Format (ASCII) The first and second lines of the file contain information on the grid size, for example: grd_sz xmin ymin xmax ymax n_cols n_rows n_months missing 0.5 0.25 -89.75 359.75 89.75 720 360 12 -9999 This is followed by 12 monthly grids that are n_cols by n_rows in size. Each record contains n_cols (=720) columns longitude values, format 720i5, starting at xmin (= 0.25 deg East) and ending at xmax (=0.25 deg West) The first record starts at January, ymax (=89.75 deg North) and the last record is the row for December, ymin (=-89.75 deg South). Co-ordinates represent the centre of each grid cell. Missing values (Antarctica, oceans and major inland water bodies) are assigned integer values of -9999. ccld6190.dat: grd_sz xmin ymin xmax ymax n_cols n_rows n_months 0.50 0.25 -89.75 359.75 89.75 720 360 12 Units: Days * 10

Problems have come to light through user feedback and routine quality control in the Climatic Research Unit (during work towards the construction of an updated version of the 1961-1990 climatology). These problems range from systematic errors (for example, due to incorrect units being assumed for a whole country) to single-station errors. Errors at single stations have varying importance depending on the density of the surrounding network. Where the network is sparse, the station error will influence a larger area because of its greater spatial influence during the interpolation. Systematic errors arise because of varying reporting methods and units adopted by different national meteorological agencies (and other data contributors). The methods used to standardise data to a common format are discussed by New et al. (1999). Some problems are due to shortcomings in the standardisation methods used. In addition, misunderstandings (or a lack of information) about the units in which data were supplied means that the incorrect (or no) standardisation was used. The majority of problems/errors identified to date are due to confusion of units and shortcomings in the standardisation towards the use of common units. The climate variable with the largest potential for confusion is wind speed. Not only do reporting units vary between metres per second, miles per hour and knots; anemometer heights can vary greatly from the desired 10m (e.g. between 2m and 20m). The following (variable) gridded files are known to have either systematic or significant single station errors which potential users need to be aware of. The errors are present in the mean-monthly 1961-90 terrestrial climatology files and are thus propagated into the gridded 1901-1995 monthly time-series. The affected part of the globe is defined where possible. * Greenland: all variables The interior of Greenland is poorly covered by meteorological observation. This coupled with the presence of the high elevation ice cap makes interpolation of climate normals very difficult due to the potential for unusual lapse rates. It is likely that significant bias may be present with all variables for the interior of thelandmass (e.g. diurnal temperature range and precipitation too high).