57-4 The Spatial Structure of Microscale Reactivity of Arsenic in Soil Material.

See more from this Division: ASA Section: Biometry and Statistical Computing
See more from this Session: Symposium--Advanced Statistical Approaches to Reach Strong Inferences From Agronomic and Environmental Studies

Monday, November 4, 2013: 3:35 PM
Marriott Tampa Waterside, Grand Ballroom E

Montserrat Fuentes1, Dean L. Hesterberg2, Matthew Polizzotto3 and Joseph S. Guinness1, (1)Statistics, NC State University, Raleigh, NC
(2)Soil Science, North Carolina State University, Raleigh, NC
(3)1272 University of Oregon, University of Oregon-Eugene, Eugene, OR
Abstract:
The existence of toxic trace elements in drinking water is a significant public health concern, especially in Southern Asia, where 100 million people consume drinking water containing levels of arsenic that exceed what is considered dangerous by the World Health Organization. Movement of arsenic in the environment is regulated in part by soils, which are complex mixtures of minerals and organic material, so it is important to understand the chemical reactions involving arsenic and other elements commonly found in soil particles. We describe an experiment in which a single grain of mineral-coated quartz is treated with arsenic solution, and the surface elemental composition is measured on a fine grid of spatial locations. We study the spatial patterns of the accumulation of arsenic on the sample and the extent to which those patterns can be explained by the spatial distribution of the other elements present in the sample. In this study, we consider the effects of iron and chromium on the accumulation of arsenic. Our analyses suggest that arsenic accumulates in regions where a large amount of iron is present, but this effect is mediated by the presence of chromium. We develop spatial statistical methodology for testing whether two components, arsenic and chromium in this application, are conditionally uncorrelated given a third, iron,  versus the alternative that their partial correlation given is negative. The test relies on the specification of  nonseparable multivariate spatial models to account for spatial correlation. The spatial models have parameters for controlling the marginal and partial correlation among the multiple elements. Our results show how spatially heterogeneous chemical microenvironments affect redox transformations and accumulation of arsenic in soil material, transcending the paradigm that the reactive soil unit is a pure molecular species, and instead presenting soil species as being integrally connected to their host micro or nanoenvironments.

See more from this Division: ASA Section: Biometry and Statistical Computing
See more from this Session: Symposium--Advanced Statistical Approaches to Reach Strong Inferences From Agronomic and Environmental Studies

<< Previous Abstract | Next Abstract