Modeling arsenic in the wells of Nepal

Groundwater is used in Terai region of Nepal as a major source of drinking water since surface water is vulnerable to microbial contamination. But the presence of arsenic, a toxic element in groundwater poses a threat to human health and environment. Arsenic in groundwater is a serious health hazard not only in Nepal but in almost all the countries in South East Asia, including the arsenic hot spots, Bangladesh and West Bengal, India. Even though mitigation plans for arsenic is extended almost everywhere, the risk of arsenic seems to be an omnipresent problem.; We suggested that the mitigation strategy for arsenic requires the quantification of this risk through a series of statistical modeling approaches that describe the possible relationship of arsenic with measurable physical quantities and with the water chemistry in subsurface water. We developed prediction models to estimate the arsenic concentration at unmeasured locations. These models discard insignificant predictors and include only the significant ones in terms of variation of groundwater arsenic. We also developed risk models to asses the risk posed to human health by the presence of arsenic in groundwater. The models indicate a dependence of high arsenic on positive linear function of depth that represents the downward transport of oxygen and negative quadratic function of depth proximates to deeper course sediments.; In the sequential development of models we found that the linear models fail to describe the arsenic process in groundwater, however, the quadratic models perform reasonably well in this aspect. The higher order polynomial models perform quite well in describing the arsenic process. Among the models considered, generalized quasi-Poisson is comparatively better and quantile model is good in modeling extreme cases. Further more, the models with the entire dataset are not that much efficient, but the models that use data sub settings are quite satisfactory. Among these models, the micro model used in the areas of arsenic hot spots based on the sub setting defined by the spatial structure of the data is most successful in prediction as well as risk assessment. These micro models are able to explain much of the variability of arsenic in groundwater tube wells. These models also allow prediction of arsenic at unmeasured locations, quantitative assessment of health risk, optimum remediation strategies, and cost benefit analysis for safe drinking water.