Qartod Temperature Salinity Manual

User Manual:

Open the PDF directly: View PDF .
Page Count: 56

Download
Open PDF In Browser	View PDF

Manual for

Real-Time Quality Control of
In-situ Temperature and
Salinity Data
A Guide to Quality Control and Quality Assurance
for In-situ Temperature and Salinity Observations

Version 2.0
January 2016

Document Validation

U.S. IOOS® Program Office Validation
Zdenka S. Willis, Director, U.S. IOOS Program Office

01/06/2016
Date

QARTOD Project Manager Validation
Joseph Swaykos, NOAA National Data Buoy Center

01/06/2016
Date

QARTOD Board of Advisors Validation

Julianna O. Thomas, Southern California Coastal Ocean Observing System

01/06/2016
Date

Table of Contents
Document Validation ....................................................................................... ii
Table of Contents ............................................................................................ iii
List of Figures .................................................................................................. iv
List of Tables .................................................................................................... iv
Revision History ................................................................................................ v
Endorsement Disclaimer ................................................................................. vi
Acknowledgements ......................................................................................... vii
Acronyms and Abbreviations .........................................................................viii
Definitions of Selected Terms........................................................................... x
1.0 Background and Introduction ................................................................. 1
2.0 Purpose/Constraints/Applications ......................................................... 3
2.1
Data Processing Methodology...................................................................................................................8
2.2
Traceability to Accepted Standards ..........................................................................................................8
2.3
Sensor Deployment Considerations .........................................................................................................8
2.3.1 Fixed Platform and Fixed Vertical Profilers .................................................................................... 11
2.3.2 Mobile Surface Vessels ........................................................................................................................ 11
2.3.3 3-D Profiler Vessels ............................................................................................................................. 11
2.4
Hardware Limitations .............................................................................................................................. 11

3.0

Quality Control ....................................................................................... 13

3.1
QC Flags .................................................................................................................................................... 13
3.2
Test Hierarchy ........................................................................................................................................... 14
3.3
QC Test Descriptions .............................................................................................................................. 15
3.3.1 Applications of QC Tests to Stationary TS Sensors ...................................................................... 15
Test 1) Timing/Gap Test (Required) ............................................................................................................. 15
Test 2) Syntax Test (Required) ........................................................................................................................ 16
Test 3) Location Test (Required) .................................................................................................................... 16
Test 4) Gross Range Test (Required) ............................................................................................................. 17
Test 5) Climatology Test (Required)............................................................................................................... 17
Test 6) Spike Test (Strongly Recommended) ................................................................................................ 18
Test 7) Rate of Change Test (Strongly Recommended) .............................................................................. 18
Test 8) Flat Line Test (Strongly Recommended) ......................................................................................... 19
Test 9) Multi-Variate Test (Suggested) ........................................................................................................... 20
Test 10) Attenuated Signal Test (Suggested) ................................................................................................. 21
Test 11) Neighbor Test (Suggested) ............................................................................................................... 21
Test 12) TS Curve/Space Test (Suggested) ................................................................................................... 22
Test 13) Density Inversion Test (Suggested) ................................................................................................ 22
3.3.2 Applications of QC Tests to Mobile TS Sensors ............................................................................ 23

4.0
5.0

Summary................................................................................................. 31
References .............................................................................................. 32

iii

Appendix A. Quality Assurance (QA)........................................................ A-1
A.1
Sensor Calibration Considerations....................................................................................................... A-1
A.2
Sensor Comparison ................................................................................................................................ A-1
A.3
Bio-fouling and Corrosion Prevention Strategies.............................................................................. A-2
A.4
Common QA Considerations............................................................................................................... A-3
A.5
QA Levels for Best Practices ................................................................................................................ A-4
A.6
Additional Sources of QA Information .............................................................................................. A-4
A.7
Example Deployment Checklists ......................................................................................................... A-5
Pre-deployment QA Checklist....................................................................................................................... A-5
Deployment Checklist..................................................................................................................................... A-5
Post-deployment Checklist ............................................................................................................................ A-6

Appendix B. QARTOD TS Manual Team ................................................ B-1

List of Figures
Figure 2-1. A profiling Sea-Bird SBE 9plus CTD mounted on a rosette with Niskin bottles is recovered
during a cruise aboard the RV OCEAN VERITAS following the 2010 Deepwater Horizon
incident (photo courtesy of Mark Bushnell). .....................................................................................5
Figure 2-2. This Sea-Bird Electronics SBE 37-IM temperature and conductivity recorder uses an inductive
modem to transmit data up the mooring cable to a surface receiver (photo courtesy of Rick
Cole, RDSEA International, Inc.). ......................................................................................................5
Figure 2-3. This towed RBR concerto CTD uses an inductive sensor with an external field and no pump
(photo courtesy of Igor Shkvorets/RBR Ltd.)..................................................................................6
Figure 2-4. The Teledyne RD Instruments CITADEL CTD-ES is an example of an inductive sensor with
an external field. Operators must be certain that additional hardware is sufficiently distant
from the toroid to avoid interference. This sensor is designed for simple and easy cleaning
(photo courtesy of Paul Devine/Teledyne RD Instruments).........................................................6
Figure 2-5. The Teledyne RD Instruments CITADEL CTD-NH is an example of an inductive sensor with
a constrained field (photo courtesy of Paul Devine/Teledyne RD Instruments). ......................7
Figure 2-6. The JFE Advantech INFINITY-CT A7CT-USB (photo courtesy of Fabian Wolk,
Ph.D./Rockland Scientific International Inc.). .................................................................................7
Figure 2-7. Slocum Glider Profiler 3-D (L) and Liquid Robotics Wave Glider (R) (photo courtesy of Dave
Fratantoni, Ph.D./NortekUSA). .........................................................................................................9
Figure 2-8. WET Labs AMP C100 In-Situ Profiler (courtesy of WET Labs) (L); RBR CTD sensor on an
Oceaneering ROV (R) (photo courtesy of Igor Shkovorets/RBR Ltd.). .....................................9
Figure 2-9. This CTD/bottle rosette shows the use of both Sea-Bird (SBE 9plus) and RBR sensors (photo
courtesy of Igor Shkvorets/RBR Ltd.). ........................................................................................... 10

List of Tables
Table 2-1. TS sensor manufacturers ...........................................................................................................................4
Table 3-1. Flags for real-time data (UNESCO 2013)............................................................................................ 14
Table 3-2. QC Tests in order of implementation and hierarchy. ........................................................................ 14
Table 3-3. Comparison of QARTOD, GTSPP, and Argo temperature and salinity QC tests ...................... 23
Table 3-4. Application of Required QC Tests (Tests 1-5) for TS Sensor Deployments................................. 24
Table 3-5. Application of Strongly Recommended QC Tests (Tests 6-8) for TS Sensor Deployments ...... 26
Table 3-6. Application of Suggested QC Tests (Tests 9-13) for TS Sensor Deployments ............................. 28

Revision History
Date

Revision Description

12/2013
01/2016

Original Document Published.
Revise cover page to include new IOOS logo.
Revise dates on Document Validation page and substitute new logo (page ii).
Add statement requesting feedback from Manual Users (page vi).

Notes

Update Acknowledgements to include Version 2.0 team members (page vii).
Update Acronyms and Abbreviations (page viii).
Update definition of real time in Definitions of Selected Terms (page x).
Revise Background and Introduction to reflect updated temperature/salinity
manual, as well as additional manuals that have been developed (page 1).
Revise section 2.0 content in various places to reflect feedback from
reviewers (pages 3-4).
Add section 2.5 with content addressing data uncertainty (page 12).
Update content in sections 3.1 and 3.2 (pages 13-14).
Update language in section 4.0, Summary (page 31).
Update References and Supporting Documents (page 32-35).
Update Temperature and Salinity Manual Team members (page B-1).

Manual
updated
with
revisions
listed
sequentially

Endorsement Disclaimer
Mention of a commercial company or product does not constitute an
endorsement by NOAA. Use of information from this publication for
publicity or advertising purposes concerning proprietary products or
the tests of such products is not authorized.

Request to Manual Users
To gauge the success of the QARTOD project, it helps to be aware
of groups working to utilize these QC tests. Please notify us of your
efforts or intentions to implement QARTOD processes by sending a
brief email to data.ioos@noaa.gov or posting a notice at
http://www.linkedin.com/groups?gid=2521409.

Acknowledgements
Special thanks go to members of the Temperature/Salinity Manual committee, who contributed their expertise
to develop the content of this manual and also to document reviewers, whose many valuable suggestions
greatly enhanced the manual content.
Specifically, thanks to Igor Shkvorets of RBR Ltd. for his review and his expertise regarding standard seawater
and instrument calibration. We appreciate the additional photographs and thoughtful review provided by
Stephanie Jaeger at Sea-Bird Electronics.

vii

Acronyms and Abbreviations
AOML

(NOAA) Atlantic Oceanographic and Meteorological Laboratory

AOOS

Alaska Ocean Observing System

AUV

Autonomous Underwater Vehicle

CDIP

Coastal Data Information Program

CeNCOOS

Central and Northern California Ocean Observing System

CO-OPS

(NOAA) Center for Operational Oceanographic Products and Services

CRC

Cyclic Redundancy Check

CTD

Conductivity, Temperature, and Depth

EuroGOOS

European Global Ocean Observing System

GCOOS

Gulf of Mexico Coastal Ocean Observing System

GLOS

Great Lakes Observing System

GOOS

Global Ocean Observing System

GPS

Global Positioning System

GTSPP

Global Temperature-Salinity Profile Program

IAPSO

International Association for the Physical Sciences of the Oceans

IMOS

Integrated Marine Observing System

IODE

International Oceanographic Data and Information Exchange

IOOS

(U.S.) Integrated Ocean Observing System

MARACOOS

Mid-Atlantic Regional Association Coastal Ocean Observing System

NANOOS

Northwest Association of Networked Ocean Observing Systems

NCEI

(NOAA) National Centers for Environmental Information (formerly NODC)

NDBC

(NOAA) National Data Buoy Center

NERACOOS

Northeastern Regional Association of Coastal Ocean Observing Systems

NIST

National Institute of Standards and Technology

NOAA

National Oceanic and Atmospheric Administration

NODC

(NOAA) National Oceanographic Data Center (now NCEI)

Pressure

PacIOOS

Pacific Islands Ocean Observing System

PMEL

Pacific Marine Environmental Laboratory

PSS-78

Practical Salinity Scale-1978

QARTOD

Quality-Assurance/Quality Control of Real-Time Oceanographic Data

Quality Assurance

Quality Control

viii

RCOOS

Regional Coastal Ocean Observing System

SCCOOS

Southern California Coastal Ocean Observing System

Standard Deviation

SECOORA

Southeast Coastal Ocean Observing Regional Association

SIO

Scripps Institution of Oceanography

Practical Salinity

Temperature/Salinity

UNESCO

United Nations Educational, Scientific and Cultural Organization

USGS

United States Geological Survey

Definitions of Selected Terms
This manual contains several terms whose meanings are critical to those using the manual. These terms are
included in the following table to ensure that the meanings are clearly defined.
Codable
Instructions

Codable instructions are specific guidance that can be used by a software
programmer to design, construct, and implement a test. These instructions also
include examples with sample thresholds.

Data Record

A data record is one or more messages that form a coherent, logical, and complete
observation.

Message

A message is a standalone data transmission. A data record can be composed of
multiple messages.

Operator

Operators are individuals or entities who are responsible for collecting and
providing data.

Practical
Salinity (SP)

A unitless ratio expressing salinity as defined by the Practical Salinity Scale 1978
(PSS-78).

Quality
Assurance
(QA)

QA involves processes that are employed with hardware to support the generation
of high quality data. (section 2.0 and appendix A)

Quality Control
(QC)

QC involves follow-on steps that support the delivery of high quality data and
requires both automation and human intervention. (section 3.0)

Real Time

Real time means that: data are delivered without delay for immediate use; time series
extends only backwards in time, where the next data points are not available; and
sample intervals may range from a few seconds to a few hours or even days,
depending upon the sensor configuration. (section 1.0)

Threshold

Thresholds are limits that are defined by the operator.

Temperature and Salinity

1.0 Background and Introduction
The U.S. Integrated Ocean Observing System (IOOS®) has a vested interest in collecting high-quality data for
the 26 core variables (U.S. IOOS 2010) measured on a national scale. In response to this interest, U.S. IOOS
continues to establish written, authoritative procedures for the quality control (QC) of real-time data through
the Quality Assurance/Quality Control of Real-Time Oceanographic Data (QARTOD) project, addressing each
variable as funding permits. This manual on the real-time QC of temperature and salinity data was first
published in December 2013 as the fourth core variable to be addressed. It is the fourth manual to be updated.
Other QARTOD guidance documents that have been published by the U.S. IOOS project to date are listed
below and are available at http://www.ioos.noaa.gov/qartod/.
1) U.S IOOS QARTOD Project Plan dated April 1, 2012.
2) U.S. Integrated Ocean Observing System, 2015. Manual for Real-Time
Quality Control of Dissolved Oxygen Observations Version 2.0: A Guide to
Quality Control and Quality Assurance for Dissolved Oxygen Observations
in Coastal Oceans. 48 pp.
3) U.S. Integrated Ocean Observing System, 2015. Manual for Real-Time
Quality Control of In-Situ Current Observations Version 2.0: A Guide to
Quality Control and Quality Assurance of Acoustic Doppler Current Profiler
Observations. 51 pp.
4) U.S. Integrated Ocean Observing System, 2015. Manual for Real-Time
Quality Control of In-Situ Surface Wave Data Version 2.0: A Guide to
Quality Control and Quality Assurance of In-Situ Surface Wave
Observations. 64 pp.
5) U.S. Integrated Ocean Observing System, 2014. Manual for Real-Time
Quality Control of Water Level Data: A Guide to Quality Control and Quality
Assurance of Water Level Observations. 43 pp.
6) U.S. Integrated Ocean Observing System, 2014. Manual for Real-Time
Quality Control of Wind Data: A Guide to Quality Control and Quality
Assurance of Coastal and Oceanic Wind Observations. 45 pp.
7) U.S. Integrated Ocean Observing System, 2015. Manual for Real-Time Quality
Control of Ocean Optics Data: A Guide to Quality Control and Quality Assurance
of Coastal and Oceanic Optics Observations. 46 pp.
8) U.S. Integrated Ocean Observing System, 2015. Manual for Real-Time
Quality Control of Dissolved Nutrients Data: A Guide to Quality Control and
Quality Assurance of Coastal and Dissolved Nutrients Observations. 56 pp.

Please reference this document as:
U.S. Integrated Ocean Observing System, 2015. Manual for Real-Time Quality Control
of In-situ Temperature and Salinity Data Version 2.0: A Guide to Quality Control and
Quality Assurance of In-situ Temperature and Salinity Observations. 56 pp.
This manual is a living document that reflects the state-of-the-art QC testing procedures for temperature and
salinity observations. It is written for the experienced operator but also provides examples for those who are
just entering the field.

Temperature and Salinity

2.0 Purpose/Constraints/Applications
This manual documents a series of test procedures for quality control (QC) of temperature and salinity (TS) data.
TS observations covered by these procedures are collected in oceans, coastal waters, and lakes in real time. The
scope of real time has expanded to accommodate the span of the 26 variables covered by U.S. IOOS. The
characteristics of real time (in no particular order) are:
 data delivered as soon as possible after acquisition for immediate use
 a time series extending only backwards in time, where the next data point is not available
 sample intervals from a few seconds to a few hours or even days, depending upon the sensor
configuration
The tests draw from existing expertise in programs such as the Global Temperature and Salinity Profile
Programme (GTSPP) and Argo. The Global Climate Observing System (GCOS) recognizes the GTSPP as one
of the international operational activities that provide essential, sub-surface climate variables of temperature and
salinity profile data. GTSPP provides timely and complete data with documented quality flags and implements
internationally accepted quality control and overall management of ocean data fully in accordance with the
GCOS action plan (www.nodc.noaa.gov/GTSPP/). The Argo program is a global array of 3,000 free-drifting
profiling floats measuring the temperature and salinity of the upper 2,000 meters (m) of the ocean. The program
provides continuous monitoring of the temperature, salinity, and velocity of the upper ocean, with all data being
relayed and made publicly available within hours after collection (www.argo.net).
This manual differs from existing QC procedures for TS data in that its focus is on real time, and it is not
constrained to deep oceans, as are GTSPP and Argo. It presents practices and procedures from these programs
as a basis for developing codable instructions and provides guidance for the broader ocean observing
community. These existing programs and others within the observing community use many of the same sensors.
The tests and codable instructions described herein are examples that might be employed. But, operators may
choose to use similar tests from existing programs (such as the MATLAB®-coded QC tests posted by the
Integrated Marine Observing System [IMOS] at https://github.com/aodn/imos-toolbox) or to develop their
own tests to accomplish the same results.
High-quality marine observations require sustained quality assurance (QA) and QC practices to ensure credibility
and value to operators and data users. Some QA practices involve processes that are employed with hardware to
support the generation of high-quality data, such as a sufficiently accurate, precise, and reliable sensor with
adequate resolution. Others include: sensor calibration; calibration checks and/or in-situ verification, including
post-deployment calibration; proper deployment considerations, such as measures for corrosion control and
anti-fouling; reliable data communications; adequate maintenance intervals; and creation of a robust QC process.
Post-deployment calibration (instrument verification after recovery) issues are not part of the scope of this
manual. Although QC and QA are interrelated and important to the process, QA issues are briefly addressed
separately in appendix A.
QC involves follow-on steps that support the delivery of high-quality data and requires both automation and
human intervention. QC practices include such things as data integrity checks (format, checksum, timely
arrival of data), data value checks (threshold checks, minimum/maximum rate of change), neighbor checks,
climatology checks, model comparisons, signal/noise ratios, the mark-up of the data, the verification of user
satisfaction, and generation of data flags (Bushnell 2005).

These procedures are written as a high-level narrative from which a computer programmer can develop code to
execute specific data flags (data quality indicators) within an automated software program. A code repository
exists at https://github.com/ioos/qartod, where operators may find or post examples of code in use. Although
certain tests are recommended, thresholds can vary among operators. The tests described here are designed to
support a range of TS sensors and operator capabilities. Some well-established programs with the highest
standards have implemented very rigorous QC processes. Others, with different requirements, may utilize
sensors with data streams that cannot support as many QC checks—all have value when used prudently. Users
must understand and appropriately utilize data of varying quality, and operators must provide support by
documenting and publishing their QC processes. A balance must be struck between the time-sensitive needs of
real-time observing systems and the degree of rigor that has been applied to non-real-time systems by operators
with decades of QC experience.
These tests apply only to the in-situ, real-time measurement of TS as observed by sensors deployed on rigidly
mounted, moored, or moving platforms (e.g., drifting buoys, autonomous marine vehicles, ships) but not to
remotely sensed TS measurements (e.g., satellite observations).
The process of ensuring data quality is not always straightforward. QA/QC procedures may be specific to a
sensor technology or even to a particular manufacturer’s model, so the establishment of a methodology that is
applicable to every sensor remains challenging.
Sensors deployed on mobile platforms such as gliders require attention to proper QA procedures both before
and after the deployment (appendix A provides general QA guidance). While outside the scope of the realtime tests described in this manual, the manufacturer’s recommendations for factory calibration schedules
and proper sensor maintenance must be followed.
The instruments described in figs. 2-1 through 2-6 are illustrations provided by manufacturers and TS
committee members. They may be referred to as TS (temperature and salinity), CTD (conductivity,
temperature and depth) or CT sensors (conductivity and temperature), and they directly measure T, C, and P
(pressure). Their measurements are used to derive salinity, depth, density, specific gravity, and specific
conductance. Table 2-1 lists companies that produce sensors covered in this manual.
Table 2-1. TS sensor manufacturers

Aanderaa
Campbell Scientific
Greenspan
Hach
In-Situ
JFE Advantech Company Ltd.
RBR Ltd.
Rockland Scientific International Inc.
Sea-Bird Electronics, Inc.
Teledyne RD Instruments
YSI

Temperature and Salinity

Figure 2-1. A profiling Sea-Bird SBE 9plus CTD mounted on a rosette with Niskin bottles is recovered
during a cruise aboard the RV OCEAN VERITAS following the 2010 Deepwater Horizon incident
(photo courtesy of Mark Bushnell).

Figure 2-2. This Sea-Bird Electronics SBE 37-IM temperature and conductivity recorder uses an inductive
modem to transmit data up the mooring cable to a surface receiver (photo courtesy of Rick Cole, RDSEA
International, Inc.).

Figure 2-3. This towed RBR concerto CTD uses an inductive sensor with an external field and no pump
(photo courtesy of Igor Shkvorets/RBR Ltd.).

Figure 2-4. The Teledyne RD Instruments CITADEL CTD-ES is an example of an
inductive sensor with an external field. Operators must be certain that additional hardware
is sufficiently distant from the toroid to avoid interference. This sensor is designed for
simple and easy cleaning (photo courtesy of Paul Devine/Teledyne RD Instruments).

Temperature and Salinity

Figure 2-5. The Teledyne RD Instruments CITADEL CTD-NH is an example
of an inductive sensor with a constrained field (photo courtesy of Paul
Devine/Teledyne RD Instruments).

Figure 2-6. The JFE Advantech INFINITY-CT A7CT-USB (photo courtesy
of Fabian Wolk, Ph.D./Rockland Scientific International Inc.).

2.1

Data Processing Methodology

The type of sensor system collecting the data and the system processing and transmitting the measurements
can affect which QC algorithms are used. In-situ systems with sufficient on-board processing power within
the sensor may process the original (raw) data and produce derived products, such as salinity, density, or
speed of sound. If ample bandwidth is available, the entire original data stream may be transmitted ashore and
subsequently quality controlled. If lacking sufficient bandwidth, the operator may not be able to apply tests
designed for raw data. Therefore, because operators have different data processing methodologies, three
levels of QC are proposed: required, strongly recommended, and suggested.

2.2

Traceability to Accepted Standards

To ensure that TS sensors are producing accurate data, rigorous calibrations and calibration checks must be
performed in addition to QC checks. Most operators rely upon manufacturer calibrations and conduct
calibration checks only before deployment. These calibration checks are critical to ensuring that the
manufacturer calibration is still valid. These procedures are currently considered QA and addressed further in
appendix A.
Calibrations and calibration checks must be traceable to accepted standards. The National Institute of
Standards and Technology (NIST) (http://www.nist.gov/index.html), a provider of internationally accepted
standards, is often the source for these standards. Calibration activities must be tailored to match data use and
resources; calibration cost and effort increase dramatically as accuracy requirements increase. NIST standards
for temperature and pressure sensors can be met using transfer references such as platinum resistance
thermometers and deadweight testers. Conductivity sensors are most commonly calibrated against the
International Association of Physical Sciences of the Ocean (IAPSO) standard seawater, certified by Ocean
Scientific International Ltd. (OSIL) in terms of the ratio K15. The ocean observing community uses
dimensionless practical salinity as defined by the Practical Salinity Scale-1978 (PSS-78), developed in 1978
(UNESCO 1981). PSS-78 is based on an equation relating salinity to the ratio K15 of the electrical
conductivity of seawater at 15 °C to that of a standard potassium chloride solution (KCl)
(http://salinometry.com/pss-78). Laboratory salinometers (http://salinometry.com/modern-oceanographicsalinometers) are used for the precise measurement of salinity samples during laboratory conductivity
calibrations and bottle samples at sea. A new absolute salinity scale was adopted in 2009 by the Scientific
Committee on Oceanic Research and the IAPSO Working Group 127 (WG127) (McDougall et al., 2009).
However, WG127 has advised the continued use of the PSS-78 for data repositories.

2.3

Sensor Deployment Considerations

TS sensors can be deployed in several ways. Stationary sensor deployments are on fixed platforms or
moorings where there is minimal horizontal or vertical movement. Mobile platforms are available in a variety
of configurations and require different real-time TS QC considerations. Mobile platforms are, in order of
increasing complexity: fixed vertical profilers, mobile surface vessels, and vessels freely operating in three
dimensions (e.g., gliders, floats, powered automated underwater vehicles or AUVs). Figures 2-7 through 2-9
illustrate examples.

Temperature and Salinity

Figure 2-7. Slocum Glider Profiler 3-D (L) and Liquid Robotics Wave Glider® (R) (photo courtesy of Dave Fratantoni,
Ph.D./NortekUSA).

Figure 2-8. WET Labs AMP C100 In-Situ Profiler (courtesy of WET Labs) (L); RBR CTD sensor on an Oceaneering ROV (R)
(photo courtesy of Igor Shkovorets/RBR Ltd.).

Figure 2-9. This CTD/bottle rosette shows the use of both Sea-Bird (SBE 9plus) and RBR sensors (photo courtesy of Igor
Shkvorets/RBR Ltd.).

Moving Platform Corrections
Mobile and profiling sensors commonly move through gradients over short time scales and require additional
QC. Therefore, two additional corrections specifically for mobile and profiling sensors should be applied prior
to the real-time QC tests described in this manual: a response time correction and a thermal mass correction. The
methods employed to make these corrections are usually developed and provided by the manufacturer, since they
are unique to each specific sensor and may even require calibration factors. The following discussion is an
overview of the complexity associated with obtaining CTD data of high accuracy but is not meant to instruct or
guide operators on these correction processes.

Response Time Correction. The first correction is made because the CT sensors on the instrument have
different measurement response times and may have different physical locations; thus, the two independent
measurements should be aligned with respect to time so that each CTD record represents a measurement on a
single parcel of water. This time shift should account for the sample rate of the instrument and for the known
constant flow rate of the pump on the CTD sensor (if pumped) or the speed of the glider through the water
column (if unpumped) (Garau et al. 2011).

Thermal Mass Correction. A second correction is needed to account for the thermal mass of the conductivity
cell and its effect on the resulting salinity calculation. The CTD sensor temperature is measured outside the
conductivity cell, while the conductivity is measured inside the cell. In addition, the conductivity cell can store
heat from the surrounding water inside the wall of the cell, resulting in a heating or cooling of new water parcels
as they pass through the cell. As a result of this thermal lag, without the corrections, the paired conductivity and

Temperature and Salinity

temperature used to calculate salinity could result in erroneous salinity values, especially across temperature
gradients. A method to correct for heating inside the cell has been developed, resulting in more accurate salinity
profiles (Morison et al. 1994). Garau et al. (2011) specifically address the additional considerations associated with
unpumped CTD sensors deployed on gliders.

2.3.1

Fixed Platform and Fixed Vertical Profilers

Fixed vertical TS profiles are obtained from a variety of systems, including rigid-mounted systems,
buoy/mooring climbers, surface- or bottom-tethered systems, or even casts from regularly scheduled manual
station observations. Tests described for a fixed sensor (section 3.3) either remain unchanged or are
conducted along the vertical (‘z’) axis, as well as along a time series of observations.

2.3.2

Mobile Surface Vessels

Examples of mobile surface vessels include manned vessels of opportunity and autonomously operated
vehicles, such as wave gliders, fitted with TS sensors. Samples are obtained at a fixed depth along a track and
may be taken at fixed temporal or spatial intervals. Tests may be conducted along the vessel path (‘s’), or the
path may be projected along ‘x’ (longitude) and ‘y’ (latitude) coordinates, as well as along a time series of
observations.

2.3.3

3-D Profiler Vessels

Sensors mounted on gliders, floats, powered AUVs, and animals can provide TS observations in a wide
variety of space/time configurations. Observations can be as simple as along path ‘s’, periodic vertical ascent
profiles recorded following at-depth drifts (Argo profilers), or real-time processed down/up profiles (gliders).
When applying increasingly complex real-time QC tests to increasingly complex deployments, challenges may
arise. However, most of the 13 tests described in section 3.3 can be applied with little modification.

2.4

Hardware Limitations

Most temperature and pressure sensors can withstand moderate bio-fouling. However, conductivity sensors
cannot, so salinity observational accuracy gradually degrades with time. Because the performance decline is
gradual or can occur as part of an event, it is difficult to detect and usually is not noticed until the fouled
sensor is replaced. Fouling most often leads to lower conductivity/salinity readings. For more information on
QA related to bio-fouling, see appendix A.
Advances in TS measurement technology have eliminated many of the problems encountered in older
devices. Sensors are smarter, smaller, more reliable, and draw less power. More sensors can be employed and
used for comparison to make corrections. Most notably, signal processing hardware and software capabilities
have grown substantially. For example, sensor response is more easily digitally characterized and calibrated, as
opposed to constructing a physical device with a known response.

2.5

Other Important Considerations

While outside the scope of the real-time tests described in this manual, quality assurance (QA) is critical to
data quality. Sensors require attention to proper QA measures both before and after the deployment
(appendix A). Operators must follow the manufacturer’s recommendations for factory calibration schedules
and proper sensor preparation and maintenance.

Also important, but beyond the scope of this document at present, is the determination and reporting of data
uncertainty. Knowledge of the accuracy of each observation is required to ensure that data are used
appropriately and aids in the computation of error bounds for subsequent products derived by users. All
sensors and measurements contain errors that are determined by hardware quality, calibration accuracy,
methods of operation, and data processing techniques. Operators should routinely provide a quantitative
measure of data uncertainty in the associated metadata. Such calculations can be challenging, so operators
should also document the methods used to compute the uncertainty. The limits and thresholds implemented
by operators for the data QC tests described here are a key component in establishing the observational error
bounds. Operators are strongly encouraged to consider the impact of the QC tests on data uncertainty, as
these two efforts greatly enhance the utility of their data.
Sensor redundancy is key to obtaining reliable measurements and ensuring that uncertainties can be assigned
to those measurements. Comparing two adjacent instruments can assist in evaluation of data quality, as well
as provide two (or more) independent estimates of a parameter of interest. Variation in the estimates of
uncertainty provided by those instruments can occur for several reasons, including water mass gradients in
the environment.

Temperature and Salinity

3.0 Quality Control
To conduct real-time QC on TS observations, the first pre-requisite is to understand the science and context
within which the measurements are being conducted. Each deployment method imposes the need for specific
QC methods, with different interpretations of ‘real time.’ A fixed TS sensor deployed in coastal waters may
report at 5-minute intervals, while deep ocean CTD casts may take hours to provide a profile. While each
sensor provides vastly different products, QC techniques can be applied broadly; with the proper selection of
thresholds, a check for temporal data spikes in the former is similar to data spike checks in the vertical profile
of the latter.
TS measurements can be used to resolve many things, such as internal waves, oceanic fronts, river runoff,
upwelling, etc., and some of these can be extreme events. Human involvement is therefore important to
ensure that solid scientific principles are applied to data evaluation to ensure that good data are not discarded
and bad data are not distributed.
The real-time QC of TS observations can be extremely challenging. For example, for real-time QC, gradual
calibration changes and long-term system responses (sensor drift) most likely cannot be detected or corrected
with real-time, automated QC. Drift correction for TS measurements during post-processing is difficult even
if a valid post-recovery calibration is obtained. Drift is often caused by bio-fouling, affecting different systems
in different ways—a sensor’s response will be affected by the added mass of bio-fouling. Another example is
the ability of some data providers to backfill data gaps. In both of these examples, the observations are not
considered to be real time for purposes of QC checks.

3.1

QC Flags

Data are evaluated using QC tests, and the results of those tests are recorded by inserting flags in the data
record. Table 3-1 provides a simple set of flags and associated descriptions. Operators may incorporate
additional flags for inclusion in metadata records to further assist with troubleshooting. For example, an
observation may fail the temperature min/max range test and be flagged as having failed. An operator could
provide an additional test to further define a failure: if the data failed the temperature min/max by exceeding
the upper limit, a “failed high” flag could indicate that the values were higher than the expected range. Such
detailed flags primarily support maintenance efforts and are presently beyond U.S. IOOS requirements for
QC of real-time data. For additional information regarding flags, see the Manual for the Use of Real-Time
Oceanographic Data Quality Control Flags (U.S. IOOS 2014) posted on the U.S. IOOS QARTOD website.
Further post-processing of the data may yield different conclusions from those reached during initial
assessments. Flags set in real time should not be changed to ensure that historical documentation is
preserved. Results from post-processing should generate another set of flags.
Observations are time ordered, and the most recent observation is n0, preceded by a value at n-1, and so on
moving back in time. The focus is primarily on the real-time QC of observations n0, n-1, and n-2.

Table 3-1. Flags for real-time data (UNESCO 2013)

Flag
Pass=1

Not evaluated=2

Description
Data have passed critical real-time quality control tests and are deemed adequate for
use as preliminary data.
Data have not been QC-tested, or the information on quality is not available.

Suspect or
Of High Interest=3

Data are considered to be either suspect or of high interest to data providers and users.
They are flagged suspect to draw further attention to them by operators.

Fail=4

Data are considered to have failed one or more critical real-time QC checks. If they are
disseminated at all, it should be readily apparent that they are not of acceptable quality.

Missing data=9

3.2

Data are missing; used as a placeholder.

Test Hierarchy

This section outlines the 13 real-time QC tests that are required, strongly recommended, or suggested for
real-time TS measurements. Salinity may be computed onboard the sensor package or after transmission of
the raw data. When possible, tests should be applied to conductivity and temperature observations, as well as
the derived salinity values, regardless of where the salinity calculation takes place. Operators should also
consider that some of these tests can be carried out within the instrument, where thresholds can be defined in
configuration files. Although more tests may imply a more robust QC effort, there are many reasons
operators could use to justify not conducting some tests. In those cases, operators need only to document
reasons these tests do not apply to their observations. Tests are listed in table 3-2 and are divided into three
groups: those that are required, strongly recommended, or suggested.
Table 3-2. QC Tests in order of implementation and hierarchy.

Group 1
Required

Test 1)
Test 2)
Test 3)
Test 4)
Test 5)

Gap Test
Syntax Test
Location Test
Gross Range Test
Climatological Test

Group 2
Strongly
Recommended

Test 6)
Test 7)
Test 8)

Spike Test
Rate of Change Test
Flat Line Test

Group 3
Suggested

Test 9)
Test 10)
Test 11)
Test 12)
Test 13)

Multi-Variate Test
Attenuated Signal Test
Neighbor Test
TS Curve/Space Test
Density Inversion Test

Temperature and Salinity

3.3

QC Test Descriptions

A variety of tests can be performed to evaluate data quality in real time. Testing the timely arrival and integrity
of the data transmission itself is a first step. If the data are corrupted during transmission, further testing may
be irrelevant. The checks defined in these 13 tests evaluate data through various comparisons to other data
and to the expected conditions in the given environment. The tests listed in this section presume a timeordered series of observations and denote the most recent observation as previously described.
Sensor operators need to select the best thresholds for each test, which are determined at the operator level
and may require trial and error/iteration before final selections are made. A successful QC effort is highly
dependent upon selection of the proper thresholds, which should not be determined arbitrarily but can be
based on historical knowledge or statistics derived from more recently acquired data. Although this manual
provides some guidance for selecting thresholds based on input from various operators, it is assumed that
operators have the expertise and motivation to select the proper thresholds to maximize the value of their
QC effort. Operators must openly provide thresholds as metadata for user support. This shared information
will help U.S. IOOS to document standardized thresholds that will be included in future releases of this
manual.

3.3.1 Applications of QC Tests to Stationary TS Sensors
These 13 tests require operators to select a variety of thresholds. Examples are provided in the following test
tables; however, operators are in the best position to determine the appropriate thresholds for their
operations. Some tests rely on multiple data points most recently received to determine the quality of the
current data point. When this series of data points reveals that the entire group fails, the current data point is
flagged, but the previous flags are not changed. This action supports the view that historical flags are not
altered. The first example of this scenario is Test 8, the Flat Line Test.

Test 1) Timing/Gap Test (Required)
Check for arrival of data.
Test determines that the most recent data point has been measured and received within the expected time
window (TIM_INC) and has the correct time stamp (TIM_STMP).
Note: For those systems that do not update at regular intervals, a large value for TIM_STMP can be assigned. The
gap check is not a solution for all timing errors. Data could be measured or received earlier than expected. This test
does not address all clock drift/jump issues.

Flags
Fail=4
Suspect=3

Condition

Codable Instructions

Data have not arrived as expected.
N/A

If NOW – TIM_STMP > TIM_INC, flag = 4
N/A

Pass=1

Applies for test pass condition.

N/A

Test Exception: None.
Test specifications to be established locally by the operator.
Example:
TIM_INC= 1 hour

Test 2) Syntax Test (Required)
Check to ensure that the message is structured properly
Received data message (full message) contains the proper structure without any indicators of flawed transmission
such as parity errors. Possible tests are: a) the expected number of characters (NCHAR) for fixed length messages
equals the number of characters received (REC_CHAR), or b) passes a standard parity bit check, cyclic redundancy
check (CRC), etc. Many such syntax tests exist, and the operator should select the best criteria for one or more
syntax tests.
Capabilities for dealing with flawed messages vary among operators; some may have the ability to parse messages
to extract data within the flawed message sentence before the flaw. A syntax check is performed only at the
message level and not within the message content. In cases where a data record requires multiple messages, this
check can be performed at the message level but is not used to check message content.

Flags
Fail=4
Suspect =3
Pass=1

Condition

Codable Instructions

Data sentence cannot be parsed
to provide a valid observation.
N/A

If REC_CHAR ≠ NCHAR, flag = 4
N/A

Expected data sentence received;
absence of parity errors.

Test Exception: None.
Test specifications to be established locally by the operator.
Example:
NCHAR = 128

Test 3) Location Test (Required)
Check for reasonable geographic location.
Test checks that the reported present physical location (latitude/longitude) is within operator-determined limits.
The location test(s) can vary from a simple impossible location to a more complex check for displacement (DISP)
exceeding a distance limit (RANGEMAX) based upon a previous location and platform speed. Operators may also
check for erroneous locations based upon other criteria, such as reported positions over land, as appropriate.

Flags
Fail=4
Suspect=3
Pass=1

Condition

Codable Instructions

Impossible location.
Unlikely platform displacement.
Applies for test pass condition.

LAT > | 90 | or LONG > | 180 |
DISP > RANGEMAX
N/A

Test Exception: Test does not apply to fixed deployments when no location is transmitted.
Test specifications to be established locally by the operator.
Example:
Displacement DISP calculated between sequential position reports, RANGEMAX = 20 km

Temperature and Salinity

Test 4) Gross Range Test (Required)
Data point exceeds sensor or operator-selected min/max. Applies to T, SP, C and P.
All sensors have a limited output range, and this can form the most rudimentary gross range check. No values less
than a minimum value or greater than the maximum value the sensor can output (T_SENSOR_MIN,
T_SENSOR_MAX) are acceptable. Additionally, the operator can select a smaller span (T_USER_MIN, T_USER_MAX)
based upon local knowledge or a desire to draw attention to extreme values.
NOTE: Operators may choose to flag as suspect values that exceed the calibration span but not the hardware limits
(e.g., a value that sensor is not capable of producing or negative conductivity).

Flags
Fail=4
Suspect=3
Pass=1

Condition

Codable Instructions

Reported value is outside of sensor
span.
Reported value is outside of operatorselected span.
Applies for test pass condition.

If Tn < T_SENSOR_MIN, or
Tn > T_SENSOR_MAX, flag = 4
If Tn < T_USER_MIN, or
Tn > T_USER_MAX, flag = 3

Test Exception: None.
Test specifications to be established locally by the operator.
Examples: The following global range min/max are applied on some climate and forecast standard-names in the
IMOS toolbox: depth: -5/12,000 m
sea_water_pressure: -5/12,000 decibars (dbar)
sea_water_pressure_due_to_sea_water: -15/12,000 dbar
sea_water_salinity: 2/41
sea_water_temperature: -2.5/40 °C

Test 5) Climatology Test (Required)
Test that data point falls within seasonal expectations. Applies to T and SP.
This test is a variation on the gross range check, where the thresholds T_Season_MAX and T_Season_MIN are
adjusted monthly, seasonally, or at some other operator-selected time period (TIM_TST). Expertise of the operator
is required to determine reasonable seasonal averages. Longer time series permit more refined identification of
appropriate thresholds. The ranges should also vary with water depth, if the measurements are taken at sites that
cover significant vertical extent and if climatological ranges are meaningfully different at different depths (e.g.,
narrower ranges at greater depth).

Flags
Fail=4

Condition

Codable Instructions

Because of the dynamic nature of T and S in some
locations, no fail flag is identified for this test.

N/A

Suspect=3

Reported value is outside of operator-identified
climatology window.

If Tn < T_Season_MIN or
Tn > T_Season_MAX, flag = 3

Pass=1

Applies for test pass condition.

N/A

Test Exception: None.
Test specifications to be established locally by operator: A seasonal matrix of Tmax and Tmin values at all
TIM_TST intervals.
Examples: T_SPRING_MIN = 12 °C, T_SPRING_MAX = 18.0 °C

Test 6) Spike Test (Strongly Recommended)
Data point n-1 exceeds a selected threshold relative to adjacent data points. Applies to T, SP,
C, and P.
This check is for single value spikes, specifically the value at point n-1. Spikes consisting of more than one data point
are difficult to capture, but their onset may be flagged by the rate of change test. The spike test consists of two
operator-selected thresholds, THRSHLD_LOW and THRSHLD_HIGH. Adjacent data points (n-2 and n0) are averaged to
form a spike reference (SPK_REF). The absolute value of the spike is tested to capture positive and negative spikes.
Large spikes are easier to identify as outliers and flag as failures. Smaller spikes may be real and are only flagged
suspect. The thresholds may be fixed values or dynamically established (for example, a multiple of the standard
deviation over an operator-selected period).

Flags
Fail=4
Suspect=3

Condition
High spike threshold exceeded.
Low spike threshold exceeded.

Pass=1

Applies for test pass condition.

Test Exception: None.
Test specifications to be established locally by the operator.
Examples: THRSHLD_LOW = 3 °C, THRSHLD_HIGH = 8 °C

Test 7) Rate of Change Test (Strongly Recommended)
Excessive rise/fall test. Applies to T, SP, C, and P.
This test inspects the time series for a time rate of change that exceeds a threshold value identified by the
operator. T, SP, C, P values can change substantially over short periods in some locations, hindering the value of
this test. A balance must be found between a threshold set too low, which triggers too many false alarms, and one
set too high, making the test ineffective. Determining the excessive rate of change is left to the local operator. The
following show three different examples of ways to select the thresholds provided by QARTOD VI participants.
Implementation of this test can be challenging. Upon failure, it is unknown which of the points is bad. Further, upon
failing a data point, it remains to be determined how the next iteration can be handled.
The rate of change between temperature Tn-1 and Tn must be less than three standard deviations (3*SD). The SD of
the T time series is computed over the previous 25-hour period (operator-selected value) to accommodate cyclical
diurnal and tidal fluctuations. Both the number of SDs (N_DEV) and the period over which the SDs (TIM_DEV) are
calculated and determined by the local operator.


The rate of change between temperature Tn-1 and Tn must be less than 2 °C +2SD.



|Tn-1 – Tn-2| + |Tn-1 – Tn| <= 2*N_DEV*SD (example provided by EuroGOOS).

Flags

Condition

Codable Instructions

Fail=4
Suspect=3

No fail flag is identified for this test.
The rate of change exceeds the selected threshold.

N/A
If |Tn – Tn-1|>N_DEV*SD, flag = 3

Pass=1

Applies for test pass condition.

N/A

Test Exception: None.
Test specifications to be established locally by operator.
Example:
N_DEV = 3, TIM_DEV = 25

Temperature and Salinity

Test 8) Flat Line Test (Strongly Recommended)
Invariant value. Applies to T, SP, C, and P.
When some sensors and/or data collection platforms fail, the result can be a continuously repeated observation of
the same value. This test compares the present observation n to a number (REP_CNT_FAIL or REP_CNT_SUSPECT)
of previous observations. Observation n is flagged if it has the same value as previous observations within a
tolerance value, EPS, to allow for numerical round-off error. Note that historical flags are not changed.

Flags
Fail=4
Suspect=3

Pass=1

Condition

Codable Instructions

When the five most recent observations are
equal, Tn is flagged fail.

For i=1,REP_CNT_FAIL Tn -Tn-i
N_T_DEV*SD_T
AND
|SPn – SPn-1|N_T1_DEV*SD1
AND
|T2n – T2n-1| SP_fit_fail, flag = 4

If |SPn-SPfit| < SP_fit_fail and |SPn-SPfit | > SP_fit warn,
flag = 3
N/A

Test Exception: The test will probably not be useful in estuaries or ocean surface waters.
Test specifications to be established locally by the operator.
Examples:
At the Bermuda Atlantic Time Series site, for a temperature of 18 °C, SPfit = 36.5
SP_fit_fail = 0.05, SP_fit_warn = 0.02

Test 13) Density Inversion Test (Suggested)
Checks that density increases with pressure (depth).
With few exceptions, potential water density σθ will increase with increasing pressure. When vertical profile data
are obtained, this test is used to flag as failed T, C, and SP observations, which yield densities that do not
sufficiently increase with pressure. A small, operator-selected density threshold (DT) allows for micro-turbulent
exceptions. Here, σθn is defined as one sample increment deeper than σθn-1. With proper consideration, the test can
be run on downcasts, upcasts, or down/up cast results produced in real time.
From a computational point of view, this test is similar to the rate of change test (test 7), except that the time axis
is replaced by depth. The same code can be used for both, using different variables and thresholds. As with the rate
of change test, it is not known which side of the step is good versus bad.
An example of the software to compute sigma-theta is available at http://www.teos-10.org/software.htm.

Flags
Fail=4

Codable Instructions
If σθn-1 +DT > σθn , flag = 4

Suspect=3

Condition
Potential density does not sufficiently increase with increasing
depth.
No suspect flag is identified for this test.

Pass=1

Potential density sufficiently increases with increasing depth.

If σθn-1 +DT ≤ σθn, flag = 1

Test Exception: None.
Test specifications to be established locally by the operator.
Examples:
DT = 0.03 kg/m3

N/A

Temperature and Salinity

3.3.2

Applications of QC Tests to Mobile TS Sensors

The specific application of the QC tests can be dependent on the way the sensor is deployed. For mobile
platforms, at least two existing programs, GTSPP and Argo, have developed QC tests that are similar to the
U.S. IOOS QARTOD tests in this manual. Manuals from both programs are available online
(UNESCO-IOC 2010; Carval et al. 2015). Operators within such programs will likely find their present QC
process to be compliant with U.S. IOOS QARTOD requirements and recommendations, which is the
intention of the QARTOD TS Committee. Table 3-3 provides a comparison of salinity and temperature QC
tests from the U.S. IOOS QARTOD, GTSPP, and real-time Argo programs.
Table 3-3. Comparison of QARTOD, GTSPP, and Argo temperature and salinity QC tests

QARTOD

GTSPP

Argo

1)
2)
3)
4)
5)
6)
7)
8)
9)
10)
11)
12)
13)

1.2
No match
1.3, 1.4
2.1
3.1, 3.2, 3.3, 3.4
2.7, 2.8
2.9, 4.1
2.4, 2.5
No match
2.4
No match
No match
2.10

2
1 (close, not identical)
3, 4, 5
6, 7
No match
9
11
14, 18
No match
16 (close, not identical)
No match
No match
14

Time/Gap Test
Syntax Test
Location Test
Gross Range Test
Climatological Test
Spike Test
Rate of Change Test
Flat Line Test
Multi-Variate Test
Attenuated Signal Test
Neighbor Test
TS Curve/Space Test
Density Inversion Test

Tables 3-4 through 3-6 provide a summary of each QC test described in section 3.3.1 and indicate any
changes necessary for the test to be applied to mobile deployment scenarios. Note that the “s” axis indicates
“along path” for mobile platforms. Each data point, whether horizontal, vertical, or along the path, is quality
controlled and assigned a flag using these tests. Operators may choose to expand upon the flagging scheme
using another tier of flags, e.g., to characterize the entire vertical profile.

Table 3-4. Application of Required QC Tests (Tests 1-5) for TS Sensor Deployments

Test

Condition

Platform

Codable
Instructions

1) Timing/Gap Test (Required)

Check for
arrival of data.

Stationary
Fixed Vertical
Mobile
3D

No change

Expected data
sentence
received,
absence of
parity errors.

Stationary
Fixed Vertical
Mobile
3D

No change

Check for
acceptable
geographic
location.

Stationery
Fixed Vertical
Mobile
3D

No change

Test determines that the most recent data
point has been measured and received within
the expected time window (TIM_INC) and has
the correct time stamp (TIM_STMP).
Note: For those systems that do not update at
regular intervals, a large value for TIM_STMP can
be assigned. The gap check is not a panacea for all
timing errors. Data could be measured or received
earlier than expected. This test does not address
all clock drift/jump issues.
2) Syntax Test (Required)
Received data message (full message) contains
the proper structure without any indicators of
flawed transmission such as parity errors.
Possible tests are: a) the expected number of
characters (NCHAR) for fixed length messages
equals the number of characters received
(REC_CHAR), or b) passes a standard parity bit
check, cyclic redundancy check (CRC), etc. Many
such syntax tests exist, and the operator should
select the best criteria for one or more syntax
tests.
Note: Capabilities for dealing with flawed
messages vary among operators; some may
have the ability to parse messages to extract
data within the flawed message sentence before
the flaw. A syntax check is performed only at the
message level and not within the message
content. In cases where a data record requires
multiple messages, this check can be performed
at the message level, but is not used to check
message content
3) Location Test (Required)
Test checks that the reported present physical
location (latitude/longitude) is within operatordetermined limits. The location test(s) can vary
from a simple impossible location to a more
complex check for displacement (DISP) exceeding
a distance limit RANGEMAX based upon a
previous location and platform speed. Operators
may also check for erroneous locations based
upon other criteria, such as reported positions
over land, as appropriate.

Temperature and Salinity

Test

Condition

Platform

Codable
Instructions

4) Gross Range Test (Required)
All sensors have a limited output range, and this
can form the most rudimentary gross range
check. No values less than a minimum value or
greater than the maximum value the sensor can
output (T_SENSOR_MIN, T_SENSOR_MAX) are
acceptable. Additionally, the operator can select
a smaller span (T_USER_MIN, T_USER_MAX)
based upon local knowledge or a desire to draw
attention to extreme values.

Data point
exceeds sensor
or operator
selected
min/max.

Stationary
Fixed Vertical
Mobile
3D

No change

5) Climatology Test (Required)
This test is a variation on the gross range check,
where the thresholds T_Season_MAX and
T_Season_MIN (for example) are adjusted
monthly, seasonally, or at some other operatorselected time period (TIM_TST). Expertise of the
local user is required to determine reasonable
seasonal averages. Longer time series permit
more refined identification of appropriate
thresholds.

Test that data
point falls
within seasonal
expectations.

Stationary
Fixed Vertical

No change
Test is conducted
along z axis
Test is conducted
along s, x, or y axis
Test is conducted
along s, x, y, or z
axis

Mobile
3D

Table 3-5. Application of Strongly Recommended QC Tests (Tests 6-8) for TS Sensor Deployments

Test

Condition

Platform

Codable
Instructions

6) Spike Test (Strongly Recommended)
This check is for single value spikes, specifically the
value at point n-1. Spikes consisting of more than
one data point are difficult to capture, but their
onset may be flagged by the rate of change test. The
spike test consists of two operator-selected
thresholds, THRSHLD_LOW and THRSHLD_HIGH.
Adjacent data points (n-2 and n) are averaged to
form a spike reference (SPK_REF). The absolute
value of the spike is tested to capture positive and
negative spikes. Large spikes are easier to identify as
outliers and flag as failures. Smaller spikes may be
real and are only flagged suspect. The thresholds
may be fixed values or dynamically established (for
example, a multiple of the standard deviation over a
specified period).

Data point n-1
exceeds a
selected
threshold
relative to
adjacent data
points.

Stationary
Fixed Vertical

No change
Test is conducted
along z axis

Mobile

No change, or test
is conducted along
s, x, or y axis

No change, or test
is conducted along
s, x, y, or z axis

7) Rate of Change Test (Strongly Recommended)
This test inspects the time series for a time rate of
change that exceeds a threshold value identified by
the operator. T, SP, C, P values can change
substantially over short periods in some locations,
hindering the value of this test. A balance must be
found between a threshold set too low, which
triggers too many false alarms, and one set too high,
making the test ineffective. Determining the
excessive rate of change is left to the local operator.
The following show two different examples of ways
to select the thresholds provided by QARTOD VI
participants. Implementation of this test can be
challenging. Upon failure, it is unknown which of the
points is bad. Further, upon failing a data point, it
remains to be determined how the next iteration
can be handled.
 The rate of change between Tn-1 and Tn must be
less than three standard deviations (3*SD). The
SD of the T time series is computed over the
previous 25-hour period (operator-selected
value) to accommodate cyclical diurnal and tidal
fluctuations. Both the number of SDs (N_DEV)
and the period over which the SDs (TIM_DEV)
are calculated and determined by the local
operator.
 The rate of change between Tn-1 and Tn must be
less than 2 °C +2SD.
 |Tn-1 – Tn-2| + |Tn-1 – Tn| <= 2*N_DEV*SD
(example provided by EuroGOOS).

Excessive
rise/fall test

Stationary
Fixed Vertical

No change
Test is conducted
along z axis

Mobile

No change, or test
is conducted along
s, x, or y axis

No change, or test
is conducted along
s, x, y, or z axis

Temperature and Salinity

Test

Condition

Platform

Codable
Instructions

8) Flat Line Test (Strongly Recommended)
When some sensors and/or data collection
platforms fail, the result can be a continuously
repeated observation of the same value. This test
compares the present observation n to a number
(REP_CNT_FAIL or REP_CNT_SUSPECT) of previous
observations. Observation n is flagged if it has the
same value as previous observations within a
tolerance value EPS to allow for numerical round-off
error. Note that historical flags are not changed.

Invariant value

Stationary
Vertical

No change
Test is conducted
along z axis
No change, or test
is conducted along
s, x, or y axis

Mobile

No change, or test
is conducted along
s, x, y, or z axis

Table 3-6. Application of Suggested QC Tests (Tests 9-13) for TS Sensor Deployments

Test

Condition

Platform

Codable
Instructions

9) Multi-Variate Test (Suggested)
This is an advanced family of tests, starting with
the simpler test described here and anticipating
growth towards full co-variance testing in the
future. It is doubtful if anyone is conducting tests
such as these in real time. As these tests are
developed and implemented, they should be
documented and standardized in later versions of
this manual.
This example pairs the rate of change tests as
described in test 7. The T (or SP or P) rate of
change test is conducted with a more restrictive
threshold (N_T_DEV). If this test fails, a second
rate of change test operating on a second variable
(salinity or current would be the most probable) is
conducted. The absolute valued rate of change
should be tested, since the relationship between
T and variable two is indeterminate. If the rate of
change test on the second variable fails to exceed
a threshold (e.g., an anomalous step is found in T
and is lacking in salinity), then the Tn value is
flagged.

Comparison to
other variables

Stationary
Fixed Vertical

No change
Test is conducted
along z axis
Test is conducted
along s, x, or y axis
Test is conducted
along s, x, y, or z axis

10) Attenuated Signal Test (Suggested)
A common sensor failure mode can provide a data
series that is nearly but not exactly a flat line (for
example, if the sensor head were to become
wrapped in debris). This test inspects for a
standard deviation (SD) value or a range variation
(MAX-MIN) value that fails to exceed threshold
values (MIN_VAR_WARN, MIN_VAR_FAIL) over a
selected time period (TST_TIM).

Inadequate
variation test

Mobile
3D

Stationary
Fixed Vertical
Mobile

No change
Test is conducted
along z axis
No change, or test is
conducted along s,
x, or y axis
No change, or test is
conducted along s,
x, y, or z axis

Temperature and Salinity

Test

Condition

Platform

Codable
Instructions

11) Neighbor Test (Suggested)
The check has the potential to be the most useful
test when a nearby second sensor is determined
to have a similar response.
Ideally, redundant sensors utilizing different
technology would be co-located and alternately
serviced at different intervals. This close neighbor
would provide the ultimate QC check, but cost
prohibits such a deployment in most cases.
However, there are few instances where a second
sensor is sufficiently proximate to provide a useful
QC check. Just a few hundred meters in the
horizontal and less than 10 meters vertical
separation can often yield greatly different
results. Nevertheless, the test should not be
overlooked where it may have application.
This test is the same as Test 9), Multi-variate
Check – comparison to other variables where the
second variable is the second sensor. The selected
thresholds depend entirely upon the relationship
between the two sensors as determined by the
local knowledge of the operator.
In the instructions and examples below, data from
one site (T1) are compared to a second site (T2).
The standard deviation for each site (SD1, SD2) is
calculated over the period (TIM_DEV) and
multiplied as appropriate (N_T1_DEV for site T1)
to calculate the rate of change threshold. Note
that an operator could also choose to use the
same threshold for each site, since they are
presumed to be similar.

Comparison to
nearby sensors of
the same variable

Stationary
Fixed Vertical
Mobile
3D

No change
Test is conducted
along z axis
No change
No change

Test 12) TS Curve/Space Test (Suggested)

Comparison to
expected TS
relationship

Stationary

No change

Fixed Vertical

Test is conducted
along z axis
Test is conducted
along s, x, or y axis
Test is conducted
along s, x, y, or z axis

The TS curve is a classic tool used to evaluate
observations, especially in the open ocean below
the thermocline. Site-specific TS curve
characteristics are used to identify outliers. The
curve could be either a fitted equation or
numerical table. For a given Tn, SPn is expected to
be within SPfit ± SP_fit_warn or SP_fit_fail,
operator-provided values. The value SPfit is
obtained from the equation or table.

Mobile
3D

Test

Condition

Platform

Codable
Instructions

Test 13) Density Inversion Test (Suggested)

Checks that
density increases
with pressure
(depth)

Stationary
Fixed Vertical

No change
Test is conducted
along z axis

Mobile

No change, or test is
conducted along s,
x, or y axis
No change, or test is
conducted along s,
x, y, or z axis

With few exceptions, potential water density σθ
will increase with increasing pressure. When
vertical profile data is obtained, this test is used
to flag as failed T, C, and SP observations, which
yield densities that do not sufficiently increase
with pressure. A small operator-selected density
threshold (DT) allows for micro-turbulent
exceptions. Here, σθn is defined as one sample
increment deeper than σθn-1. With proper
consideration, the test can be run on downcasts,
upcasts, or down/up cast results produced in real
time.

From a computational point of view, this test is
similar to the rate of change test (test 7), except
that the time axis is replaced by depth. The same
code can be used for both, using different
variables and thresholds. As with the rate of
change test, it is not known which side of the step
is good versus bad.
An example of the software to compute sigmatheta is available at http://www.teos10.org/software.htm.

Temperature and Salinity

4.0 Summary
The QC tests in this TS manual have been compiled using the guidance provided by the TS committee and
valuable reviewers (appendix B), all QARTOD workshops (www.ioos.noaa.gov/qartod), and earlier U.S.
IOOS/QARTOD manuals. Test suggestions came from both operators and TS data users with extensive
experience. The considerations of operators who ensure the quality of real-time data may be different from
those whose data are not published in real time, and these and other differences must be balanced according
to the specific circumstances of each operator. Although these real-time tests are required, strongly
recommended, or suggested, it is the operator who is responsible for deciding which tests are appropriate.
Each operator selects thresholds based on the specific program requirements that must be met. The scope of
requirements can vary widely, from complex data streams that support myriad QC checks to ensure precise
and accurate measurements - to basic data streams that do not need such details. Operators must publish
their QC processes via metadata so that data users can readily see and understand the source and quality of
those data.
The 13 QC tests identified in this manual apply to TS observations from a variety of sensor types and
platforms that may be used in U.S. IOOS. At least two existing programs, GTSPP (UNESCO-IOC 2010) and
Argo (Carval et al. 2015), have developed QC tests for mobile platforms that are similar to the U.S. IOOS
QARTOD tests in this manual. The QARTOD TS committee intends for the QC tests of these programs to
be compliant with U.S. IOOS QARTOD requirements and recommendations. The individual tests are
described and include codable instructions, output conditions, example thresholds, and exceptions (if any).
Selection of the proper thresholds is critical to a successful QC effort. Thresholds can be based on historical
knowledge or statistics derived from more recently acquired data, but they should not be determined
arbitrarily. This manual provides guidance for selecting thresholds based on input from various operators, but
also notes that operators need the subject matter expertise and motivation to select the proper thresholds to
maximize the value of their QC effort.
Future QARTOD reports will address standard QC test procedures and best practices for all types of
common and uncommon platforms and sensors for all the U.S. IOOS core variables. Some test procedures
may take place within the sensor package. Significant components of metadata will reside in the sensor and be
transmitted either on demand or automatically along with the data stream. Users may also reference metadata
through Uniform Resource Locators (URLs) to simplify the identification of which QC steps have been
applied to data. However, QARTOD QC test procedures in this manual address only real-time in-situ
observations. The tests do not include post-processing, which is not in real time but may be useful for
ecosystem-based management, or delayed-mode, which might be suitable for climate studies
Each QARTOD manual is envisioned as a dynamic document and will be posted on the QARTOD website
at www.ioos.noaa.gov/qartod/. This process allows for QC manual updates as technology development
occurs for both upgrades of existing sensors and new sensors.

5.0 References
Bushnell, M., Presentation at QARTOD III: November 2005. Scripps Institution of Oceanography, La Jolla,
California.
Carval Thierry, Keeley Robert, Takatsuki Yasushi, Yoshida Takashi, Schmid Claudia, Buck Justin, Wong
Annie, Thresher Ann, Tran Anh (2015). Argo user’s manual V3.2. http://dx.doi.org/10.13155/29825
Garau, B., Ruiz, S., Zhang, W., Pasucal, A., Heslop, E., Kerfoot, J., and Tintore, J., 2011. Thermal Lag
Correction on Slocum CTD Glider Data, Journal of Atmospheric and Oceanic Technology, 28, 1065 –
1071.
GTSPP Real-Time Quality Control Manual, First Revised Edition. UNESCO-IOC 2010. (IOC Manuals and
Guides No. 22, Revised Edition.) (IOC/2010/MG/22Rev.)
Note: The initial version of the manual was published in 1990 (SC-90/WS-74) Published in 2010 by the
United Nations Educational, Scientific and Cultural Organization, 7, Place de Fontenoy, 75352, Paris 07
SP UNESCO 2010.
http://www.nodc.noaa.gov/GTSPP/document/qcmans/MG22rev1.pdf
Interagency Ocean Observation Committee (IOOC), 2012. Integrated Ocean Observing System (IOOS)
Certification Criteria. 11 pp. http://www.iooc.us/wp-content/uploads/2012/05/IOOS-CertificationCriteria_4-25-12.pdf
McDougall, T.J., Feistel, R., Millero, F.J., Jackett, D.R., Wright, D.G., King, B.A., Marion, G. M., Chen, CT.A., and Spitzer, P., 2009. Calculation of the Thermophysical Properties of Seawater, Global Ship-based
Repeat Hydrography Manual. IOCCP Report No. 14, ICPO Publication Series No. 134.
http://www.marine.csiro.au/~jackett/TEOS-10/Thermophysical_Manual_09Jan09.pdf.
Morison, J., R. Andersen, N. Larson, E. D'Asaro, and T. Boyd, 1994: The correction for thermal-lag effects in
Sea-Bird CTD data. J. Atmos. Ocean. Technol., 11, 1151–1164.
Paris. Intergovernmental Oceanographic Commission of UNESCO. 2013. Ocean Data Standards, Vol. 3:
Recommendation for a Quality Flag Scheme for the Exchange of Oceanographic and Marine
Meteorological Data. (IOC Manuals and Guides, 54, Vol. 3.) 12 pp. (English.) (IOC/2013/MG/54-3)
http://www.nodc.noaa.gov/oceanacidification/support/MG54_3.pdf.
UNESCO, Tenth Report of the Joint Panel on Oceanographic Tables and Standards. UNESCO Technical
Papers in Marine Science 36. Sydney, BC, 1981.
U.S. IOOS Office, November 2010. A Blueprint for Full Capability, Version 1.0, 254 pp.
www.ioos.noaa.gov/library/us_ioos_blueprint_ver1.pdf
U.S. Integrated Ocean Observing System, January 2014. Manual for the Use of Real-Time
Oceanographic Data Quality Control Flags. 19 pp.
http://www.ioos.noaa.gov/qartod/temperature_salinity/qartod_oceanographic_data_quali
ty_manual.pdf

Temperature and Salinity

Additional References to Related Documents:
Alliance for Coastal Technologies (ACT) 2012. Accessed September 20, 2012 at http://www.actus.info/evaluations.php
Boyer, T.P., J. I. Antonov, O. K. Baranova, C. Coleman, H. E. Garcia, A. Grodsky, D. R. Johnson, R. A.
Locarnini, A. V. Mishonov, T.D. O'Brien, C.R. Paver, J.R. Reagan, D. Seidov, I. V. Smolyar, and M. M.
Zweng, 2013: World Ocean Database 2013, NOAA Atlas NESDIS 72, S. Levitus, Ed., A. Mishonov,
Technical Ed.; Silver Spring, MD, 209 pp., http://doi.org/10.7289/V5NZ85MT

http://www.nodc.noaa.gov/OC5/WOD/pr_wod.html
CALCOFI: Seabird dual SBE43 oxygen sensors (O2); rated to 7000m used with Seabird CTD in conjunction
with temperature and salinity sensors to calculate all pertinent data.
http://calcofi.org/references/ccmethods/283-art-ctdatsea.html
Scheme on QC flags, which is a general document that discusses how to write the results of tests, but does
not discuss the actual tests. http://www.oceandatastandards.org/
The ocean data standards resource pool can be found at:
http://www.oceandatastandards.org/resourcepool-mainmenu-7
http://www.oceandatastandards.org/index.php?option=com_content&task=view&id=22&Itemid=28 is
the higher level page (see menu to the right for sub-pages). There is a sub page for T and S profiles that
contains a lot of good information including names and reference documents. Some of the references
under T and S also apply to DO.
National Oceanographic Partnership Program (NOPP) January 2006. The First U.S. Integrated Ocean
Observing System (IOOS)Development Plan – A report of the national Ocean Research Leadership
Council and the Interagency Committee on Ocean Science and Resource Management Integration. The
National Office for Integrated and Sustained Ocean Observations. Ocean US Publication No. 9.
Dickson, A.G., Sabine, C.L. and Christian, J.R. (Eds.) 2007. Guide to best practices for ocean CO2
measurements. PICES Special Publication 3, 191 pp.
Guidelines on Quality Control Procedures for Data from Automatic Weather Stations
https://www.wmo.int/pages/prog/www/IMOP/meetings/Surface/ETSTMT1_Geneva2004/Doc6.1(2).pdf
IODE recommendation on flagging scheme:
http://www.oceandatastandards.info
http://www.iode.org/index.php?option=com_oe&task=viewDocumentRecord&docID=10762
The OceanSITES manual also has a description of flagging schemes.
http://www.oceansites.org/docs/oceansites_data_format_reference_manual.pdf

Sun, C. & Co-Authors (2010). "The Data Management System for the Global Temperature and Salinity
Profile Programme" in Proceedings of OceanObs.09: Sustained Ocean Observations and Information for
Society (Vol. 2), Venice, Italy, 21-25 September 2009, Hall, J., Harrison, D.E. & Stammer, D., Eds., ESA
Publication WPP-306, doi:10.5270/OceanObs09.cwp.86
NODC Website
http://www.nodc.noaa.gov/GTSPP/
Argo Website – Reference Documentation
http://www.argodatamgt.org/Documentation
Thermodynamic Equation of Seawater - 2010 (TEOS-10
http://www.teos-10.org
Absolute Salinity and TEOS-10: Sea-Bird's Implementation
http://www.seabird.com/application_notes/AN90.htm
National Data Buoy Center (NDBC) Technical Document 09-02, Handbook of Automated Data Quality
Control Checks and Procedures, August 2009. National Data Buoy Center, Stennis Space Center,
Mississippi 39529-6000.
NOAA, 2005. Second Workshop Report on the QA of Real-Time Ocean Data, July 2005. 48 pp. Norfolk,
Virginia. CCPO Technical Report Series No. 05-01
NOAA, 2005, Workshop Report on Salinity Data Management Best Practices, August 2005, 37 pp,
Charleston, SC, Hosted by Coastal Services Center
NOAA, 2009. Fifth Workshop on the QA/QC of Real-Time Oceanographic Data. November 16-19, 2009. 136
pp. Omni Hotel, Atlanta, Georgia.
Ocean.US, 2006. National Office for Integrated and Sustained Ocean Observations. The First U.S. Integrated
Ocean Observing System (IOOS) Development Plan, Publication 9, January 2006.
U.S. IOOS QARTOD Project Plan, February 18, 2012.
http://www.ioos.noaa.gov/qartod/meetings.html
Draft DBCP TD. No. 42 - Sea Surface Salinity Quality Control processes for potential use on Data Buoy
observations
http://www.jcomm.info/index.php?option=com_oe&task=viewDocumentRecord&docID=5656
CLIVAR-GSOP Coordinated Quality-Control of Global Subsurface Ocean Climate Observations
http://www.clivar.org/organization/gsop/activities/clivar-gsop-coordinated-quality-control-globalsubsurface-ocean-climate
GTSPP Real-time Quality Control Manual Revised Edition, 2010 (Same as lines 534-539.)
http://www.nodc.noaa.gov/GTSPP/document/qcmans/index.html
Data QC Flags from CSIRO Cookbook
http://www.nodc.noaa.gov/GTSPP/document/qcmans/csiro/csiro.htm

Temperature and Salinity

IMOS Toolbox
https://github.com/aodn/imos-toolbox
Salinometry
Informational website about the laboratory salinity measurements, history of salinity, PSS-78
http://salinometry.com/

Supporting Documents Available from the QARTOD Website:
http://www.ioos.noaa.gov/qartod/temperature_salinity/welcome.html
These documents were particularly useful to the committee and reviewers when developing this manual. They do not contain
copyright restrictions and are posted on the U.S. IOOS QARTOD website for easy reference.
Argo User’s Manual, V3.2. December 29, 2015.
GTSPP Real-Time Quality Control Manual, First Revised Edition. UNESCO-IOC 2010. (IOC Manuals and
Guides No. 22, Revised Edition.) (IOC/2010/MG/22Rev.)
NOAA, 2005, Workshop Report on Salinity Data Management Best Practices, August 2005, 37 pp,
Charleston, SC, Hosted by Coastal Services Center
U.S. IOOS Development Plan
NDBC Handbook of Automated Data Quality Control
Data Quality Control in the U.S. IOOS
National Water Quality Monitoring Council Water Quality Data Elements: A User Guide

Temperature and Salinity

Appendix A.

Quality Assurance (QA)

A major pre-requisite for establishing quality control standards for TS measurements is a strong quality
assurance program. Remember the mantra that good QC requires good QA, and good QA requires good
scientists, engineers, and technicians. A good QA effort continuously seeks to ensure that end data products
are of high value and strives to prove they are free of error.
The following sections suggest ways to ensure QA by using specific procedures and techniques. Operators
should document the processes they use to perform QA. Additionally, details of QA for sensors associated
with specific observations should be captured and made available to data consumers as part of the
accompanying metadata (e.g., sensor calibration date, sensor service history).

A.1

Sensor Calibration Considerations

Observations must be traceable to one or more accepted standards through a calibration performed by the
manufacturer and/or the operator. If the calibration is conducted by the manufacturer, the operator must
also conduct some form of an acceptable calibration check.
NIST provides a wealth of information on standards and calibrations for many variables, including time,
temperature, and pressure. Virtually all manufacturers provide calibrations traceable to NIST standards as
part of their standard product services. However, this is not the case for salinity. Salinity/conductivity sensors
are most commonly calibrated against IAPSO standard seawater, which is available from OSIL. The ocean
observing community uses the practical salinity unit (PSU) as defined by the practical salinity scale, developed
in 1978 (UNESCO 1981).
An often overlooked calibration or calibration check can be performed by choosing a consensus standard.
For example, deriving the same answer (within acceptable levels of data precision or data uncertainty) from
four different sensors of four different vendors, preferably utilizing several different technologies, constitutes
an acceptable check. Because of the trend towards corporate conglomeration, those wishing to employ a
consensus standard should ensure that the different vendors are truly independent.

A.2

Sensor Comparison

An effective QA effort continuously strives to ensure that end data products are of high value and to prove
they are free of error. Operators should seek out partnering opportunities to inter-compare systems by colocating differing sensors. Agreement of multiple systems would provide a robust observation, while
disagreement may offer a measure of data uncertainty. If possible, operators should retain an alternate sensor
or technology from a second vendor for similar in-house checks. For resource-constrained operators,
however, it may not be possible to spend the time and funds needed to procure and maintain two systems.
For those who do so and get two different results, the use of alternate sensors or technologies provide several
important messages: a) a measure of corporate capabilities; b) a reason to investigate, understand the different
results, and take corrective action; and c) increased understanding that when variables are measured with
different technologies, different answers can be correct, and they must be understood in order to properly
report results. For those who succeed, the additional sensors provide a highly robust demonstration of
capability. Such efforts form the basis of a strong QA/QC effort. Further, it provides the operator with an
expanded supply source, permitting less reliance upon a single vendor and providing competition that is often
required by procurement offices.

A-1

A.3

Bio-fouling and Corrosion Prevention Strategies

Bio-fouling is the most frequent cause of sensor failure, so the following strategies may be useful for
ameliorating the problem:














Use anti-fouling paint with the highest copper content available (up to 75%) when possible (not on
aluminum).
Use tributyltin (TBT), anti-foulant cartridges to slow fouling of the conductivity cell.
Wrap body of sensor with clear packing tape for a small probe or plastic wrap for a large instrument.
This keeps the PVC tape from leaving residue on the sensor. Heavy PVC underground cable tape is
the best for bad bio-fouling.
Wrap with copper tape (again, beware of aluminum).
Coat with zinc oxide (e.g., Desitin ointment).
Remember that growth is sensor, depth, location, and season dependent.
Plan for routine changing or cleaning of sensor as necessary.
Check with calibration facility on which anti-foulants will be handled (allowed) by the calibrators.
Avoid or isolate dissimilar metals.
Maintain sacrificial anodes and ensure they are properly installed (good electrical contact).
Maximize use of non-metallic components.
Use UV-stabilized components that are not subject to sunlight degradation.

A-2

Temperature and Salinity

A.4

Common QA Considerations

The following lists suggest ways to ensure QA by using specific procedures and techniques:








Pre-deployment calibrations on every sensor
Post-deployment calibrations on every sensor, plus in-situ comparison before recovery
Periodic calibration of ready-to-use spares
Monitor with redundant sensors whenever possible
Take photos of sensor fouling for records
Record all actions related to sensors – calibration, cleaning, deployment, etc.
Monitor battery voltage and watch for unexpected fluctuations

When evaluating which instrument to use, consider these factors:











Selection of a reliable and supportive manufacturer and appropriate model
Operating range (i.e., some instruments won’t operate at a certain temperature, depth or pressure range)
Resolution/precision required
Sampling frequency – how fast sensor can take measurements
Reporting frequency – how often the sensor reports the data
Response time of the sensor – sensor lag – time response
Instrument check – visual inspection for defects, bio-fouling, etc.
Power check – master clock, battery, etc.
Standardize sensor clock to a reference such as GPS timing
Capability to reveal a problem with data

When evaluating which specifications must be met:





State the expected accuracy
Determine how the sensor compares to the design specifications
Determine if the sensor meets those specifications
Determine whether result is good enough (fit for purpose: data are adequate for nominal use as
preliminary data)

General comments regarding QA procedures:




A diagram (http://www.ldeo.columbia.edu/~dale/dataflow/), contributed by Dale Chayes (LDEO)
provides a visual representation of proper QA procedures.
Require serial numbers and model ID from the supplier.
Develop a checklist; do not make it so detailed that it will not be used.



Do not assume the calibration is perfect (could be a calibration problem rather than a sensor
problem).





Keep good records of all related sensor calibrations and checks (e.g., temperature).
Use NIST-traceable instrumentation when conducting calibrations or calibration checks.
A sensor that maintains an internal file of past calibration constants is very useful since it can be
downloaded instead of transcribed manually (which introduces human error).

The calibration constants or deviations from a standard should be plotted over time to determine if the
sensor has a drift in one direction or another. A sudden change can indicate a problem with the sensor or the
last calibration.

A-3

A.5

QA Levels for Best Practices

A wide variety of techniques are used by operators to assure that sensors are properly calibrated and
operating within specifications. While all operators must conduct some form of validation, there is no need to
force operators to adhere to one single method. A balance exists between available resources, level of
proficiency of the operator, and target data reproducibility requirements. The various techniques span a range
of validation levels and form a natural hierarchy that can be used to establish levels of certification for
operators (table A-1). The lists in the following sections suggest ways to ensure QA by using specific
procedures and techniques.
Table A-1. Best practices indicator for QA

QA Best Practices
Indicator

Description

Good Process

Sensors are swapped and/or serviced at sufficient regular intervals.
Sensors are pre- and post-deployment calibration checked.

Better Process

Good process, plus an overlapping operational period during sensor
swap-out to demonstrate continuity of observations.

Best Process

Better process, and follow a well-documented protocol or alternative
sensors to validate in-situ deployments. Or, the better process
employing manufacturer conducted pre- and post-calibrations.

A.6

Additional Sources of QA Information

TS sensor operators also have access to other sources of QA practices and information about a variety of
instruments. For example, the Alliance for Coastal Technologies (ACT) serves as an unbiased, third party test
bed for evaluating sensors and platforms for use in coastal and ocean environments. ACT conducts
instrument performance demonstrations and verifications so that effective existing technologies can be
recognized and promising new technologies can become available to support coastal science, resource
management, and ocean observing systems (ACT 2012). The NOAA Ocean Systems Test and Evaluation
Program (OSTEP) http://tidesandcurrents.noaa.gov/ostep.html also conducts independent tests and
evaluations on emerging technology as well as new sensor models. Both ACT and OSTEP publish findings
that can provide information about QA, calibration, and other aspects of sensor functionality. The following
list provides links to additional resources on QA practices.







Manufacturer specifications and supporting Web pages/documents
QARTOD http://www.ioos.noaa.gov/qartod/
ACT http://www.act-us.info/
NOAA CO-OPS - http://tidesandcurrents.noaa.gov/pub.html under the heading Manuals and Standards
WOCE http://woce.nodc.noaa.gov/wdiu/
NOAA NDBC http://www.ndbc.noaa.gov/

A-4

Temperature and Salinity

A.7

Example Deployment Checklists

The following samples provide hints for development of deployment checklists taken from QARTOD IV:

Pre-deployment QA Checklist
 Read the manual.
 Establish, use, and submit (with a reference and version #) a documented sensor preparation
procedure (protocol). Should include cleaning sensor according to the manufacturer’s procedures.
 Calibrate sensor against an accepted standard and document (with a reference and version #).
 Compare the sensor with an identical, calibrated sensor measuring the same thing in the same area (in
a calibration lab).
 View calibration specifications with a critical eye (don’t presume the calibration is infallible). Execute
detailed review of calibrated data.
 Check the sensor history for past calibrations, including a plot over time of deviations from the
standard for each (this will help identify trends such a progressively poorer performance). Control
chart calibrations.
 Check the sensor history for past repairs, maintenance, and calibration.
 Consider storing and shipping information before deploying.
o Heat, cold, vibration, etc.
 Provide detailed documentation.
 Record operator/user experiences with this sensor after reading the manual.
 Search the literature for information on your particular sensor(s) to see what experiences other
researchers may have had with the sensor(s).
 Establish and use a formal pre-deployment checklist.
 Ensure that technicians are well-trained. Use a visual tracking system for training to identify those
technicians who are highly trained and then pair them with inexperienced technicians. Have a data
quality review chain.

Deployment Checklist













Scrape bio-fouling off platform.
Verify sensor serial numbers.
Deploy and co-locate multiple sensors (attention to interference if too close).
Perform visual inspection; take photos if possible (verify position of sensors, connectors, fouling,
and cable problems).
Verify instrument function at deployment site prior to site departure. Allot sufficient time for
temperature equilibration.
Monitor sensors for issues (freezing, fouling).
Automate processing so you can monitor the initial deployment and confirm the sensor is working
while still on-site.
Specify date/time for all recorded events. Use GMT or UTC.
Check software to ensure that the sensor configuration and calibration coefficients are correct. Also
check sampling rates and other timed events, like wiping and time averaging.
Visually inspect data stream to ensure reasonable values.
Compare up and down casts and/or dual sensors (if available).
Note weather conditions and members of field crew.

A-5

Post-deployment Checklist









Take pictures of recovered sensor as is for metadata.
Check to make sure all clocks agree or, if they do not agree, record all times and compare with NIST.
Post-calibrate sensor and document before and after cleaning readings.
Perform in-situ side by side check using another sensor.
Provide a mechanism for feedback on possible data problems and/or sensor diagnostics.
Clean and store the sensor properly or redeploy.
Visually inspect physical state of instrument.
Verify sensor performance by:
o Checking nearby stations;
o Making historical data comparisons (e.g., long-term time-series plots, which are particularly
useful for identifying long-term bio-fouling or calibration drift.)

A-6

Temperature and Salinity

Appendix B.

QARTOD TS Manual Team

Version 2.0 Temperature/Salinity Manual Team
Name

Organization

Mark Bushnell, Lead Editor
Helen Worthington, Editor
Janet Fredericks
Stephanie Jaeger
Mathias Lankhorst
Igor Shkvorets
Julie Thomas

CoastalObsTechServices LLC/CO-OPS
REMSA/CO-OPS
Woods Hole Oceanographic Institution
Sea-Bird Electronics, Inc.
OceanSITES/SIO
RBR Ltd.
SIO/CDIP

B-1

Version 1.0 Temperature/Salinity Manual Committee
Name

Organization

Mark Bushnell, Chair
Ray Toll, Editor
Helen Worthington, Editor
Mathieu Belbeoch
Fred Bingham
Julie Bosch
Rich Bouchard
Richard Butler
Paul Devine
Janet Fredericks
Gustavo Goni
Karen Grissom
Bill Harrington
Bob Heitsenrether
Chris Heyer
Fred Holloway
Matt Howard
Betty Howell
Leonid Ivanov
Carol Janzen
Bob Jensen
Scott Kendall
Sung-Chan Kim
Josh Kohut
Matthias Lankhorst
Mike Lizotte (retired)
Hank Lobe
Rick Lumpkin
Bryan Mensi
Terry Merrifield
Ellyn Montgomery
Chris Paternostro
Dan Rudnick
Claudia Schmid
Igor Shkvorets
Vembu Subramanian
Charles Sun
Julie Thomas
Doug Wilson
Fabian Wolk

CoastalObsTechServices LLC/CO-OPS
Old Dominion University
REMSA/CO-OPS
JCOMMOPS
SECOORA/UNCW
NOAA/NCDDC (now NCEI)
NOAA/NDBC
Aanderaa
Teledyne RD Instruments
Woods Hole Oceanographic Institution
NOAA/AOML
NOAA/NDBC
Hach
NOAA/CO-OPS
Aquatic Informatics
Greenspan
GCOOS/Texas A & M University
NAVOCEANO
Woods Hole Group
AOOS
U.S. Army Corps of Engineers
GLOS
U.S. Army Corps of Engineers
MARACOOS/Rutgers University
OceanSITES/Scripps Institution of Oceanography
YSI
Severn Marine Technology
NOAA/AOML
Naval Oceanographic Office
In-Situ
USGS
CO-OPS
University of California San Diego
NOAA/AOML
RBR Ltd.
SECOORA
NODC
SIO/CDIP
Caribbean Wind LLC/MARACOOS
Rockland Scientific International, Inc.

B-2

Temperature and Salinity

Temperature/Salinity Manual Reviewers
Name

Organization

Charly Alexander
Rob Bassett
Fred Bingham
Julie Bosch
Rich Bouchard
Gisbert Breitbach
Janet Fredericks
Grace Gray
Bob Heitsenrether
Scott Kendall
Josh Kohut
Matthias Lankhorst
Jim Manning
Ellyn Montgomery
Jennifer Patterson
Jim Potemra
Sylvie Pouliquen
Claudia Schmid
Igor Shkvorets
Vembu Subramanian
Charles Sun
Julie Thomas

NOAA
NOAA/CO-OPS
UNCW
NCDDC
NOAA
Institute of Coastal Research/Germany
WHOI
NOAA/CO-OPS
NOAA/CO-OPS
GVSU
Rutgers University
OceanSITES/SIO
NERACOOS
USGS
CeNCOOS
PacIOOS
IFREMER/ France
NOAA/AOML
RBR Ltd.
SECOORA
NOAA
SIO/CDIP

QARTOD Board of Advisors
Name

Organization

Joe Swaykos - Chair
Kathy Bailey
Julie Bosch
Eugene Burger
Janet Fredericks
Matt Howard
Bob Jensen
Chris Paternostro
Julie Thomas

NOAA/NDBC
U.S. IOOS
NOAA/NCEI
NOAA/PMEL
Woods Hole Oceanographic Institution
GCOOS
U.S. Army Corps of Engineers
NOAA/CO-OPS
SIO/CDIP

B-3

DMAC Committee
Name

Organization

Rob Bochenek
Eric Bridger
Jorge Capella
Jeremy Cothran
Matt Howard
Kelly Knee
Emilio Mayorga
Jennifer Patterson
Jim Potemra
Rob Ragsdale
Tad Slawecki
Derrick Snowden
Shane StClair
Vembu Subramanian
Darren Wright

AOOS/Axiom Consulting & Design
NERACOOS/GMRI
CariCOOS/University of Puerto Rico
SECOORA/University of South Carolina
GCOOS/Texas A&M University
MARACOOS/RPS ASA
NANOOS/University of Washington
CeNCOOS/MBARI
PacIOOS/University of Hawaii
U.S. IOOS
GLOS/LimnoTech
U.S. IOOS
AOOS/Axiom Consulting & Design
SECOORA
SCCOOS/SIO

U.S. IOOS Regional Associations
Name

Organization

Josie Quintrell
David Anderson
Debra Hernandez
Barbara Kirkpatrick
Gerhard Kuska
Molly McCammon
Julio Morell
Ru Morrison
Jan Newton
Chris Ostrander
Kelli Paige
Julie Thomas

IOOS Association
CeNCOOS
SECOORA
GCOOS
MARACOOS
AOOS
CariCOOS
NERACOOS
NANOOS
PacIOOS
GLOS
SCCOOS

B-4

Source Exif Data:

File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.5
Linearized                      : Yes
Author                          : Helen Worthington
Create Date                     : 2016:01:06 15:54:35-05:00
Modify Date                     : 2018:10:18 14:45:49-07:00
Language                        : en-US
Tagged PDF                      : Yes
XMP Toolkit                     : Adobe XMP Core 5.6-c015 84.159810, 2016/09/10-02:41:30
Format                          : application/pdf
Creator                         : Helen Worthington
Creator Tool                    : Microsoft® Word 2013
Metadata Date                   : 2018:10:18 14:45:49-07:00
Producer                        : Microsoft® Word 2013
Document ID                     : uuid:2e9bb24d-759e-4137-9b9e-e7b081f0a522
Instance ID                     : uuid:a7691f8d-3f4c-40bb-827f-e31e9fecf027
Page Count                      : 56

EXIF Metadata provided by EXIF.tools

Qartod Temperature Salinity Manual

Navigation menu

Versions of this User Manual:

Views

Navigation