nerc.ac.uk

Nested sampling and spatial analysis for reconnaissance investigations of soil: an example from agricultural land near mine tailings in Zambia

Lark, R.M.; Hamilton, E.M.; Kaninga, B.; Maseka, K.K.; Mutondo, M.; Sakala, G.M.; Watts, M.J.. 2017 Nested sampling and spatial analysis for reconnaissance investigations of soil: an example from agricultural land near mine tailings in Zambia. European Journal of Soil Science, 68 (5). 605-620. 10.1111/ejss.12449

Before downloading, please read NORA policies.
[img]
Preview
Text (Open Access Paper)
Lark_et_al-2017-European_Journal_of_Soil_Science.pdf - Published Version
Available under License Creative Commons Attribution 4.0.

Download (1MB) | Preview

Abstract/Summary

A reconnaissance survey was undertaken on soil near mine tailings to investigate variation in the content of copper, chromium and uranium. A nested sampling design was used. The data showed significant relations between the content of copper and uranium in the soil and its organic matter content, and a significant spatial trend in uranium content with distance from the tailings. Soil pH was not significantly related to any of the metals. The variance components associated with different scales of the sample design had large confidence intervals, but it was possible to show that the random variation was spatially dependent for all spatial models, whether for variation around a constant mean, or with a mean given by a linear effect of organic matter or distance to the tailings. For copper, we showed that a fractal or multifractal random model, with equal variance components for scales in a logarithmic progression, could be rejected for the model of variation around the fixed mean. The inclusion of organic matter as an explanatory factor meant that the fractal model could no longer be rejected, suggesting that the effect of organic matter results in spatial variation that is not scale invariant. It was shown, taking uranium as a case study, that further spatially nested sampling to estimate scale-dependent variance components, or to test a non-fractal model with adequate power, would require in the order of 200–250 samples in total.

Item Type: Publication - Article
Digital Object Identifier (DOI): 10.1111/ejss.12449
ISSN: 13510754
Date made live: 08 Aug 2017 13:33 +0 (UTC)
URI: http://nora.nerc.ac.uk/id/eprint/517503

Actions (login required)

View Item View Item

Document Downloads

Downloads for past 30 days

Downloads per month over past year

More statistics for this item...