 |
|
The M-Scale model: A multi-scale model for decision support of on-site remediation
Meng-Ying Li
Paperback. ProQuest, UMI Dissertation Publishing 2011-09-02.
ISBN 9781243488558
|
|
|
|
Buy from Amazon.com
|
Publisher description
Remedial decisions for sediment management involve knowledge of the biogeochemical processes affecting contaminant fate and transport in the sediment, as well as the spatial distribution of the contaminant. Spatial statistical models provide estimates of the spatial distribution, and the results depend on validity of the assumptions inherent to the selected statistical tools, and the appropriateness of these tools with respect to the objectives of the estimation. Most decision tools for site assessment depend on spatial estimation/simulation that either misinterpret the extent of exceedance, or assigns a single decision map corresponding to a given uncertainty criterion. The specific objectives of this work are (i) to provide a spatial statistical approach, the M-Scale model, for characterization of the spatial structure and spatial distribution of an attribute, such as contaminant concentration or microbiological parameters; (ii) to investigate the applicability of the developed model to field data relevant to contaminated sediments using various performance diagnostics; and (iii) to explore the sensitivity of the M-Scale model and other methods to the nugget effect (artificially induced error and micro-scale variability) using laboratory and field data from the Anacostia River (NJ). Results using artificial data indicate the developed model generates estimates that (i) reproduce spatial variability evident in the sample, with reasonable precision for classifying exceedance/non-exceedance of a design threshold, and (ii) reproduce the overall attribute value distribution. Cross-validation results using datasets from the Passaic River yield similar performance metrics for the M-Scale model relative to CK in the reproduction of the overall value distribution, and relative to OK in the precision of classification. Estimation results using samples at both the site-scale and the micro-scale from the Anacostia River further indicate the possibility of reducing the uncertainty associated with estimates by characterizing the actual micro-scale variability. Cross-validation results using the same datasets indicate that each data point in a small-size sample set is essential in the estimation process. The reproduction of spatial variability demonstrated in this dissertation indicates improvement of spatial estimation by characterizing multi-scale covariances of means. The model has broad applicability for situations where multi scale characterization issues drive spatial management decisions
Similar books
Rate the book
Write a review and share your opinion with others. Try to focus on the content of the book. Read our instructions for further information.
The M-Scale model: A multi-scale model for decision support of on-site remediation
Book reviews » The M-Scale model: A multi-scale model for decision support of on-site remediation
|
|
 |
|
|
|
|
