Chemometric analyses of atmospheric precipitation in the WMO background air pollution monitoring network

Abstract

A detailed chemometric analysis has been made of the atmospheric precipitation data from the World Meteorological Organization Background Air Pollution Monitoring Network (BAPMoN) for the period 1980-1985 in an attempt to characterize precipitation chemistry in remote sites of the World. Data evaluation by appropriate quality control procedures reveals that the quality of the data is inconsistent among different stations. Unreasonable pH values have been recorded for atmospheric precipitation in Kecskemet and French stations, which indicate presence of errors in pH measurements. In addition, decimal point errors have been found in the measured electrical conductivity, Ca²+ and C1ˉ data of Ejde station. At Cameron Highlands, the poor ion balance may be associated with poor sensitivity of nephelometric turbidimetry used in the determination of SO4²ˉ concentration. Anion deficit is common in precipitation samples of majority of the remote BAPMoN sites and shows spatial variability. The large anion deficits for the Italian stations suggests that certain important anions are not measured. It is speculated that the distribution of the ion and conductivity balances will be non-normal owing to the missing anions. As such both ion and conductivity balances may be inappropriate quality control parameters for the validation of precipitation chemistry data from such sites, especially those that are remote from urban area. Results from Kolmogorov-Smirnov test indicate that 2-parameter Gamma distribution appears to be better in describing majority of the ion concentration data. Non-parametric and 'robust' methods are employed in analysis of the data. Both multiple regression and principal component analyses based on rank-transformed data have been demonstrated to be more 'robust' to the non-normal data and are better statistical tools as compared to analysis based on the screened data. A pronounced Ca²+ enrichment can be found in most of the Italian stations including the southern sites, Trapani Birgi and Santa Maria di Leuca, which may result from long-range transport of calcium containing dust originating from arid regions in the north of Africa to these Mediterranean area. It is found that free acidity cannot be fully neutralized by NH4+ alone in most of BAPMoN sites. More complete neutralization of the free acidity may be achieved by reaction with additional Ca²+ and Mg²+. Results of both stoichiometric analysis and rank correlation analysis suggest that NH4NO3 is predominant chemical species among ammonium compounds and nitrate compounds in the atmosphere. Other fully or partially neutralized compounds that may also exist include (NH4)2SO4, NH4HSO4, CaSO4, Ca(NO3)2, MgSO4 and Mg(NO3)2. Multiple regression analysis based on rank-transformed data indicates that the strong inorganic acids (H2SO4, HNO3 and HC1) cannot significantly account for the precipitation acidity in most of the BAPMoN sites. Instead, organic acids are believed to be major chemical species which may control the free acidity of the precipitation in isolated sites. Several source-types have been identified through principal component analysis of the precipitation chemistry data. 'Sea' components and 'neutralization' components are dominant components in the isolated sites, and they account for a significant proportion of total variances of the data. On the other hand, the 'acid' components, dominated by H+ alone, can explain only small amount of the total variances. Owing to the neutralization processes in the atmosphere, only a few 'agricultural/soil' components have been extracted. The 'source' type being identified depends not only on emission sources, but also on meteorological and climatic factors, other processes such as dissolution, and physical/chemical changes. The BAPMoN stations considered in this study exhibit considerable spatial variability in the chemical composition of atmospheric precipitation and can be divided into three clusters through cluster analysis. Cluster 1 is represented by 5 coastal stations and characterized by high level of Na+, Clˉ, Mg²+ and K+, which suggests a strong marine influence on the atmospheric precipitation. Cluster 2 is represented by continental stations which are mainly distributed in northern and western Europe, and is characterized by relatively low levels of the major ions. On the other hand, relatively high levels of NH4+, NO3ˉ, SO4²ˉ and H+ are found for Cluster 3 with its member stations concentrated in central continental Europe.

Date of Award	15 Oct 1999
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Ronald SEQUEIRA (Supervisor)