HadUK-Grid Frequently Asked Questions
Further information about the UK gridded climate datasets
- How accurate are the grids?
- What quality control has been applied?
- How many observations were used to produce the grids?
- Why are there no data before 1961 for many variables?
- How often are new grids added to the website and when will data for the most recent years be added?
- Why are the monthly temperature (rainfall) grids not equal to the average (sum) of the corresponding daily grids?
- Why are the daily temperature maxima and minima for different 24-hour periods?
- The Met Office uses two types of sunshine recorder. How does this affect the sunshine grids?
- How are the monthly wind statistics calculated?
- Why is the period 1961 to 1990 used for the long-term averages?
- What advantages do gridded data have, compared to station data?
An indication of the quality and accuracy of the grid values is given by the root-mean-square Errors (RMSE) at verification stations. For the daily grids, the RMSE values averaged over each day of a test year are as follows:
|Climate variable||Verification RMSE|
|Rainfall amount||1.23 mm|
|Mean temperature||0.94 °C|
|Maximum temperature||1.06 °C|
|Minimum temperature||1.27 °C|
For the monthly grids, the average RMSE values for 12 test months are as follows:
|Climate variable||Verification RMSE|
|Rainfall amount||16 mm|
|Mean temperature||0.36 °C|
|Maximum temperature||0.66 °C|
|Minimum temperature||0.45 °C|
RMSE values for monthly grids of other climate variables and for annual grids are given in the paper 'Generation of monthly gridded data sets for a range of climatic variables over the UK' (PDF, 590 kb).
The station observations used to produce the grids have undergone a quality control process which corrects or removes erroneous data. An additional check comparing station observations to the existing UKCP09 gridded dataset was conducted to remove any remaining questionable historical data. No attempt was made to homogenise the input data. All the daily grids were checked to ensure that no erroneous station data had been used, as described in 'The generation of the daily gridded data sets of temperature and rainfall for the UK' (PDF, 2.37 MB).
The number of stations used as input to the gridding varies with time, partly due to changes in the size of the observing network and partly due to the availability of digitised data. A typical starting year for daily and monthly data is 1961, although some monthly data have been digitised further back (see Q4). During the period 1961 to 2000, the average number of stations used is shown in the following table.
|Climate variable||Average no. of stations|
|Pressure, cloud, wind speed||70|
The variation in the numbers of rainfall and temperature stations used from 1960 to 2006 is shown in the following graph.
To create gridded data sets, observations from an evenly distributed network of stations are needed in digital format. For many climate variables, sufficient data only exist in the Met Office electronic databases from 1961. However, as a result of an ongoing programme to digitise published station records, the monthly sunshine series extend back to 1929 and the monthly air temperature to 1884 and rainfall series back to 1862.
Daily and monthly data for most variables are updated routinely soon after the calendar end. These near real time data are provisional and ssubject to change so have limited distribution, please direct enquiries for these data to the customer centre. After the full data acquisition and quality control process the HadUK-Grid gridded and regional land surface climate observations dataset will be updated with the previous years data and will be hosted at the Centre for Environmental Data Archiving.
Why are the monthly temperature (rainfall) grids not equal to the average (sum) of the corresponding daily grids?
The monthly grids were prepared using monthly station data as input and then, later, daily grids were produced using daily station data as input. Hence these were separate interpolation exercises. Comparisons between the monthly grids and the average/sum of the daily grids were carried out, confirming that the largest differences tend to occur over upland areas. However, no adjustments have been made to make the two types of grid equivalent.
Conventionally, maximum and minimum temperatures are recorded for 24-hour periods ending at 0900 UTC each day. Maximum temperatures tend to occur during mid-afternoon, so the relevant maximum for a given calendar day is the one recorded between 0900 on the day in question and 0900 on the following day. However, minimum temperatures generally occur around dawn, so the relevant minimum temperature for a given calendar day is the one recorded between 0900 on the previous day and 0900 on the day in question.
The majority of the stations in the Met Office network that record sunshine duration use a Campbell-Stokes recorder. However, a significant proportion of the stations that report in real time now use a Kipp & Zonen sunshine sensor (the first such sensors were installed in 2000). The characteristics of the two instruments are not identical and therefore some adjustment of the observations to a common standard is required. Currently, Kipp & Zonen monthly totals are adjusted to Campbell-Stokes equivalent values prior to generating the sunshine grids. The correction methodology is described in a paper by A. Kerr and R. Tabony: 'Comparison of sunshine recorded by Campbell-Stokes and automatic sensors' (Weather, April 2004, vol. 59,90-95), but additional overlapped data was used by M. Perry in 2007 to produce improved monthly adjustment factors: Perry M. 2007. 'Updated Comparison of Sunshine Duration Recorded by Campbell–Stokes and Automatic Sensors' (Published as National Climate Information Centre Memorandum no. 27 (2011). National Meteorological Library: Exeter, UK).
The standard exposure for wind sensors is at 10 m above ground in an open, level area (such as an airfield). Hourly observations of mean wind speed are used to calculate monthly means which are then interpolated to the 5 x 5 km grid, taking into account altitude. The resulting grid therefore reflects general topography (e.g. hills, and coasts) but not the sheltering effect of urban or forested areas. The speeds are in knots, the internationally agreed meteorological unit (1 knot = 0.514 m/sec = 1.15 m.p.h.).
The 30-year period 1961 to 1990 has been designated as the international standard reference period for climate averages by the World Meteorological Organization. Averages for the periods 1971 to 2000 and 1981 to 2010 have also been produced for the UK, but the earlier period has been chosen for the UKCIP grids as it represents a better baseline for placing recent climate change into context.
The advantages of gridded data include:
- regional values can be produced for any arbitrary area (county, unitary authority, river catchment, etc.) with greater accuracy and consistency
- they facilitate the creation of contoured or colour-shaded maps, e.g. of 30-year averages
- they can be combined with other spatial data, such as transport routes, to study links between them
- less reliance on a weather station network that is irregularly spaced and changing with time
- reliable climate estimates for points some distance from weather stations, e.g. in upland areas
- a consistent series of climate data, enabling comparisons to be made in time and space
- a complete series of climate data, without missing values (a problem to which weather stations can be prone)