An Intelligent Temperature-Conductivity Curve Test for Autonomous Ocean Mooring Applications

Abstract:

In coastal environments the continuous measurement of salinity is useful for detecting water mass movements, coastal circulation patterns, and freshwater input. However, in long-term instrument deployments some instrument conductivity cells can become obstructed with debris and marine growth. Such obstructions in water flow are a particular problem in nutrient-rich coastal waters. On a temperature-conductivity (TC) curve, these blocks are characterized by divergences from the linear relationship between temperature and salinity that defines a water mass. While such divergences may be identified through visual inspection over short time periods, the appearance of new water masses and conditions may confound attempts at parameterized quality control measures, leading to intensive manual protocols. This work applies statistical and machine learning-based techniques to analyze separations in TC space and identify conductivity blocks for quality control purposes. The results show that a relatively simple algorithm can automate and improve the flagging of conductivity cell blockages. This flagging approach is now being piloted within the Community Observatories program at Ocean Networks Canada.