Temperature

Temperature#

The “Background” section summarises typical characteristics of temperature encountered in the ocean : the temperature range, vertical profile, and spatial and temporal scales encountered in the ocean. Temperature measurements can be taken by “Contact sensing” (using sensors such as thermistors, thermocouple and RTDs (resistance temperature detectors)) or with “Remote sensing” by measuring thermal radiation (e.g. satellite measurements or thermal cameras).

The “Historical and high-technology sensors” section presents technologies no longer used today or that are not suitable to DIY approache. Finally, a table of sensors and instruments available on the market is provided in the “Sensors databse” section.

Contributions:

Louis Marié, Aurélien Ponte, Adèle Moncuquet

Background#

Ocean temperature ranges between -2°C and 40°C globally. It controls the evolution of ocean circulation, climate systems, and marine habitats. Temperature determines water density (together with salinity and pressure) and hence conditions the oceanic circulation [section 1.2.4 Siedler 2001]. It reflects the amount of energy stored and absorbed by oceans [Cheng et al. 2020].

You can measure temperature for a variety of purposes and key climate indicators directly linked to temperature measurements include sea surface temperature (SST), ocean heat content, and indicators of marine heat waves. Discover typical variations of temperature depending on location, height above sea floor and time and the evolution of temperature standard! Use this section to warm-up before learning about temperature sensors technology!

Now you know what to expect in terms of temperature measurement wherever you are! (Or close enough). You should understand that temperature fluctuate at various time and spatial scales. So make sure you clearly define what you want to look at! At this point you can decide what you want to learn from temperature measurements. The next two sections provide simplified theoretical insights into the different sensing elements to help you master the technology and understand its advantages and limitations.

Contact sensing#

Contact thermal measurements mainly use an electric component which react to temperature changes. These sensors are inexpensive and compact. There are three main types:

Thermistors are semiconductors whose resistance changes with temperature. They are commonly used in oceanography due to their good sensitivity. However, the resistance variation with respect to temperature is not linear. See section Thermistor.

Thermocouples produce an electrical current by Seebeck effect which depends on the temperature environment. Thermocouples are less used then thermistors because of the complex electrical circuit involved. However, they are robust and have a fast time response (around 10 ms). See section Thermocouple.

RTDs are metallic wires whose resistance changes with temperature (like the PT100). The resistance variations are small, so they require a more sensitive acquisition system tend than the thermistor. See section RTD.

Thermistor#

The word “thermistor” derives from the description “thermally sensitive resistor”. It was first discovered by M. Faraday in 1833. At this time, mass production of thermistors was challenging, so commercial production only started in the 1930s. The development of high-performance semiconductors in the late 20th century significantly increased sensor sensitivity and precision, as well as expanded the operational temperature range. Today, thermistors are used for a wide range of applications. They are the main type of temperature sensor for oceanographic measurements due to their good precision and low cost.

Thermocouple#

Thermocouples produce an electrical signal due to temperature changes that can be transmitted and processed electronically. Thermocouples are inexpensive and easy to construct (see video How to make a Thermocouple or the blog Making a thermocouple).

Thermocouples are robust sensors that can operate over a wide temperature range and be smaller than a few micrometres. However, any change in lattice structuree or impurity modifies the amount of generated voltage making the sensor less accurate [Pavlasek et al. 2015].

They are typically employed in industrial processes as well as in healthcare applications [Zhang et al. 2022 ; Leonidas et al. 2022]. In the metallurgy industry, thermocouples are slower and less precise than the infrared thermometers and fibre obtics [Leonidas et al. 2022] and drift (0.22°C per hour above 100°C - [Pavlasek et al. 2015]).

In oceanography, thermocouples offer an interesting spatial resolution and a fast response times (in the order of the milliseconds) compared to thermistors and to RTDs. They are well suited to use as microstructure sensors at sites with pronounced thermal stratification [Nash et al. 1999]. Currently, no studies using thermocouples for long-term measurements have been found. Despite the need for complex electronic development, thermocouples are suggested as an interesting sensor to develop in oceanography. Contributions are welcome.

../_images/temperature_thermocouple_prez.png — Type K Exposed Junction Thermocouple from RS Pro#

RTD#

RTD stands for Resistance Temperature Detector. It differs from a thermistor in terms its composition and behaviour. RTDs are made of pure metal, the resistance of which changes linearly with temperature. This contact sensor offers relatively good accuracy (around 0.2°C) over time and over a large range of temperatures (usually used between -50°C and 50°C) but it does generate self-heating. RTDs are most commonly used in atmospheric sciences [Lin & Hubbard 2004].

../_images/temperature_Pt100.png — PT100 RTD Sensor from RS Pro#

Remote sensing#

Remote sensing measures sea surface temperature (SST) without physical contact. All thermal instruments use the principle of thermal radiation. Different sensor types are based on this principle:

Radiometers
Thermal cameras
Infrared thermometers

Usually, thermal cameras and pyrometers output a temperature with low accuracy (the best I found so far is ±2°C for a thermal camera around 500€). However, combined measurements and dedicated post-processing may significantly increase accuracy [Robinson & Donjon 2010], Zhang & Wang 2022. Remote sensing allows the measurement of skin temperature (0–500 μm below the surface; The thermal stucture within the first few micrometres below the sea surface is complex and different methods of measuring the SST may record different values. Furthermore, object smaller than the IR wavelength may not be approximated by the black body approximation Cuevas 2019). Therefore the following text may not apply to the SST_skin measurement and contributions on SST_skin are more than welcome.

Radiometers measure radiance over several thin spectral bands and provide radiation output. Radiometers are expensive and relatively massive instruments used on satellites or on Earth to measure sea and atmospheric properties. More information about radiometers can be found on the Apogee instrument website and in the Flir website. Radiometers are not considered in the following sections; contributions are more than welcome.

This section first presents the “General principle” of thermal radiation and the “measurement process”. Then we describe the two main affordable instruments: “Thermal camera” and “Infrared thermometer”.

▶ General principle of thermal radiation

Remote temperature measurement aims to capture the intensity of long wavelength emissions from an observed body. Matter naturally emits electromagnetic radiation at different wavelengths which constitutes a spectrum (see Figure solar spectrum). The shape of the spectrum depends on the temperature of the body. The peak emission for instance shifts towards shorter wavelengths as temperature increases as described by Wien’s displacement law (see Figure solar spectrum). The sun, at a temperature of 5900 K, has a maximum luminous intensity in the “visible” wavelengths (from a human point of view), whereas SST whose temperature is between 2 and 32°C does not spontaneously emit in the visible range; but in the infrared and microwave wavelength range. A fun interactive simulation of the blackbody spectrum can be find in the Phet website from university of Colorado. Note that, as you decrease the temperature, you will have to reduce the \(y\) axis limit (i.e reduce the power density) and the \(x\) axis limit (i.e reduce the wavelength towards the IR) in order to accurately observe the shape of the spectrum.

The radiation captured by the instrument is a combination of what is emitted by the body of interest and the environment. When looking at the SST, measurements also capture what the atmosphere emits (and all the other bodies that may be present on the way). The atmosphere emits where it absorbs sunlight (see Figure solar spectrum). Usually, filters are chosen between 10.5–12.5 μm (atmospheric window) to avoid atmospheric absorption bands without being too broad. Once the radiation is corrected, one can compute the temperature associated with the emission using Stefan-Boltzmann’s Law. The approximations made at this stage are : 1) the SST mostly emits in the measured range 2)the spectrum is approximated by the black-body spectrum, described by Planck’s Law.

To better represent the spectrum of a real body, measurements can be taken across several bands (such as in satellite measurements). To better fit the black body spectrum to the real body spectrum more accurately, a correction factor called the emissivity is used. For instance, emissivity of the sea surface will depend on the angle of observation and the surface wind speed Masuda et al. 1988. More information can also be found in [DeWitt & Nutter 1988]. A complete course about atmospheric radiation written by J. Lefrere can be found here. For english readers [Bohren 2005] describe fundamentals of atmospheric radiation with plenty of exercices.

../_images/temperature_Solar_spectrum_en.png — (left) Solar spectrum above the atmosphere (yellow) and at the surface (red), Black body spectrum at the sun’s temperature (black curve) - Adapted from Solar spectrum by Robert A. Rohde. (right) Black body spectral radiance curve for various temperatures adapted from Black body illustration by Darth Kule - Own work, Public Domain#

Compared to contact sensing, thermal radiation is a non-intrusive measurement, which makes it ideal for SST measurement.

Thermal camera#

Thermal cameras cost around 160–1500€ and can be lightweight (153g), wireless, or plugged into smartphones. Cameras can be technically challenging, and training may be needed to fully handle them. Thermal sensor array can also be purchased see Melexis sensing element which drastically reduce the cost.

../_images/temperature_camera.png — IR camera from FLIR#

Infrared thermometer#

An infrared thermometer is an inexpensive alternative to measure SST at a single point. The measurement spot diameter depends on the distance-to-object ratio. The closer the sensor is to the surface, the smaller the measurement point (see instrument testo 830-T1 documentation for more information).

Instruments (such as pyrometers or medical IR thermometers) cost between 10-1000€. IR sensors can also be purchased separately and integrated into custom systems. Examples of IR sensors with 0.5°C accuracy (specified by the manufacturer) can be found for around 23 USD. A combination of measurements may help achieve better accuracy. [Zhang & Wang (2022)] propose a methodology to obtain SST measurements using integrated infrared thermometers, MLX90614 and Heitronics KT19. A more detailed methodology would be needed.

../_images/temperature_IR_sensor.jpg — Temperature and humidity sensor MLX90614ESF-BCC-000-TU by Melexis website#

Historical and high-technology sensing techniques#

This section presents a non-exhaustive list of techniques that are either no longer used or represent high-tech solutions that are not well suited to DIY instruments. These techniques are included to privde insight into the evolution of oceanographic measurements techniques, and because some of them may become relevant again for low-tech development. Any contributions to completing these descriptions are more than welcome.

The following methods are covered:

Liquid dilatation: Mercury and alcohol thermometers used in early oceanography
Bimetallic strips: Temperature sensors based on differential thermal expansion of metals
Bathythermograph: Historical profiling instruments from the 1930s-1970s
Fiber optic sensing: Advanced optical techniques for subsurface temperature measurement
Sound velocity: Acoustic methods for indirect temperature measurement

▶ BathyThermographs

The bathythermograph was developed in 1938 and extensively used from the 1930s to 1970s for measuring temperature profiles with depth. For the first time, an instrument recorded continuous profiles rather than discrete point measurements. During WWII, BT measurements were used to predict sonar range. More than 1,300,000 temperature-depth profiles have been acquired using this method [Couper & Lafond 1970].

../_images/temperature_exploded_BT.jpeg — Mechanical BathyThermograph (MBT or BT) exploded view, Source: Neumann and Pierson (1966).#

The initial mechanical bathythermographs (Figure MBT) were then replaced by expendable bathythermographs (XBT) (Figure XBT). XBT allowed for the first synoptic view of the Japan/East Sea thermocline structure [Gordon et al. 2002]. Nowadays, XBT data are still used in oceanographic studies (see the NOAA/AOML XBT Network). XBTs are also used by volunteer observing ships [Sect 16.4.2.5 from Lynne 2011]. Similar objects were deployed in the atmosphere (see NOAA website).

../_images/temperature_exploded_XBT.gif — eXpendable BathyThermograph (MBT or BT) exploded view, Source: NOAA website.#

Principle

To measure temperature profiles, the MBT operates on mechanical principles using the thermal expansion of xylene in capillary tubing. These mechanical movements are transferred through a linkage system to a stylus that traces curves on a smoked glass slide. The glass slide was moved at right angles with respect to the stylus by a pressure-sensitive bellows, creating a permanent record of temperature versus depth (Figure MBT). The MBT accuracy was excellent for the time as it could reach ±0.2 K and ±2 m with good calibration [sect 16.4.2.4 Lynne 2011], but doubtfully down to 0.06°C which was sometimes stated [sect 1.3.2 Emery 2001].

../_images/temperature_BT_hazard.png — Illustration from the manufacturer’s instruction book - from [Couper et al. 1970].#

The MBT had numerous limitations. After retrieval, data were entered into files by hand and the opportunities for operator recording errors were considerable [Emery 2001]. The MBT was also restricted to depths less than 300 m and intended to be used with a ship traveling at a few knots maximum. The lowering and operating of the BT had to be done with care and loss of BTs was frequent (Figure BT hazard). The BT is a delicate instrument which would be easily damaged and danger points given in the instruction book were not to: “lower the BT while the ship is expected to turn;” “leave in a temperature greater than 105 °F;” “bump, drop or jar.” [Couper et al. 1970].

Nowadays the XBT uses a thermistor and electronic data acquisition system which reduced the MBT uncertainties and was able to go deeper (1500 m for XBT).

The different types of sensors

Standard MBT: Basic model for depths up to 285m
Deep MBT: Extended range version for depths up to 400m
Fine-structure MBT: Enhanced sensitivity for detecting small temperature variations
Expendable bathythermographs (XBT): Used thermistors which enhanced accuracy (around 0.1°C in laboratory) [Emery 2001].

Advantages and limitations

Advantages:

Continuous temperature-depth profiles
MBT : No electrical components to fail
Important historical dataset for climate studies

Limitations:

MBT : The data were entered into files by hand leading to opportunities for operator recording errors were considerable [Emery 2001]
Limited depth range
Ships must travel at no more than a few knots
Metal thermometers are subject to permanent deformation

References

You can access publications and book references via the library’s search tool.

Books

Lynne D. Talley, George L. Pickard, William J. Emery, James H. Swift, 2011. Chapter S16 - Instruments and Methods, Editor(s): Lynne D. Talley, George L. Pickard, William J. Emery, James H. Swift, Descriptive Physical Oceanography (Sixth Edition), Academic Press, Pages 1-83, ISBN 9780750645522, pdf.
W. J. Emery & R. E. Thomson, 2001. Data Analysis Methods in Physical Oceanography - Elsevier Science, Revised Second Edition, ISBN 978-0-444-50756-3

Scientific reports and publications

Couper, B. K. and Lafond, E. C, 1970. Mechanical bathythermograph, an historical review. Adv. Instrumn. Instrum. Soc. Am., 25, paper 735–70, PDF
Spilhaus, A. F., 1938: A bathythermograph. J. Mar. Res., 1, 95-100.
Gordon, A.L., C.F. Giulivi, C.M. Lee, H.H. Furey, A. Bower, and L. Talley. 2002. Japan/East Sea intrathermocline eddies. Journal of Physical Oceanography 32(6):1,960–1,974.

Websites

NOAA Historical Bathythermograph Data
https://www.aoml.noaa.gov/hrd/about_hrd/HRD-P3_exp.html?print=yes- Neumann, G. and Pierson, W.J. (1966) Principles of Physical Oceanography. Prentice-Hall, Englewood Cliffs, 230-233.

▶ Acoustic thermometry

Acoustic thermometry is widely used to measure atmospheric boundary layer characteristics or air characteristics in indoor environments [Arnold et al. 1999; Othmani et al. 2023]. In the ocean, acoustic thermometry allows for deep temperature measurements with 0.1°C precision at 500-1500 m [Woolfe et al. 2015]. The current description covers a range of different technologies. Any contributions with clarification and organization is welcome.

Thermometry is part of ocean acoustic tomography measurements, which were first developed based on theoretical work for ocean monitoring down to the mesoscale [Munk & Wunsch 1979; Behringer et al. 1982; ATOC technical report]. Thermometry initially required an acoustic emitter within the ocean, which was suspected to impact marine life [New York Times archive] but was demonstrated to be harmless to marine life, especially compared to boats [ATOC]. Later, it was demonstrated that passive measurements were sufficient to measure temperature changes [Woolfe et al. 2015]. Passive ocean acoustic thermometry (POAT) observations showed 0.007°C/year warming around 900 m over twenty years [Läslo et al. 2015]. Coastal thermometry is more challenging due to high-frequency changes in ocean characteristics and internal waves. Current work on vertical profiles of sound speed variations in coastal oceans presents measurements with 25 ms accuracy and 1 m vertical resolution even in the presence of linear and nonlinear internal waves [Tan et al. 2021].

Principle

Sound velocity in water changes with temperature, salinity, and pressure.

Measurement process

Acoustic sensors emit sound waves and measure their travel time or frequency shift. The sound velocity is calculated and correlated with temperature using established relationships.

The different types of methodology

Two main methodologies are used to measure sound variations in the ocean. The first method developed consists of measuring the propagation of acoustic signals between sources and receivers, refered to as Active Thermometry of Ocean Climate [ATOC]. The second method is Passive Ocean acoustic Thermometry which use ambient noise correlation [Woolfe et al 2015; Howe et al. 2019].