Cleaned UK rainfall chemistry data (1986-2011)

Stored data in the UK-AIR database, operated on behalf of the Department of Environment, Food and Rural Affairs (Defra) at http://uk-air.defra.gov.uk/ were downloaded in 2012 for 20 of the PrecipNet sites with the longest data record, starting in 1986. Each sample entry was examined for internal chemical consistency and evidence of contamination, using the following rules: 1 ion balance: The ion balance was calculated as (Sum of(cations) - Sum of(anions)) / (Sum of(cations) + Sum of(anions)). Some imbalance may occur because not all ions are analysed, e.g. bicarbonate or weak organic acids, so that a small positive balance is usually observed. Data were regarded as acceptable where the ion balance lay in the range -10% to +20%, except where the overall ion concentration (Sum of(cations + anions)) was less than 200 microequiv/l, where the uncertainties in individual ion analyses at low concentrations could lead to a greater uncertainty in calculating the ion balance; in such cases the range was extended from -10% to +30%. 2 evidence of contamination: Samples with measured phosphate concentrations above 10 microequiv/l were eliminated. All samples with ammonium concentrations above 100 microequiv/l were examined, and those with potassium concentrations above 8 microequiv/l were eliminated. Finally, all samples with calcium concentrations greater than 50 microequiv/l which had higher calcium concentrations than sodium concentrations (in microequiv/l) were excluded as being contaminated with wind-blown dust. This last condition is rather conservative, i.e. some dust-contaminated samples may remain in the dataset; these usually were samples with very low sample volumes, so contributed little to the annual deposition. Following identification and removal of contaminated sample data, the missing data were estimated, where possible, using statistical interpolation across both time and space with the GENSTAT procedure MULTMISSING. The final datasets contain the accepted and estimated data values, flagged as appropriate.