Where the information comes from and the techniques we have used to analyse it


Most of the information used in this study is from 1961 until the end of 2004, although there are exceptions to this. In particular, we have much longer records for temperature, rainfall and sunshine. To assess whether we can see trends in records of measure of climate for Scotland, we try to use as long a record as possible. On the other hand, using a consistent period of history for all variables allows us to produce a comprehensive picture of change, and to identify any links between variables. For this reason, we analysed all information for the period 1961 to 2004 (or less if the records are not available throughout the full period). However, if we have records covering a longer period, we also analysed these and present them alongside the 1961 to 2004 analysis. If we did not have information or it is difficult to interpret because of irregularities in the information (for example, a change in instrumentation or observing site) we used individual site records.


The Met Office has a historical database containing observations of different weather elements. These observations come from a network of meteorological stations across the United Kingdom which is constantly evolving. From this information we have produced a consistent series of climatic statistics which allows us to compare weather and climate across space and time. In order to do this, methods were developed to create gridded datasets from the information gathered at each station.


The process for analysing this information applied geographical information systems (GIS) capabilities and included a range of factors such as altitude, terrain shape and proximity to sea or urban areas. The gridded results were also checked with values we have from observations. The full method is described in Perry and Hollis (2005a, 2005b). Although the gridded datasets have been checked using independent observations, they are the result of interpolation techniques, and so you should treat the maps of trends from this dataset only as an estimate of local patterns of change.


The Met Office dataset is notable for the range of elements included. The Met Office have produced gridded datasets at 5km by 5km resolution over the UK for 36 monthly or annual climate variables, for the period 1961 to 2000. The Met Office chose the start date of 1961 because there is a significant increase in the availability of digitised information from this point. A number of these variables are routinely updated with the latest month while we have updated several other variables to 2004 for this study.


The Met Office have used all available monthly meteorological information to make best use of the information available and make sure that we can make the most accurate possible representation of the climate for each month. As a result, the network of stations used changes slightly each month, and the methods we have used are designed to reduce, as far as possible, the effect of these changes on the consistency of the datasets through time. Table A1 shows, by variable, the average number of stations included in a month of the dataset. We also give the average number of stations for the variables available before 1961, clearly showing the increase in digitised information since this time.


Table A1 - Average number of stations included each month in the gridded dataset.


Climate variable Before 1961 1961 onwards
Precipitation 102 740
Days of 10 mm or more rain n/a 693
Rainfall intensity / greatest five-day rainfall n/a 458
Air temperature 76 158
Consecutive dry days each year n/a 145
Extreme temperature range each year n/a 141
Heating and growing degree-days n/a 126
Days of snow cover n/a 106
Sunshine 59 75
Length of heat-waves and cold spells n/a 52
Average sea level pressure / cloud n/a 19


We can estimate the values of climate variables between observing stations to a good degree of accuracy, producing detailed and representative maps of the Scottish climate. We can investigate changes in patterns over time as well as trends in climate using the processed information. However, the accuracy depends on the nature of the variable plus how many stations there are providing information for an area. Inaccuracies occur most often in areas where there are few stations, particularly the highlands of Scotland, which are also areas of complicated mountainous terrain. The process does not take account of localised effects on climate such as frost hollows, and effects caused by the type of soil and forests.