OPS Performance Report
Polarstern Cruise ARK XXV/1, 2010
Gereon Budéus, Andreas Wisotzki
29.06.2010
The Optimare Precision Salinometer (OPS) s/n 003 was purchased in early 2010 and, for a short time, was put into service in the lab on land. Performance there was sufficiently good to take it on the cruise for further sea expedition tests of the instrument. The cruise started on 10 June 2010.
The OPS is a relatively recent development. It aims at the determination of salinity with accuracies better than 0.001 in rooms without temperature control. Sample intake and processing are highly automated. User interface is a touch screen (supplemented by a mouse and a keyboard if desired). A pre-bath is used to temperate the sea water samples. The conductivity cell is located in the main bath, which is extremely well temperature-controlled. This goes to the extent that internal lights are switched off during the measurements, as their radiant power deteriorates the measurements on this precision level. The temperature of the main bath is constantly monitored. Thus, the validity of measurements is not guesswork but fully proofed. The instrument records data only when the thermal management is perfect. This status is reached usually after only a few minutes from switching on power. Data are exported via network or USB.
Via the ‘Main Menu’, the choice ‘Measurement’ leads to either ‘New Sample’, ‘Repeat Sample’, or ‘Standardize’. After the choice of ‘New Sample’, a box and a sample number are requested before the intake is lowered into the sample bottle and pumping is started to flush the cell and measure thereafter. ‘Repeat sample’ can be used to continue with measurements from the same bottle and reduces flushing in comparison to ‘New Sample’. ‘Standardize’ is used in the usual manner with Standard Sea Water of known salinity. Most actions can be adjusted to the particular needs of the user. Critical parameters can be changed only by a ‘super user’ who has to log in.
1. Setup
The OPS was transported in its Zarges alu-box, the sample intake and cell being filled with standard sea water. Pre bath and main bath were emptied for the transport. An ordinary dry lab on the starbord side of Polarstern (E-deck) was chosen as the place to set up the OPS. The room is not air conditioned and the door to the adjacent corridor is mostly left open. Main and pre bath were filled on 11.6.2010, the instrument was switched on in the morning of 12.6.. After 5 minutes the instrument reached operational status and the first samples were measured.
Fig. 1: General setup of the Optimare Precision Salinometer in the dry lab of Polarstern
Fig. 2: Opposite side of the lab with sample bottles waiting for evaluation
2. Sample consistency
The first step of the performance test was to check if water samples are measured consistently or show adversely large ranges in this environment without airconditioning. For this purpose, residual water samples from the previous cruise were arbitrarily combined to full bottle samples. The absolute salinity of these samples is of no concern. Bottles were shaken well before measuring them.
Measurements were performed with the following setup parameters (‘standard procedure’):
- Filling pump strokes: 15
- Drain pump strokes: 2
- Rinse cycles: 5
- Pre-bath offset: -0.2 K
- Residence time (of sample water in the cell): 11 sec
- Measuring time: 3 sec
- Measurements per sample bottle: 3
- Allowed salinity range for acceptance of these 3 consecutive measurements: 0.0003
The first sample bottle contained water with a salinity of above 37.0, presumably due to the fact that it had been previously left open for a long time. It was immediately apparent that with this salinity difference of more than 2.0 (from 35. to >37.) the number of rinse cycles did not suffice to obtain consecutive salinities in the range of 0.0003 immediately after 5 rinse cycles. A few more flushing cycles were necessary to flush tubes and cell completely. After 4 additional flushings, the salinity measurements were stable within a range of 0.0001.
Four additional bottles were measured during the two following days with the same result, except that the 5 rinse cycles were sufficient when salinity differences of consecutive samples were small. Two of these bottles were measured under adverse weather conditions with wind force 10 and 5m wave heights (12.6., close to Westkapp Norway).
In order to check the consistency of measurements beyond the preselected 3 measurements per sample, a sample bottle was measured as long as it was possible to take in water. Meplat bottles of 200 ml content were used, the sample intake ending about 1.5 cm above the bottle bottom. After 5 rinse cycles the result was:
36.4974
36.4970
36.4967
36.4965
36.4963 This shows that 5 rinse cycles do not suffice if salinities of bottles differ markedly
36.4963
36.4963
36.4963
36.4962
36.4962
36.4962
36.4962
36.4963
36.4963
36.4962
36.4962
36.4962
36.4962
36.4962
36.4962
36.4962
36.4963
36.4962
36.4962
36.4962
36.4962
36.4961
36.4962
-- abort ---
Various similar measurements of entire sample bottle contents showed that evaporation during the evaluation of a sample bottle is of no concern when working according to the standard procedure. One operates on the safe side by not using the last quarter of the bottle volume.
On 14.6.2010, the standard procedure was used again for two bottles. In addition to the ‘New sample’ evaluation, ‘Repeat sample’ was also used. Bottle 3 showed
‘New sample’: 35.7223
‘Repeat sample’: 35.7223
After leaving the bottle open for about 15 min:
‘Repeat sample’: 35.7229
A new mixture of leftovers was put into the same bottle and evaluated. Number of rinse cycles were set to 10 cycles to ensure complete flushing of the old sample water on the given precision level.
‘New sample’: 35.8944 (35.8944, 35.8944, 35.8943)
‘Repeat sample’: 35.8946 (35.8946, 35.8946, 35.8947)
3. Principal sample treatment
Out of curiosity, this bottle – only the half empty one – was shaken again. The expected result was that the salinity should increase by this action due to forced evaporation and washing off leftovers from the upper, empty part of the bottle.
‘Repeat sample’: 35,8973 (35.8972, 35.8973, 35.8973)
Same procedure repeated:
‘Repeat sample’: 35,9002 (35.9001, 35.9002, 35.9002)
Thus, each of these shaking actions increased the salinity of the remaining water by roughly 3/1000.
Next try was to check whether it is possible to use a cold standard water bottle for standardization, or whether such a bottle has to be temperature-conditioned before using it. We simply measured the bottle until it was empty. Bottle temperature was about 13°C (personal estimate, no measurement), room temperature 21°C, main bath temperature slightly higher. The bottle was shaken very well. Nevertheless, it was impossible to achieve consistent readings within limits of 1/1000. Values were wildly varying by about 5/1000. We suspect that the temperature difference between bottle and room reinforces any existing inhomogeneity in the bottle, leading to such adverse results. This topic has clearly to be investigated further. It seems clear that the OPS had no difficulties to temperate the sample perfectly (main bath temperature remained constant to better than 0.5 mK), so the problem seems to be caused by the bottle content itself. This conjecture is supported by previous observations in the land lab, that a bottle taken out of a fridge shows grossly varying salinities, while the same water stored at room temperature leads to consistent results.
15.6.2010
A standard water bottle with S=34.994 from 2007 was stored at room temperature for 12 h. Measurements with the OPS were then consistent within a range 0.0003. The last standardization of the OPS was on 22.4.2010, i.e. seven weeks ago on land. Readings were now 0.003 low with respect to the nominal salinity of the standard water bottle used today (reading: 34.991).
16.6.2010
Standardize with OSIL standard water bottle of S=34.994.
The instrument showed S=34.9928 with the old standardization of 22.4.2010.
Slope had to be corrected from 1.0002164 to 1.00025065.
Measurement as ‘new sample’ from this bottle: 34.9938, 34.9939, 34.9939.
‘Repeat’: 34.9939, 34.9940, 34.9940.
Next important step was to use fresh ocean water samples from the stations performed on our cruise. This revealed important insights about bottle treatment. Deep water samples from at least 1,600 m depth were used.
We measured:
a) Box 20, Bottle 2, not well adjusted to room temperature
Salinity values decrease first for many measurements and remain later at fluctuating values between 34.905 and 34.902. (Conjecture at this time was that room-sample temperature gradients are adverse to precise measurements. This is not necessarily true, see below.)
b) Box 9, bottle 1, standing over night on top of the desk in the lab (17.6.)
Very well shaken.
During ‘rinse’ the values decrease, decreasing further during measurements, total range many 1/1000, close to 0.01. No stable readings. Measurement ended after half of the bottle volume was used up. Bottle was then shaken again very well with a thumb used to close it. Salinities were than more stable:
34.9150, 34.9150, 34.9149
But ‘Repeat’ was not consistent with this result:
34.9158, 34.9159, 34.9155, 34.9158, 34.9158, 34.9157
c) Andreas measured more examples of this type without getting results that were stable enough.
We were getting nervous about the fact that standard water revealed perfect measurements, and leftovers from the previous cruise, too (relative values), but newly sampled water did not perform similarily.
d) Box 8, bottle 3, special treatment: Warmed to about 30°C, cooled to room temperature. This did not improve measurements.
Initial values: 34.3671, decreasing to 34.3585 after 16 measurements, then rising again to values around 34.363 to 34.366. Values vary during the 10 sec residence time in the cell for about 0.001. This is not acceptable.
The waste water from the cell was collected in another random bottle which included some small residue from previous measurements. Normal shaking action. Already during the rinse cycles the salinity values were approaching a stable level.
‘New sample’: 34.4924, 34.4924, 34.4924
‘Repeat’: 34.4924, 34.4924, 34.4925
e) Box 8, bottle 9
In order to check whether this development reflects a stratification in the sample bottle, we proceeded with the same procedure as d), but with the intake manually adjusted close to the sinking water surface in the bottle. Results were similar to those of d), with the same pattern in time. Waste of these measurements showed again immediately stable values. Two ‘repeat measurements’ confirming the result. We conclude that the sequence of readings does not mirror a stratification in the sample bottle.
f) Box 8, bottle 4 (20.6.2010)
This measurement revealed the breakthrough with respect to the identification of sample treatment problems:
On the previous evening, an injection needle was pushed through the rubber stopper of the bottle to release the overpressure which is present in every sample bottle when it was filled with cold water and later adjusted to room temperatures. The idea was to check for a potential influence of probably invisible outgassing after the opening of the bottle, which would result in erroneous readings if this gas is included in the analysed water volume.
Results were immediately stable (34.3690, 34.3691, 34.3691),
‘Repeat’ (set to infinite measurements) showing values between 34.3695 and 34.3699, later including 34.3701 and 34.3693, with a very smooth development.
This experiment indicates that it is inhibited to open a sample bottle which shows overpressure and analyse it immediately. A pressure release has to be applied some time before the measurement in order to allow for an equilibration of gas concentrations and to avoid a contamination of the sample water by micro gas bubbles.
4. Sample treatment details: equilibration time and tempering
The question is, how much time is needed for the equilibration process. The fact, that the waste from the measured unequilibrated samples showed, in contrast to these, immediately stable readings, suggests that less than one hour might suffice for it. In order to define this time approximately we measured the following samples:
a) Box 8, bottle 5
Release overpressure: 08:55 h, start measurement: 09:16 h
The readings approach a stable value already during rinsing.
‘New measurement’ (infinite number of measurements):
34.3684
34.3685
34.3684
34.3683
82
82
82
80
80
79
79
79
78
77
76
76
75
75
74
75
74
Seemingly perfect at the beginning, the further development shows that the measurements are not free of artefacts from handling, which are in the range of 0.001 only, but could presumably be avoided.
(During this measurement, air temperature varied wildly due to the lab’s open door and the swift air exchange there with zero degrees C outside arctic air caused by open outer doors. Temperature in the main bath showed no sign of being affected, and salinity readings were stable.)
Various deep ocean water sample bottles have been treated slightly differently, until a reliable procedure was found which leads to readings that are consistent to a level of a few 0.0001:
The procedure includes a slight warming of the sample bottles in addition to pressure release and shaking. The tempering serves as a means to avoid gas saturation and microbubbles in the conductivity cell of the OPS.
On day 1: Shake and put sample bottle in a water bath that is hand-hot (use a bucket e.g.). After the water bath has cooled to room temperature, take the sample out, shake well, and release pressure by means of a injection-needle. Let bottle stand for further equilibration over night.
On day 2: Shake well. Release pressure again. Wait about 2 minutes. Measure with OPS.
Fig. 3: Sample bottles being tempered in lukewarm water
Fig. 4: Infusion needle applied for pressure release.
Bottles are made from hydrolyse class 1 glass quality, stoppers are butyl and secured by aluminium crimp caps
Results are as follows:
b) Box 11, bottle 5 (22.6.2010)
34.6611
34.6611
12
12
13
13
12
11
11
12
12
12
c) Box 11, bottle 7
34.3730
34.3729
30
30
30
30
30
31
32
31
31
31
These results show clearly that both tempering and pressure equilibration are mandatory to achieve ultimate accuracy when measuring salinity of an ocean sample.
With the above results, we were confident that we are able to measure ocean water samples to the desired accuracy. Thus, we standardized the OPS and measured samples with the above described procedure applied (equilibrating and tempering). Rinse cycles were set to 10, as we preferred perfect rinsing over insignificant time savings.
5. Standardization (23.6.2010)
With a standard water bottle of S=34.994 (batch P145 from 5.10.2007) we immediately got 3 identical readings, giving a slope correction of
1.0001426426
with readings that were calculated with the old slope correction (16.6.2010, see above) being 0.004 too high.
As this value differs markedly from the previous, but resembles much the one preceding it, we repeated the standardisation with yet another standard water bottle, giving readings:
34.9939, 34.9940, 34.9940
thus confirming the actual result.
6. Sample measurements
During the next few days, we evaluated 34 deep water samples which we collected during the preceding days. The OPS was left running (though we could have switched it off as it is quickly opertional), and no new standardization was applied. 10 rinse cycles were used, 3 consecutive measurements with an allowed range of 0.0003 were requested. The OPS was still residing in the dry lab, door to the corridor left open.
Two examples are shown below:
Box 35 bottle 12 First 3 measurements: 34.9161, 34.9162, 34.9161
Box 35 bottle 35 First 3 measurements: 34.9129, 34.9130, 34.9130
All other measurements were of the same quality as these. All samples were immediately stable (first 3 measurements were within the allowed range). Maximum range within a set of three measurements of a sample was 0.0002 for two sample bottles. The others revealed readings with a range of 0.0001 or 0.0000.
Fig. 5: Micro bubbles are hardly detectable
Fig. 6: Results should resemble the quality described in the text
7. Further questions
The fact that stratification develops quickly inside a sample bottle if this is not at about the same temperature than its surrounding, establishes the question, whether a similar stratification process can develop inside a Niskin bottle. The ocean water is cold, while the ambient temperatures at the place where samples are taken is not far from standard room conditions. There might be a critical time interval for that. This problem should be evaluated further. A second concern is whether salinity analysis results differ between sample bottles filled via a tube (similar to taking oxygen samples to avoid evaporation) from a rinse through the air.
8. Conclusion
The Optimare Precision Salinometer OPS provides reference salinity measurements with unprecedented accuracy not only in the lab but also in a less than ideal environment as usually encountered on ship based expeditions. Automated sample processing allows to operate it by less trained persons without compromising the results. A precondition for ultimate accuracy is the correct handling of the ocean sample, which has to be tempered and equilibrated. The sample also needs to be roughly at the temperature of the room where it is being processed. Otherwise stratification in the sample bottle develops quickly (more investigation is needed on this subject). The room temperature has not to be identical with the salinometer’s bath temperature. Open questions remain particularly with respect to stratification in the Niskin bottles and the best practice to take samples. If difficulties arise concerning the stability of the OPS readings, one should consider the treatment and status of the water sample rather than direct attention to the instrument.