Temperature Bias Correction

For the detection of density inversions measured at independent instruments, it is important to account for any bias errors in the measurements. Averaged over the 3 months (8 months) of the experiment, temperatures show slight departures from a linear depth profile at mooring DS (DN) (Figure 2.3). Because the tidal vertical displacements are of the same magnitude as the height of both moorings, we assume that the sensors all sample water with similar properties, and so the departures from a smooth linear vertical profile are taken to be instrument bias errors. We therefore calculate corrections for each of the sensors so that the 3 months (8 months) averaged vertical profile is linear. The resulting linear profile (Figure 2.3) requires no more than a 5 millidegree shift in mean temperature for any given sensor. The manufacturer rating of the bias is 2 millidegrees.

For DS, we calculated the effects of the bias removal on the overturn detection and the estimated dissipation (Figure 5.3). After the bias removal, of the 41761 profiles analyzed, 14700 or 34 % contained at least one overturn $>24m$ . The correction applied to the temperature profile minimizes the detection of spurious overturns arising from sensor errors. For comparison, without the temperature correction, 23000 profiles (or 54 % of the total) contain at least one overturn $>24m$ , but most of these overturns are associated with small temperature differences approaching the accuracy of the sensors (Figure 5.3a). The temperature correction also increases the mean size of small overturns and decreases the size of large overturns (Figure 5.3b). The correction reduces the overall mean dissipation from $3.6\times10^{-8}$ to $1.2\times10^{-8} Wkg^{-1}$ , while the large overturns are responsible for a relatively larger percentage of the dissipation (Figure 5.3c). At mooring DN, the effect of the temperature bias is essentially the same as at DS (Figure 5.5). We note however, that the number of large energetic overturns is significantly lower at mooring DN than at mooring DS, consistent with a lower overall average dissipation (Table 5.1).