The maps have been generated using geographical information system (GIS) software from values of the long-term average calculated for each station in the Met Office's observing network. The station averages are interpolated on to a regular grid which is then used to produce the final map. The interpolation scheme uses topographic data to capture the effects of altitude, terrain shape and proximity to the coast and to urban areas.

Description of the analysis methods used to generate the averages

The new maps are based on grids with a resolution of 1 km (compared with 5 km for the previous 1961-1990 maps). The use of solid colour shading (rather than contours) is designed to reveal as much of the detail in these grids as can reasonably be achieved using a static A4-sized image. However this is not intended to imply that all spatial variations at a scale of 1 km have been portrayed. Where there are large gradients in the climatological average the variations are now better resolved (e.g. in upland and coastal areas). However, the quality of the gridded data is still limited by the interpolation scheme and certain features, such as frost hollows, are known to be captured imperfectly.

A total of 442 maps have been produced (twelve monthly, four seasonal and one annual map for each of 13 climate variables and each of two averaging periods). It was not possible to produce such a large number of maps efficiently without a fair degree of automation and this has influenced some of the presentational aspects of the maps.

For example, the number of data classes is fixed and the boundaries between the classes are chosen automatically, such that each colour occupies an approximately equal area of the map. Although equal-interval classes are more traditional, it was found that these often resulted in maps that had large areas of lowland Britain in a single colour. Given that the interpolation process will generally be more accurate in lowland areas than in topographically-complex upland areas, it was decided to use equal-area classes as this helps to bring out the detail in those areas where the interpolation is most robust.

A feature of equal-area classes is that in some cases the top and bottom classes span quite wide ranges of values while the classes in between encompass much smaller ranges of values (in some cases just a single value). This is simply a reflection of the statistical distribution of the data. In practice the number of grid points with values in the tails of the distribution will generally be very small. For example, although the grid point values of the 1961-1990 average maximum temperature for February vary between -3.5°C and +9.6°C, over 90% of the values lie in the much narrower range of +3.0°C to +8.0°C.

The use of equal-area classes also means that the boundaries between classes are arbitrary values which differ from map to map. As a consequence, it has been necessary to use a relative colour scheme, where the same set of colours is used on all maps (for a given climate variable) but individual colours correspond to different ranges of values on different maps.

Note that these climate maps are intended to be no more than a guide to the climate of the UK. Averages for selected UK stations, districts and regions are also available from our web site. Additional averages (including gridded versions) or bespoke versions of the maps and any associated charges can be obtained by contacting our 24-hour Customer Centre.