10645     42    2218    41 59.8   14 00.7    812       6     6     1       
10646     43    1043    42 14.9   12 59.8   1209       5     5     1       
10647     43    1557    42 14.5   12 20.1   1208       6     6     1       
10648     44    2057    43 56.9   15 53.1   5081       7     7     3       
10649     45    0253    44 17.7   16 08.0   4036       7     7     3       
10650     47    0007    40 59.1   20 25.4   1233       4     4     1       
                                                                           
                                                                           
Table 2 - Calibration Constants                                            
                                                                           
               Initial calibration constants            Final corrections  
                                                                           
          Pressure  Temperature    Cond.    Oxygen     Temp   Sal.   Oxygen
                                                                           
Cast(s)   C1    C2  C3    10(3)*C4 10(3)*C5 10(3)*C6   C7     C8     C9    
                                                                           
10620     0.025  45 0.000 0.5      1.0      1.65       +0.017 +0.035 0.906 
10621     0.1   120 0.034 0.5005   1.0003   1.65       +0.017 -0.008 0.885 
10624-26  0.1    50 0.034 0.5005   1.002    1.46       +0.017 -0.076 1.0   
10627-38  0.1    50 0.034 0.5005   1.00013  1.46       +0.017  0.0   1.0   
10639-47  0.1    50 0.034 0.5005   1.00013  1.48       +0.017  0.0   1.0   
10648-49  0.1    50 0.034 0.5005   1.002    1.46       +0.017 -0.076 1.0   
10650     0.1    50 0.034 0.5005   1.00013  1.46       +0.017  0.0   1.0   
                                                                           
P(CAL)(dbar) = C1*(P(RAW)-C2)                                              
P(CORR) = P(CAL)                                                           
T(CAL)(degC) = C3 + C4*T(RAW)                                              
T(CORR) = T(CAL) + C7                                                      
C(CAL) (mmho/cm(3)) = C5*C(RAW)                                            
S(CORR) (psu) = S(CAL) + C(8)                                              
0(CAL) (ml/l) = C6* 0(RAW)* EXP(-0.036*T(L)-0.000155 P(CAL)) * 0(Saturated)
(T(CAL),S(CAL))                                                            
where T(L) is T(CAL) lagged with a 300 second time constant.               
0(CORR) = C9 x 0(CAL)                                                      
                                                                           
                                                                           
Table 3 - Calibration statistics                                           
                                                                           
(Calibration sample - CORRected CTD)                                       
                                                                           
                            Deep CTD              Shallow CTD (10620)      
                                                                           
                      P     T     S    0         P    T     S     0        
                      dbar  mK    ppm  ml/l      dbar mK    ppm   ml/l     
                                                                           
no in sample           26    32   129   111      3     3     7     8       
mean                  -2.6  -0.1  0.3  -0.002    0.0  -2.7  -0.1  -0.029   
standard deviation     2.5   8.4  2.8   0.087    1.0   6.0   2.7   0.192   