Observed features of salinity bias with negative pressure dependency for measurements by SBE 41CP and SBE 61 CTD sensors on deep profiling floats

Abstract

The salinity measured by SBE 41CP and SBE 61 conductivity-temperature-depth (CTD) sensors on deep floats was evaluated by comparison with shipboard CTD casts at deployment, after temperature and pressure were verified as unlikely to be biased. Salinity biases with a negative pressure dependency, expressed as ΔS = ΔSoffset + ap × pressure and ap < 0, were identified statistically even in the first profiles. For SBE 61 sensors, the average dependency coefficient ap was −0.54 × 10−6 dbar−1 and ΔSoffset was slightly positive (+0.8 × 10−3). The results for SBE 41CP sensors showed a more severe dependency: ap and ΔSoffset were −1.80 × 10−6 dbar−1 and −0.012 on average, respectively. The ap negativity was statistically significant for both sensors. For SBE 41CP, sensors with a fresher ΔSoffset tended to have a smaller ap. The fresh salinity bias found for SBE 41CP changed toward saline over time, mainly because ΔSoffset became less fresh with a decreasing rate. Stable ap for a long time was expected for the majority of sensors, but sometimes ap increased to a specific value within the first several profiles. Meanwhile, for SBE 61 sensors, these indicators seemed fairly stable for a longer period. Most Argo floats had the SBE 41/41CP CTD sensor, which was the original CTD model for deep floats. An analysis of data from 383 Argo floats did not identify a statistically significant pressure dependency for salinity. Conclusively, the SBE 41CP and SBE 61 CTD sensors on deep floats, in their present states, generally did not meet the target accuracy of Deep Argo for salinity due to the salinity bias. The present study suggests that both CTD sensors could almost achieve the target accuracy by the following improvements: aging treatment with high pressure on the sensor, accurate calibration of pressure-aged sensors, and a suitable canceling factor for pressure (CPcor; −12.6 × 10−8 dbar−1 for SBE 41CP and −11.0 × 10−8 dbar−1 for SBE 61). These operations can greatly reduce the salinity biases derived from the pressure dependency, the offset at the sea surface, and their changes over time.