Cast specific comments, quality assessment, analytical methods as prepared
by L. Codispoti
All stations
See Codispoti documentation regarding data quality see section on DATA
QUALITY below.
Station 1 cast 1
This station was taken on the way to the Arabian Sea from Singapore
near Sri Lanka to the South of the Bay of Bengal.
Station 5 cast 1
This station was designed to check the flushing characteristics of the
10 liter Niskins on the hydrographic rosette. The rosette was pulled
through a strong gradient into a fairly uniform layer and bottles
were fired immediately, after 21 sec, etc. until 149 seconds. Based
on these data, it was decided that a 20 second soak time was ample
for flushing the 10 liter Niskin bottles. On later cruises, longer
flushing times than would be suggested by the data were used. This is
because the CTD sensors are surrounded by additional sensors added
for other JGOFS investigators.
Station 6 cast 1
This cast was made with the large bottle rosette for special chemical
samples. The rosette was equipped with a mixture of bottles. Only
salinities and Winkler O2's on selected bottles, no nutrients. The
three replicate Oxygens near the surface agreed well, but one of the
two oxygens at 28-29 decibars is suspect.
Station 6 cast 3
This was a large bottle rosette cast for special chemistry
samples. The data from bottle 23 (Seq.23) were deleted because
this bottle appeared to be a mis-trip.
Station 6 cast 4
Another large bottle rosette cast for special chemistry samples.
As usual, the large bottles were paired with 10 L Niskins from which
the samples for the chemical data were taken. CTD "spiking" problems
occured which could compromise the data, particularily the depths of
bottles 13 and 15.
Station 7 cast 1
All oxygen samples were drawn by trainess from Oman and
Pakistan. Bubbles were noted in flasks from bottles
18-24. These questionable data have been deleted.
Station 8 cast 1
We had "spiking" problems with the CTD/rosette that could have
caused mis-trips at this station, but the depths seem o.k.
The higher colorimetric oxygens seem to be systematically lower
than the Winklers which probably means that they were outside the
high range for accuracy with the colorimetric method. These values
have been deleted.
The elevated NH4 value at 1514.7 db occurred on several casts. Competing
hypotheses for this peak are contamination from the Niskin bottle or a
concentration of zooplankton activity in this layer.
Station 8 cast 4
Casts 02 and 03 at Station 008 were too badly comprimised by
electrical "spikes" to make it worthwhile to collect water.
This shallow cast was taken because it was orginally thought
that the shallow values from cast 01 were comprimised by spiking.
We believe that the depths for samples for casts 01 and 04 are o.k.
but there is a possibility that that they are in error because of
the spiking problem.
Station 9 cast 1
No significant problems on this cast, but some questionable oxygens
are not reported.
Station 10 cast 1
The Silicate (Silicic Acid) from bottle 3 is probably incorrect and
was deleted.
Station 11 cast 1
The surface oxygen saturation at this station was 125% which is possible
given the relatively high nutrients at the sea surface. The salinity
at 252.3 db is questionable and was deleted.
Station 12 cast 1
Another station with appreciable nutrients and O2 supersaturation
at the surface.
Station 18 cast 1
Special Chemistry cast with large bottle rosette and mixture of bottles.
All Winklers questionable and have been deleted because flasks were
only shaken once before running. All data from bottle 13 are questionable
because of leaking Niskin bottles.
Station 18 cast 3
Special Chemistry cast with large bottle rosette and mixture of bottles.
Station 18 cast 5
Another special chemistry cast.
Station 18 cast 6
Another special chemistry cast with large bottle rosette.
Station 18 cast 9
The bottles were not tripped in order of their sequence on the rosette.
Station 19 cast 1
Electronic spiking in CTD/rosette system make depths between 26.8 and
454.4 db somewhat uncertain. The depths listed are our present "best guess".
Station 21 cast 1
Spiking problems could have caused mis-trips, but depths look
o.k. Because of the problems two bottles were tripped at 1014.5 db.
Station 21 cast 2
Another special chemistry cast. Air leak at top of bottle 9
and relatively high result makes Winkler values questionable. Spiking
problems occurred which makes bottle mis-trips a slight possibility.
Station 21 cast 4
Slight possibity of mis-trips due to electrical spliking in CTD/rosette
system, but data look good.
Station 22 cast 1
Once again, electrical spiking in the CTD/rosette system introduces a
possibility of mis-trips, particularly between 26-300db.
Bottle 1 was definitely a mis-trip and the data have been eliminated.
The bottle 5 "hung up".
Station 23 cast 1
Another special chem. cast with large bottle rossette.
Station 23 cast 2
The bottle 1 mis-tripped again.
Station 23 cast 3
Another special chemistry cast with the large bottle rosette.
Station 23 cast 4
Another special chemistry cast with the large bottle rossette.
The bottle 9 leaked but the data look o.k.
Station 23 cast 6
This is another special chemistry cast with the large bottle CTD/rosette.
DATA QUALITY
JGOFS Arabian Sea Cruise TN039 (Set-up and Calibration Cruise)
Sept-Oct. 1994: QA/QC Report for the Niskin and Go Flow Bottle
Data (Bottle Salinities, Oxygens and Nutrients)
L.A. Codispoti (lou@ccpo.odu.edu)
Old Dominion University, May 1995
General Comments:
This "readme" file pertains to the salinity, dissolved oxygen,
and nutrient data taken from sampling bottles during RV T.G.
Thompson cruise TN039. This cruise took advantage of the sampling
and training opportunities provided by the Thompson's transit leg
from Singapore to Oman. The purposes of this cruise included:
1)testing equipment and methods that would be used on the
subsequent JGOFS Arabian Sea process cruises,
2)finalizing the hydrographic and data-processing protocols that
would be used on subsequent JGOFS Arabian Sea process cruises,
3)training participants from Pakistan and Oman,
4)collecting as much data as possible to extend the temporal and
spatial coverage of the time-series observations included in the
JGOFS Arabian Sea process study.
Because the JGOFS data base system does not have a system for
"flagging" questionable data, no questionable data are included
in the files sent to the JGOFS Data Management Office. These data
are available by sending an Internet message to "lou@ccpo.odu.edu".
No units are given for salinity in this report because the most
recent definitions of salinity define it as a dimensionless
number. To accomodate every preference, Winkler oxygen values
are reported in ml/l, micromolar and micromoles per kg. The
latter values can only be calculated with a knowledge of the
oxygen sample temperatures when the samples were drawn. These
"draw temperatures" are not reported here, but can be obtained by
contacting lou@ccpo.odu.edu. Nutrient values are reported in
micromolar. They can be converted to micromoles per kg, by
combining lab. temperature on the Thompson (approx. 23.5 deg C)
and the salinity of the sample to compute density and then
dividing the value in micromolar by this number.
Methods:
In general, the methods employed for the bottle Salinity, Winkler
dissolved oxygen, and nutrient analyses did not differ
significantly from those described in the JGOFS protocols that
were distributed in June, 1994. Minor differences included the
following:
1) Sea Bird CTD systems and bottle carousels were employed (SBE-
9+ underwater units, SBE-11 deck units, SBE-32 carousels). These
units represent a newer generation of equipment than the units
described in the JGOFS protocols.
2) The weights of the salts used for primary standards for
dissolved oxygen and nutrients were not adjusted to an "in vacuo"
basis as suggested in the protocols. It is unlikely that this
departure from procedure would cause significant errors. Our
calculations suggest that any differences arising from our
decision to not correct to an "in vacuo" basis would range from
0.02% (oxygen standards) to 0.06%(ammonium standards).
3) The protocols give one a choice of adjusting nutrient methods
so that calibration curves are strictly linear, or opting for
more response and taking into account non-linearities. We choose
the latter method.
4) No corrections were made for "carryover" between nutrient
samples run on the Technicon Autoanalyzer. Data from this cruise
and a subsequent cruise suggest that carryover effects in our
nutrient analyses are generally less than 1-2% of the
concentration difference between adjacent samples.
5) Calibration and re-calibration of volumetric ware was not as
rigorous as described in the JFOFS protocols, but this was
largely compensated for by comparing independent standards
diluted with independent volumetric ware.
6) Duplicate oxygen samples were not drawn from every Niskin or
Go-Flo bottle, but there were several comparisons of bottles
tripped at the same depth and numerous comparisons of the Winkler
and colorimetric oxygen values.
7) Azide was added to the Winkler oxygen pickling reagents to
destroy nitrite that can be present in relatively high
concentrations in the Arabian Sea.
Cruise TN039 contains some oxygen determinations made using the
colorimetric method of Broenkow in Cline (1969) which is
optimized for low dissolved oxygen concentrations. This method
is not described in the JGOFS protocols. Similarly, a method for
the automated determination of ammonium is not included in the
JGOFS protocols. The method described by Whitledge et al. (1981)
as modified by K. Krogsland of the University of Washington
(kkgrog@u.washington.edu) was employed for this analysis.
Temperature Data:
The temperature data associated with each bottle data depth were
taken by the CTD system during the bottle tripping process.
Consult the CTD data report for this cruise to learn more about
the CTD system.
Sampling Bottles:
Most of the samples in this report were taken from 10 liter
Niskin bottles. A few samples were taken from 20 and 30 liter Go-
Flo or Niskin bottles. Information about what type of bottle a
sample came from can be obtained by sending an Internet message
to lou@ccpo.odu.edu.
Because there is little or no lag time between triggering a
bottle and bottle closure with the new SeaBird rosette systems
a bottle flushing experiment was performed. The rosette
was raised through a strong gradient into a mixed layer and
then a sequence of bottles was tripped over about a two
minute period. This experiment suggested that the bottles
flushed fairly well and that a 20 second "soak time" should
be sufficient before tripping a bottle at a given depth.
On a subsequent cruise (TN043), it was found that bottle soak
times had to be increased largely because of relatively slow
response times for the CTD sensors. The bottles were probably
flushing relatively rapidly but the companion CTD data for
salinity showed some variation that disappeared with longer soak
times. This was probably because of the additional equipment
mounted near the CTD sensors during the subsequent cruises. This
equipment can act as a heat source/sink and interfere with
flushing and equilibration of the CTD sensors on the up cast.
NOTE THAT THE MID-POINT OF THE SAMPLING BOTTLES WAS ONE METER
ABOVE THE CTD SENSORS. THE DATA HAVE NOT BEEN CORRECTED FOR THIS
ONE METER DIFFERENCE BETWEEN CTD SENSOR AND SAMPLING BOTTLE
POSITIONS.
Salinity:
Salinities were run on almost every bottle sample with new vials
of standard sea-water used before and at the end of every run
(12-36 samples). These runs, suggested that drift during runs
was usually less than 0.0005. Agreement between bottle salinities
and the recently calibrated sensors on the Sea Bird CTD systems
was usually better than 0.01 after final data processing.
For depths greater than 500 db, the standard deviation between
bottle salinities and the CTD results after final calibration was
0.002 for two of the three CTD systems. The third system that
was used only to collect a few "special chemistry" samples had a
standard deviation of 0.005 for this depth range. Consult the
companion TN039 CTD data report for a fuller discription of these
data.
Dissolved oxygen:
The Winkler dissolved oxygen set-up was built and supplied by the
SIO/ODF group. This system is computer controlled and detects
the end-point photometrically. Temperature of the thiosulfate
and standard solutions is automatically monitored by this system.
A primary standard provided by Lou Codispoti was compared with
the SIO/ODF primary standard. The agreement between these
standards was plus or minus 0.02 per cent. These standards were
made up at different institutions and diluted at sea with totally
independent volumetric ware.
We tested the effects of using silicone vs tygon Tubing to draw
dissolved oxygen samples for the benefit of Ed. Peltzer who was
concerned about DOC contamination from Tygon tubing. There
appeared to be no difference between the results.
Because we will not have the person power to perform colorimetric
dissolved oxygen concentrations routinely during the process
legs and because of the existence of suboxic water in the Arabian
Sea, we did a comprehensive comparison of colorimetric vs
automated Winkler oxygen analyses. Generally, the results agreed
within better than plus or minus 0.005 ml/l with perhaps a
tendency for the automated Winkler to overestimate by about
0.005ml/l in the less than 0.01 ml/l (about 0.5micromolar) range
compared to the colorimetric method.
NOTE THAT THE OBSERVATIONS WERE MADE WITH HIGHLY EXPERIENCED ANALYSTS
DRAWING THE SAMPLES AND BY ALLOWING AT LEAST THREE BOTTLE VOLUMES TO
OVERFLOW THE WINKLER OXYGEN FLASK (CONSUMING ABOUT AT LEAST 0.7L
OF WATER)WHEN DRAWING WINKLER SAMPLES.
We performed some iodine blanks on sea-water. The results were
intriguing and suggest small positive and negative blanks
(about 0.5 micromolar)in the suboxic waters. Stay tuned for
further developments.
Nutrients:
Terminology describing nutrients has become somewhat loose
over the years, so it may be useful to point out that for our
purposes silicate=silicic acid, and phosphate=reactive phosphorus.
Nutrient analyses were performed on a 5-channal Technicon II AA
system that was modified and provided by the SIO/ODF group.
In assessing the nutrient standard comparisons outlined below,
note that the full-scale ranges for nutrients were as follows:
Nitrate =0 to 45 micromolar
Nitrite =0 to 5 "
Phosphate =0 to 3.6"
Silicate =0 to 180 "
These ranges were arrived at after an Internet pole of PI's and
cover the full depth concentration range for the Arabian Sea.
The SIO/ODF nitrate and nitrite standards and standards from the
National Institute of Oceanography (NIO) in India (provided by S.W.A.
Naqvi) were compared with the following results:
NIO Nitrate Std.= 22.6 micromolar; SIO/ODF=22.5 micromolar
NIO Nitrite Std.= 2.42 micromolar; SIO/ODF = 2.50 micromolar
As can be inferred from the above, the nitrate plus nitrite
values were almost identical in the mixed standards;
25.02 (NIO) vs 25.00 (SIO) micromolar.
Lou Codispoti prepared independent primary nitrate, nitrite,
silicate and phosphate standards for comparsion with SIO/ODF primary
standards, and made dilutions using glassware entirely independent of the
SIO/ODF glassware.
The results were as follows:
Codispoti SIO/ODF
Nitrate 26.96 micromolar 26.85 micromolar
Nitrite 2.90 " 2.86 "
Silicate 86.4 " 85.8 "
Phosphate 2.36 " 2.36 "
All of the above results are within plus or minus 0.5% of the
full scale values, and with the exception of nitrite, the rest
are within plus or minus 0.2% of the full scale values.
Because nitrite values in the suboxic waters of the Arabian Sea
can attain values of approximately 5 micromolar, we tested the
efficiency of the Cd column that reduces nitrate to nitrite in
the nitrate analysis towards the end of the cruise. The
efficiency was 98.3 per cent. The column may have been more
efficient at the beginning of the cruise. We assumed that the
column was 100% efficient. A 2% error in assumed column
efficiency would in the worst case introduce an error of 0.1
micromolar in nitrite (nitrite=5 micromolar), but most of the
errors would be much smaller.
The ammonium results were the least precise as expected
given the state of the methods available. Three primary
standards were compared with agreement of about plus or minus
three per cent of the full-scale value. Based on our experience,
we feel that the standards probably agreed within the precision
of the method, but we found a significant salinity effect on the
ammonium results that might explain some of these differences
since the salinities of the comparison standards varied a bit.
Experiments on the first JGOFS Arabian Sea process study cruise
(TN043) suggest that the ammonium signal decreases by
approximately 3.5% for a salinity increase of 1.00. Thus,
salinity differences between samples and standards have to be
taken into account when calculating the final ammonium
concentrations. The ammonium values in this report have been
corrected for this effect.
Acknowledgements:
I thank everyone who helped me with the above work,
particularly K. Krogsland, J. Kinder, R. Kohrman, D. Masten, W.
Martin, S.W.A. Naqvi, R. Patrick, W. Peterson, J. Aftab, G.
White, and M. Realander.
References:
Broenkow, W.W. and J.D. Cline (1969) Colorimetric determinaton of
dissolved oxygen at low concentrations. Limnology and
Oceanography, 14, 450-454.
Whitledge, T.E., S.C. Malloy, C.J. Patton, and C.O. Wirick (1981)
Automated Nutrient Analysis in seawater. Brookhaven National
Laboratory Report 51398, 216pp.