6.6: Measurement of Temperature

Temperature in the ocean is measured many ways. Thermistors and mercury thermometers are commonly used on ships and buoys. These are calibrated in the laboratory before being used, and after use if possible, using mercury or platinum thermometers with accuracy traceable to national standards laboratories. Infrared radiometers on satellites measure the ocean’s surface temperature.

Mercury Thermometer

This is the most widely used, non-electronic thermometer. It was widely used in buckets dropped over the side of a ship to measure the temperature of surface waters, on Nansen bottles to measure sub-sea temperatures, and in the laboratory to calibrate other thermometers. Accuracy of the best thermometers is about \(\pm 0.001^{\circ} \text{C}\) with very careful calibration.

One very important mercury thermometer is the reversing thermometer (figure \(\PageIndex{1}\)) carried on Nansen bottles, which are described in the next section. It is a thermometer that has a constriction in the mercury capillary that causes the thread of mercury to break at a precisely determined point when the thermometer is turned upside down. The thermometer is lowered deep into the ocean in the normal position, and it is allowed to come to equilibrium with the water. Mercury expands into the capillary, and the amount of mercury in the capillary is proportional to temperature. The thermometer is then flipped upside down; the thread of mercury breaks, trapping the mercury in the capillary; and the thermometer is brought back. The mercury in the capillary of the reversed thermometer is read on deck along with the temperature of a normal thermometer, which gives the temperature at which the reversed thermometer is read. The two readings give the temperature of the water at the depth where the thermometer was reversed.

Figure \(\PageIndex{1}\): Left: The constricted part of the capillary in set and reversed positions. Right: Protected and unprotected reversing thermometers is set position, before reversal. After von Arx (1962: 259).

The reversing thermometer is carried inside a glass tube which protects the thermometer from the ocean’s pressure because high pressure can squeeze additional mercury into the capillary. If the thermometer is unprotected, the apparent temperature read on deck is proportional to temperature and pressure at the depth where the thermometer was flipped. A pair of protected and unprotected thermometers gives temperature and pressure of the water at the depth the thermometer was reversed.

Pairs of reversing thermometers carried on Nansen bottles were the primary source of sub-sea measurements of temperature as a function of pressure from around 1900 to 1970.

Platinum Resistance Thermometer

This is the standard for temperature. It is used by national standards laboratories to interpolate between defined points on the practical temperature scale. It is used primarily to calibrate other temperature sensors.

Thermistor

A thermistor is a semiconductor having resistance that varies rapidly and predictably with temperature. It has been widely used on moored instruments and on instruments deployed from ships since about 1970. It has high resolution and an accuracy of about \(\pm 0.001^{\circ} \text{C}\) when carefully calibrated.

Bucket temperatures

The temperature of surface waters has been routinely measured at sea by putting a mercury thermometer into a bucket which is lowered into the water, letting it sit at a depth of about a meter for a few minutes until the thermometer comes to equilibrium, then bringing it aboard and reading the temperature before water in the bucket has time to change temperature. The accuracy is around \(0.1^{\circ} \text{C}\). This is a very common source of direct surface temperature measurements.

Ship Injection Temperature

The temperature of the water drawn into the ship to cool the engines has been recorded routinely for decades. These recorded values of temperature are called injection temperatures. Errors are due to ship’s structure warming water before it is recorded. This happens when the temperature recorder is not placed close to the point on the hull where water is brought in. Accuracy is \(0.5^{\circ} - 1^{\circ} \text{C}\).

Advanced Very High Resolution Radiometer

The most commonly used instrument to measure sea-surface temperature from space is the Advanced Very High Resolution Radiometer (AVHRR). The instrument has been carried on all polar-orbiting meteorological satellites operated by NOAA since Tiros-N was launched in 1978.

The instrument was originally designed to measure cloud temperatures and hence cloud height. The instrument had, however, sufficient accuracy and precision that it was soon used to measure regional and global temperature patterns at the sea surface.

The instrument is a radiometer that converts infrared radiation into an electrical voltage. It includes a mirror that scans from side to side across the sub-satellite track and reflects radiance from the ground into a telescope; a telescope that focuses the radiance on detectors; detectors sensitive to different wavelengths that convert the radiance at those wavelengths into electrical signals; and electronic circuitry to digitize and store the radiance values. The instruments observes a 2700-km wide swath centered on the sub-satellite track. Each observation along the scan is from a pixel that is roughly one kilometer in diameter near the center of the scan and that increases in size with distance from the sub-satellite track.

The radiometers measures infrared radiation emitted from the surface in five wavelength bands: three infrared bands: \(3.55-3.99 \ \mu\text{m}\), \(10.3-11.3 \ \mu \text{m}\), and \(11.5-12.5 \ \mu \text{m}\); a near-infrared band at \(0.725-1.10 \ \mu \text{m}\); and a visible-light band at \(0.55-0.90 \ \mu\text{m}\). All infrared bands include radiation emitted from the sea and from water vapor in the air along the path from the satellite to the ground. The \(3.7 \ \mu\text{m}\) band is least sensitive to water vapor and other errors, but it works only at night because sunlight has radiance in this band. The two longest wavelength bands at \(10.8 \ \mu\text{m}\) and \(12.0 \ \mu\text{m}\) are used to observe sea-surface temperature and water vapor along the path in daylight.

Data with 1-km resolution are transmitted directly to ground stations that view the satellite as it passes the station. This is the Local Area Coverage mode. Data are also averaged to produce observations from 4 × 4 km pixels. These data are stored by the satellite and later transmitted to NOAA receiving stations. This is the Global Area Coverage mode.

The swath width is sufficiently wide that the satellite views the entire earth twice per day, at approximately 09:00 AM and 9:00 PM local time. Areas at high latitudes may be observed as often as eight or more times per day.

The most important errors are due to:

1. Unresolved or undetected clouds: Large, thick clouds are obvious in the images of water temperature Thin clouds such as low stratus and high cirrus produce much small errors that are difficult or almost impossible to detect. Clouds smaller in diameter than 1 km, such as trade-wind cumuli, are also difficult to detect. Special techniques have been developed for detecting small clouds (figure \(\PageIndex{1}\)).
2. Water vapor, which absorbs part of the energy radiated from the sea surface: Water vapor reduces the apparent temperature of the sea surface. The influence is different in the \(10.8 \ \mu\text{m}\) and \(12.0 \ \mu\text{m}\) channels, allowing the difference in the two signals to be used to reduce the error.
3. Aerosols, which absorb infrared radiation. They radiate at temperatures found high in the atmosphere. Stratospheric aerosols generated by volcanic eruptions can lower the observed temperatures by up to a few degrees Celsius. Dust particles carried over the Atlantic from Saharan dust storms can also cause errors.
4. Skin temperature errors. The infrared radiation seen by the instrument comes from a layer at the sea surface that is only a few micrometers thick. The temperature in this layer is not quite the same as temperature a meter below the sea surface. They can differ by several degrees when winds are light (Emery and Schussel, 1989). This error is greatly reduced when AVHRR data are used to interpolate between ship measurements of surface temperature.

Figure \(\PageIndex{1}\): The influence of clouds on infrared observations. Left: The standard deviation of the radiance from small, partly cloudy areas each containing 64 pixels. The feet of the arch-like distribution of points are the sea-surface and cloud-top temperatures. After Coakley and Bretherton (1982). Right: The maximum difference between local values of \(T_{11} - T_{3.7}\) and the local mean values of the same quantity. Values inside the dashed box indicate cloud-free pixels. \(T_{11}\) and \(T_{3.7}\) are the apparent temperatures at \(11.0\) and \(3.7 \ \mu\text{m}\) (data from K. Kelly). After Stewart (1985: 137).

Maps of temperature processed from Local Area Coverage of cloud-free regions show variations of temperature with a precision of \(0.1^{\circ}\text{C}\). These maps are useful for observing local phenomena including patterns produced by local currents. Figure \(10.8.4\) shows such patterns off the California coast.

Global maps are made by the U.S. Naval Oceanographic Office, which receives the global AVHRR data directly from NOAA’s National Environmental Satellite, Data and Information Service in near-real time each day. The data are carefully processed to remove the influence of clouds, water vapor, aerosols, and other sources of error. Data are then used to produce global maps between \(\pm 70^{\circ}\) with an accuracy of \(\pm 0.6^{\circ}\text{C}\) (May et al 1998). The maps of sea-surface temperature are sent to the U.S. Navy and to NOAA’s National Centers for Environmental Prediction. In addition, the office produces daily 100-km global and 14-km regional maps of temperature.

Global Maps of Sea-Surface Temperature

Global, monthly maps of surface temperature are produced by the National Centers for Environmental Prediction using Reynolds et al (2002) optimal-interpolation method. The technique blends ship and buoy measurements of sea-surface temperature with AVHRR data processed by the Naval Oceanographic Office in \(1^{\circ}\) areas for a month. Essentially, AVHRR data are interpolated between buoy and ship reports using previous information about the temperature field. Overall accuracy ranges from approximately \(\pm 0.3^{\circ}\text{C}\) in the tropics to \(\pm 0.5^{\circ}\text{C}\) near western boundary currents in the northern hemisphere where temperature gradients are large. Maps are available from November 1981. Figures \(6.3.2-6.3.4\) were made by NOAA using Reynolds’ technique. Other data sets have been produced by the NOAA/NASA Pathfinder program (Kilpatrick, Podesta, and Evans, 2001).

Maps of mean temperature have also been made from ICOADS data (Smith and Reynolds, 2004). Because the data are poorly distributed in time and space, errors also vary in time and space. Smith and Reynolds (2004) estimated the error in the global mean temperature and found the 95% confidence uncertainty for the near-global average is \(0.48^{\circ}\text{C}\) or more in the nineteenth century, near \(0.28^{\circ}\text{C}\) for the first half of the twentieth century, and \(0.18^{\circ}\text{C}\) or less after 1950. Anomalies of sea-surface temperature were calculated using mean sea-surface temperature from the period 1854–1997 using ICOADS supplemented with satellite data since 1981.