科研之友
微信
新浪微博
Facebook
分享链接
摘要
A comprehensive forensic analysis of AS contamination in Bangladesh was undertaken using geographic information system (GIS), univariate, and bivariate statistics. The assessment provided the background needed for fine-tuning emerging/evolving methods of spatial mapping based on nonlinear deterministic dynamics and multivariate factorial analyses. Data were derived from a nationwide AS survey completed in 1999 by the British Geological Survey (BGS) in collaboration with the Department of Public Health Engineering (DPHE) of Bangladesh. It was found that the standard deviation of arsenic concentration is highest along the river flood plains of the major rivers. The distribution of arsenic concentration was found to be symmetric in regions of very high mean concentration, while the regions of low and moderate contamination reveal a highly skewed distribution, suggesting the need for data transformation at these points prior to spatial interpolation. The mean value of Ca and Mg concentrations showed a strong spatial similarity in the southwest region of Bangladesh. This spatial similarity was also reflected in terms of skewness of the data. It was inferred that the mapping of AS concentrations at nonsampled locations on the basis of multivariate factorial or regression models may not require the dual presence of both Ca and Mg parameters; hence, one of them is redundant. Both the skewness and kurtosis of the distribution of Si concentration appeared very similar and highly heterogeneous in space, suggesting that Si data require careful transformation prior to modeling only in the smaller scale but not for country-wide/regional purposes. The age of wells was found to be nonlinearly correlated in the southeast region. The correlation of AS with Fe appeared strongest in the southwest region, indicating that the oxidation-reduction hypothesis may be more appropriate as a release mechanism in the southwest region. In another regard, the sulphate concentration was found to be more correlated to AS in the northwest Pleistocene region of Bangladesh, suggesting that the pyrite oxidation hypothesis may be appropriate. Overall, our forensic analysis using GIS revealed useful findings for data transformation and selection of parameters towards spatial mapping based on multivariate factorial and nonlinear deterministic models in the resource-poor settings of developing countries.
- 出版日期2008
