An experiment was conducted to measure the drift in salinity of seawater samples kept in plastic and glass bottles stored in an air conditioned room for a period of 40 days. Assuming that the salinity of samples kept in glass bottles did not change significantly over this period of time, the salinity of samples kept in plastic bottles changed by approximately 1x10¯³ every 14 days. It is hypothesized that this change results from evaporation through the plastic bottles.
Twenty-four plastic bottles and twenty-four glass bottles were filled from a batch of homogenous sea water. Over a period of forty days the conductivity ratios of these samples were measured using the Autosal. Both the glass and plastic samples showed a drift over time; however, the machine was not standardized during this period, so we assume for simplicity that the glass samples were not actually drifting with time. Although the possibility that the glass samples drifted with time cannot be ruled out, the machine is expected to drift, and the conductivity of the glass samples over time provides a reasonable reference for estimating the drift in the plastic bottles.
Measurements were taken on seven separate occasions from October 4 to November 13, 1990. One of the measurements taken from plastic bottles on October 30 (approximately day 26) was believed to be bad and was consequently not included in the analysis. Because the bottles were too salty it seems likely that they were not properly capped or sealed.
Table 5.1 gives the average readings for each day for glass and plastic bottles, and the resulting differences in salinity (with standard error estimates). A regression fit to the differences gives 0.084x10¯³ psu day¯¹ or 0.0025 psu month¯¹.
Assuming that the seawater stored in glass bottles did not change in salinity over the 40- day period, the calculated drift in salinity differences is an estimate of the increase in salinity caused by evaporation in plastic bottles. For bottles stored more than a month this effect may be significant (more than 2 parts per million) when high accuracy is required.
The evaporation rate of 0.084x10¯³ psu day¯¹ was applied to the EQ-2 bottle data and compared to the previous COARE 156-1 and EQ-1 cruises salinities. The comparison indicated an over-correction to the EQ-2 salinities, showing fresher values than those in the other two cruises. A final correction of 0.05x10¯³ psu day¯¹ was applied to match the EQ-2 salinities to those from previous cruises. The difference in the evaporation rate in EQ-2 bottles compared to that obtained in the lab experiment may be attributed to the different storage conditions in both cases. The EQ-2 samples were stored in a closed container exposed under the sun whereas the lab samples were in an air conditioned room. Another factor in the discrepancy in evaporation rate may be that the experiment used samples stored for a maximum of 40 days, and the EQ-2 bottles were in storage for 100 days. Thus the extrapolation of the lab results to the EQ-2 salinities is approximate.