1 of 15

Update for ADMT-21

Dr. Kim Martini

Dr. Jochen Klinke

December 3^rd, 2020

2 of 15

Outline

Update on Dynamic Corrections to SBE 41cps (5 min)

List of Dynamic Corrections applied
Field results compared to tank test

Long term temperature drift for Argo SBE 41 and SBE 41cp CTDs (5 min)

Update on Status of Salty Drift (10 min)

What Sea-Bird is doing to solve salty drift
Types of drift identified by Sea-Bird
What we've learned historically regarding drift

2

3 of 15

Methods for Determining CTD Dynamic Corrections

Tests in the stratified tank at WHOI

Accurate corrections for thermistor thermal mass and TC alignment, but artificial conditions are more extreme than the ocean.

3

Analysis of Argo float field data

New analysis on Solo-II floats profiling the upper 50 meters at 1 Hz guided by tank results.

5 °C and 15 PSU

change over 10 cm

2 °C and 0.1 PSU

change over 50 cm

4 of 15

Dynamic Corrections to Sea-Bird CTDs�SBE 41cp and SBE 61

There are three dynamic corrections to CTD data that can be made:

Thermistor thermal mass
Temperature and conductivity alignment
Conductivity cell thermal mass

Order of operations matter.

I am picky with data in order to not correct one source of error with another (i.e. Salinity spiking with cell thermal mass correction).

4

Modeled effects of dynamic errors on a step change in stratification

2 °C and 0.1 PSU

change over 50 cm

5 of 15

Results

Obtained similar corrections from tank and float data.

Cell thermal mass corrections similar to values obtained by John Lyman and Greg Johnson (PMEL) using a different method.

Next steps:

Changing the temperature and conductivity sampling sequence on the SBE 41cp to reduce alignment in post-processing.
Yank tests to test alignment at Sea-Bird
In situ alignment tests on R&D Navis float.

5

Citation	Data	Profiling Speed [m/s]
Johnson et al. 2007	ITP 1 Hz Profiles	0.3	0.39	0.05	6.68	0.141
Martini et al. 2019	Stratified Tank	0.1	0.16	-0.26	0.078	11.0
Johnson 2019	Stratified Tank	0.1	0.59	0.02	0.241	3.99
Martini et al. 2020	Stratified Tank	0.1	0.16	-0.26	0.090	9.16
Martini 2020	SOLO-II 1 Hz Profiles	0.1	0.16	-0.27

Dynamic corrections from the stratified tank applied to 1 Hz float data

6 of 15

Temperature Drift of 17 SBE41 & 41cp

Comparison of temperature drift in CTDs kept at the factory and deployed in the field.

Factory 41s have calibrations are averaged for each year.

Field 41s were calibrated ~6 years after pre-deployment calibrations.

Lifetime drift is within Argo program requirements for all but one example and it was within for ~5 years

6

Dots are SBE 41s that stayed in the factory

Squares are SBE 41s that went in the field

Thanks to Herve Claustre, Antoine Poteau and the rest of the team for retrieving these floats and returning the CTDs to Bellevue.

7 of 15

Work being done at Sea-Bird solve Salty Drift

Cross-functional team (science, engineering and calibration) to understand what causes salty drift and correct it.

What we have examined so far:

Calibration histories
Encapsulation material
Manufacturing process
Failure analysis of salty drifters returned to Bellevue
Historical analysis of Argo fleet performance

Biggest steps

Salty drift can be forced at Bellevue.
11 alternate cell constructions are built and are in progress of testing.

7

SCIENCE

Jochen Klinke, Kim Martini, Charlie Branham, Dave Murphy

ENGINEERING

Rob Ellison, Matt D’Asaro, Dave Walter, Matt Scanlon

CALIBRATION

Genevieve Howell, James Bauman

DBS FACILITATOR

Barry Lyon

8 of 15

Three Types of Salty Drift

Slow Salty Drifters

Core Argo SBE 41cp CTDs that slowly drift in a linear trend over long timescales
<0.01 PSU at ~70 profiles

Deep Salty Drifters

Deep Argo SBE 61 CTDs that slowly drift fresh, then rapidly drift salty
>0.05 PSU at ~60 profiles

Premature Salty Drifters

Core Argo SBE 41cp CTDs rapidly drift salty early in float lifetime
>0.1 PSU at ~20-40 profiles

8

There are three modes of salty drift, each with different characteristics and limits of correctability.

20

40

60

80

100

0.05

0.10

# OF PROFILES

∆S [PSU]

9 of 15

Slow Salty Drifters�Core Argo (>0.01 PSU at ~70 profiles)

Slow salty drift in Sea-Bird Moored CTDs is correctable and therefore correctable in SBE41cps.

9

Salty conductivity drift follows the linear model of drift that is well understood at Sea-Bird and is currently used by the DMQC teams.

Analysis of >20,000 Moored CTD calibrations over 21 years indicates slow salty drift as large as 0.1 PSU can be corrected if well-behaved, i.e. follows the linear model of conductivity drift.

The data is correctable.

There is no clear limit of correctability for slow salty drifters.

Isotherm

Pressure on

the isotherm

Salinity on that isotherm

Example of a slow salty drifter when tracking along an isotherm.

10 of 15

Deep Salty Drifters�Deep Argo (>0.1 PSU at ~20-40 profiles)

Deep salty drift does not follow a linear model of drift. We are working towards understanding the cause.

Analysis by the DMQC team shows that the in addition to drift, the correction shows a pressure dependence. While there is still work to be done for a final recommendation, we believe this kind of drift is not correctable.

Understanding the cause and making manufacturing changes to solve this is currently high priority research at Sea-Bird and will be described.

10

Deep salty drifters do not follow the well-understood linear models of drift. Correctability TBD.

Examples of SBE 61 deep salty drifters on Deep Argo Floats.

Image courtesy of Nathalie Zilberman.

11 of 15

Premature Salty Drifters�Core Argo (>0.1 PSU at ~20-40 profiles)

PMEL alerted us in February that several SBE 41cp Argo CTDs on Navis Floats are drifting prematurely in the field.

Premature salty drift affects SBE 41cps built over a period of 6 months in 2018…but are limited to individual batches.

Changes to manufacturing were implemented on October 3, 2018. Field tests and data indicates conductivity cells manufactured after this date are will not prematurely drift salty drift (SBE41cp Serial numbers 11250 and higher).

Sea-Bird has contacted customers and manufacturers regarding the affected CTDs.

SBE 61s Deep Argo CTDs are not affected.

11

Premature salty drifters do not follow the well-understood linear models of drift. Correctability TBD.

Example of a premature salty drifter on a Sea-Bird Navis float.

12 of 15

Historical Argo Fleet analysis

12

We realized from several lines of analysis that there have always been salty drifters.

Analysis of >20,000 Moored CTD calibrations over 21 years shows persistent salty drift, with increases where there have been changes in manufacturing.

Using the data set of salinity error against climatology, we did a similar analysis.

Comparison of fresh and salty drift in moored CTDs over time.

Salty Drift has always occurred in Moored CTDs

SBE 37 Moored CTDs

13 of 15

Truncated ARGO Data: GPS Position

13

All Float Data

Down Sampled Data

1251 Floats

13539 Floats

90.76% reduction of data

ARGO data truncated to only contain reliable drift data.

Historical analysis is constrained to a “boring” part of the ocean, the Pacific Ocean.

14 of 15

Truncated ARGO Data: Slope Evaluation

Eigenvalue analysis of drift indicates most drift is linear.

Drift can be estimated by fitting a line to the salinity error. Here the slope of the line equals the drift.

Percent with Negative Slope: 22.1% (Freshy)
Percent with Positive Slope: 77.9% (Salty)

There are cases with non-linear drift, but a similar result is obtained (75%) if we omit them.

14

Slope (aka drift) is determined by linearly fitting the salinity error estimated from CSIRO ARGO climatology against Profile Number.

There have always been salty drifters.

15 of 15

Final thoughts…

Planning field tests or deployments? Let Sea-Bird know and how we can help.

Questions?

15