search

euroargodev / publicQCforum

A public forum to talk about Quality Control of Argo measurements
GNU General Public License v3.0
14 stars 1 forks source link

< example of sensor failure> Kistler pressure sensor issue #9

Open cabanesc opened 4 years ago

cabanesc commented 4 years ago

The Argo 3901931 float was deployed in 2017. It is an Arvor float equipped with an SBE41CP (8497) and a Kistler pressure sensor (4940374).

This float had received a MIN/MAX warning starting around cycle 105 with fresher salinity values (-0.35 psu at depth). After analysis, it appears that there is a problem with the pressure sensor, which reports higher pressure values than the actual values.

image Temperature (left) and Salinity (right) for float 3901931 in function of pressure. Cycles 99 to 112

image Theta/S diagram for float 3901931. Cycles 99 to 112

Pressure problem appears clearly on these three plots: note the large temperature shift in the thermocline, the salinity minimum shift from 850db to 1150db, the shift in the theta/S diagram with the profiles acquired after cycle 105 that are no longer parallel to the previous profiles. Pressure does not seem correctable with a simple pressure offset as the magnitude of the pressure error increased with depth.

Based on email exchanges between K. Martini and R. Cancouet, this is probably related to a defect detected in the Kistler pressure sensors in 2016 : a sudden shift in the pressure span (the calibration slope of pressure). The problem concerns CTDs built between January and July 2016, and this particular CTD was built in May 2016. The magnitude of the pressure span shift, is 1-30 %, pivoting at 0 pressure, and always one sign – resulting in higher reported pressure than actual pressure. Although the pressure sensors were checked during the routine manufacturing process for the defect, there was a small chance that the check would not catch all the affected sensors. More information can be found here: http://www.argo.ucsd.edu/Kistler_report_SBE.pdf

Because pressure is not correctable, we put this float into the grey list for PRES, PSAL, et TEMP with a flag 4 from cycle 104.

Have you found similar cases?

RomainCancouet commented 4 years ago

this is a MOCCA float

BSHbirgit commented 4 years ago

I will look for my files, I think I have seen something similar