         CTD SENSOR CALIBRATION CIROLANA 1/1993 (JONUS EXERCISE 11)        
                                                                           
This  cruise used the Guildline CTD s/n 45056 at the JONUS stations.  Three
stations  (85,  87 and 88) used the new FSI CTD,  the first occasion it had
been  used,  and  the  results from these  three  stations  are  summarised
separately.  The following is a presentation of the data returned from  the
JONUS stations, 1-80.                                                      
                                                                           
Thirty-eight CTD profiles were obtained using the Guildline CTD during  the
cruise.  Reversing thermometers were used to measure water temperature  and
samples were collected for salinity analysis                               
                                                                           
                                                                           
(1) Thermometer data                                                       
Left and right thermometer temperatures were available on 41 occasions, but
three pairs were subsequently deleted because of acclimatisation  problems:
the  water temperatures were varying over the four minutes allowed for this
and so invalidated the thermometer values.                                 
                                                                           
All thermometers were graduated at 0.02 deg C intervals and a comparison of
the differences is as follows:                                             
                                                                           
Difference deg C        No.        Accumulative %                          
     0.00                 7              18                                
     0.01                19              68                                
     0.02                 8              89                                
     0.03                 2              95                                
     0.04                                95                                
   > 0.04                 2             100                                
                                                                           
Thus  95%  of  the  differences lie within the range  +/-0.02  deg  C,  the
thermometer scale gradation.                                               
                                                                           
                                                                           
(2) Sample salinity data                                                   
Duplicate water samples were collected on 80 occasions for analysis with  a
Guildline salinometer. The distribution of the differences is as follows:  
                                                                           
  Difference         No.        Accumulative %                             
0.000 - <0.003        68              85                                   
0.003 - <0.006         3              89                                   
0.006 - <0.009         1              90                                   
0.009 - <0.012         2              93                                   
    >0.12              6             100                                   
                                                                           
This  is not as good as usual for we normally have 95% of  the  differences
within 0.006.                                                              
                                                                           
                                                                           
(3) Sensor Calibration for Guildline CTD                                   
                                                                           
(a) Pressure                                                               
The laboratory calibration of March 1992,  sensor at 6 deg C,  was used  to
correct the Guildline CTD pressure sensor:                                 
                                                                           
P(cor) = P(ctd) - 1.6 db                                                   
                                                                           
(b) Temperature                                                            
Fig. 1 shows the differences between the (mean) thermometer and uncorrected
CTD  temperature  when  the  left/right  thermometer  difference  was  less
than/equal to 0.04 deg C (36 occasions). The mean difference calculated for
the  34 instances when the left and right thermometers themselves  differed
by less than/equal to 0.02 was 0.005 deg C.  This is in excellent agreement
with the laboratory pre-cruise calibration of 30 December,  which estimated
the CTD correction to be 0.008 deg C at the temperatures encountered during
the  cruise.  The sensor was therefore corrected using the laboratory  pre-
cruise calibration coefficients:                                           
                                                                           
 T(cor) = T(ctd) + dT                                                      
     dT = a*T(ctd)*T(ctd) + b*T(ctd) + c                                   
                                                                           
where a = 4.33738e-5                                                       
      b = -6.45026e-4                                                      
      c = 1.0307e-2                                                        
                                                                           
Fig.  2  shows  the  differences  between  these  36  thermometer  and  CTD
temperatures after the latter have been corrected. The mean difference is -
0.003  deg C and all differences lie within +/- 0.03 deg C,  supporting the
belief that if the thermometers are accurate to +/- 0.02 deg C then the CTD
temperatures are accurate to +/- 0.01 deg C.  In fact, only two differences
are outside +/- 0.025 degC.                                                
                                                                           
(c) Salinity                                                               
Fig.  3 shows the difference between the water sample salinity as  measured
with  the  Guildline salinometer and that derived from the CTD  before  any
calibrations  have been applied to the latter's sensors.  Note that  a  few
data  values have been removed prior to plotting Fig.  3 because they  were
not suitable for use in determining calibration coefficients:              
                                                                           
(i)  Stations  23,  34 to 39,  60 and 69 were from the Humber  Estuary,  an
anchored  station at the mouth of the River Humber,  the Thames Estuary and
the  entrance  to  the  River Humber  respectively.  Throughout  the  JONUS
exercise,  it  has  been found that the CTD salinity estimates  from  these
areas   are  highly  suspect,  perhaps  because  of  high  sediment   loads
influencing  the  instrument conductivity measurement,  and CTD  data  from
these stations have not been used.                                         
                                                                           
(ii)  The conductivity ratio recorded by the CTD at stations 2 (bottom),  3
(bottom) and 33 (surface) was variable during the time that the calibration
sample was collected and the CTD data have not been used.                  
                                                                           
Fig. 3 suggests that the CTD has a tendency to over-estimate salinity.     
                                                                           
Fig. 4 shows the ratio of CTD:water sample conductivity ratio after the CTD
pressure  and  temperature  sensors have been  corrected  using  the  above
coefficients.  A  notable  feature  of Figs.  3 and 4 is  in  the  relative
difference  in  the  distribution of the ratio at stations 69  to  80  when
compared  with  that at stations 2 to 63 (the former were worked some  five
days after station 63 following a mid-cruise break).                       
                                                                           
This  difference  between the two groups is clearly seen in Fig.  5,  which
shows  the water sample/CTD salinity difference after both CTD pressure and
temperature have been corrected.  Calculation of the 95% confidence  limits
for the mean differences of the two groups suggests a significant change in
CTD response:                                                              
                                                                           
stations 2 to 63: mean difference -0.008 +/- 0.003 (i.e. -0.011 to - 0.005)
stations 69 to 80: mean difference -0.025 +/- 0.004 (i.e. -0.029 to - 0.021
                                                                           
To  satisfy this,  two sets of coefficients have been derived to  calibrate
the  CTD conductivity sensor,  using a least square linear fit between  the
ratio  of  water sample and CTD conductivity and the  CTD  temperature  and
pressure.  Those  salinometer  samples with duplicates  which  differed  by
>0.006 have not been used to determine the conductivity calibration.       
                                                                           
Coefficient        Stations 2 to 63          Stations 69 to 80             
     a               -1.699686e-5              -6.096531e-5                
     b                7.705632e-7               2.124100e-6                
     c                  1.000082                  0.999895                 
                                                                           
RMS salinity difference between water sample and corrected CTD:            
  Stations 2 to 63        0.009                                            
  Stations 69 to 80       0.005                                            
number of data values:                                                     
  Stations 2 to 63           39                                            
  Stations 69 to 80          10                                            
                                                                           
Figs.  6 and 7 illustrate how effectively the CTD conductivity and  derived
salinity  have been corrected.  The discontinuity between the two groups of
samples  has  disappeared  and the tendency of  the  CTD  to  over-estimate
salinity has been overcome.                                                
                                                                           
The  histograms in Fig.  8 show how well the CTD conductivity is  corrected
since  the  upper  frame  has been derived after the  CTD  temperature  and
pressure have been corrected,  but before the CTD conductivity  calibration
has been applied.                                                          
                                                                           
If  it  is assumed that the salinometer is accurate to 0.006  and  the  CTD
salinity to 0.01,  then differences between +/- 0.016 are acceptable.  5 of
49  values lie outside this range after the calibrations have been applied;
6 of 49 lie outside +/- 0.013.                                             
                                                                           
                                                                           
(4) Transmission/suspended load                                            
A 25 cm path length transmissometer (s/n 198) was fitted to the CTD rosette
and  calibrated in terms of suspended load (mg/1) by comparing  the  logged
%transmission with laboratory determinations of suspended load.            
                                                                           
suspended load = a*loge(%Trans) + b                                        
                                                                           
where a = -12.42                                                           
      b = 45.82                                                            
                                                                           
                                                                           
Ken Medler                                                                 
7 April 1993                                                               
