Accuracy of Observations
Observational accuracy is always under scrutiny as it is important to keep a level of trust between the Met Office and the public. As new technology is developed this accuracy must be enforced with each piece of new equipment.
Checking our instruments
Before purchasing meteorological instruments for regular use at a weather station we conduct detailed trials to establish their accuracy and reliability. Since every country in the world makes meteorological measurements there is a common interest in finding the best instruments. Over the years the World Meteorological Organization has played an active role in co-ordinating instrument trials.
Once the choice of instrument has been made we need to be sure that it continues to perform to its specification throughout the time it is in use. To ensure correct measurement, all instruments are calibrated at regular intervals, testing them against a standard instrument under controlled laboratory conditions. In this way the air temperature measurement has been linked to measurements from the national standard held at the National Physical Laboratory, an unbroken practice dating back to 1858. Similar calibrations are made for anemometers, humidity sensors and barometers.
For instruments where no national standard exists (e.g. instruments measuring radiation) the Met Office participates in international trials to establish a common measurement baseline against which national instruments may be compared.
Measuring air temperature
The most important factor for accurate air temperature measurement over the land surface is the choice of site. The environment should be representative of the larger area and should be unchanging over long periods of time. For this reason Met Office sites are almost invariably in rural locations well away from large buildings or other structures that might act as a large heat stores. An open location is chosen well away from trees or large shrubs where the dominant land type is grass or other vegetation. The ground under and surrounding the Stevenson screen is grass kept well cropped by a local contractor.
For the last 20 to 30 years air temperature has been measured electronically using platinum resistance thermometers, high precision instruments accurate to within a hundredth of a degree. Before then mercury thermometers were read at set times by trained meteorological observers.
A measurement of air temperature is only meaningful if it is made consistently from site to site in a way that ensures that sunlight and rain do not interfere. For this reason the thermometer is housed in a Stevenson screen which allows the free circulation of air while excluding the weather elements. Stevenson lived over a hundred years ago and screens have been used to his original design ever since. In recent times it was decided to construct the screen in plastic rather than wood, to achieve better long term wear. To investigate what effect, if any, such a change might have on the temperature measurement, a year long trial was conducted comparing measurements from the different screens sited two metres apart. Mean temperature differences were less than 0.05 °C over the whole period.
Measuring the upper air
Accurate measurements of the upper air are essential for monitoring long term changes in the climate and for acting as a baseline against which satellite measurements may be compared. Before each launch, check readings of temperature, humidity and pressure are made from the radiosondes used at each station. Any small departures from the reference readings are applied as corrections to all subsequent data during the flight. Where the departures are large, the radiosonde is returned to the manufacturer. International trials of radiosonde instruments are undertaken to understand the common sources of error. From these trials, the total uncertainty (2 sigma) for the instrument used in the UK is estimated as follows:
· Temperature (+60 to -90 °C) 0.5 °C
· Relative humidity (0 to 100%) 5 %
· Pressure (1080 to 100 hPa) 1 hPa
· Pressure (100 to 3 hPa) 0.6 hPa
· Wind speed 0.15 m/s
· Direction 2 degrees
· Height 10 m