[CPEO-MEF] UXO Spatial Distribution

[Lenny Siegel <lsiegel@cpeo.org>]

Assessing Sites Contaminated with Unexploded Ordnance:
Statistical Modeling of Ordnance Spatial Distribution 

An important new article on the spatial distribution of unexploded
ordnance (UXO) on ranges is available on line at
http://pubs.acs.org/cgi-bin/abstract.cgi/esthag/asap/abs/es051168t.html
Unfortunately, there's a fee unless you're a subscriber to Environmental
Science & Technology.

Assessing Sites Contaminated with Unexploded Ordnance: Statistical
Modeling of Ordnance Spatial Distribution 
Jacqueline A. Macdonald and Mitchell J. Small 
Department of Civil and Environmental Engineering and Department of
Engineering and Public Policy, Carnegie Mellon University 

Abstract:

More than 40 000 km2 of former military land in the United States are
contaminated with unexploded ordnance (UXO). Cleanup costs are estimated
to total as much as $140 billion. The amount of contaminated acreage and
total costs are likely to increase as the U.S. Department of Defense
(DOD) follows through on recently announced plans to close an additional
22 domestic military bases. The U.S. Environmental Protection Agency
(EPA) and DOD disagree on how these sites should be characterized to
assess their risks and plan for cleanup. As a result, much potentially
valuable land remains idle while remediation decisions are pending. One
of the sources of disagreement is how the locations of UXO should be
characterized, given that the exact spatial distribution of UXO is
unknown in advance of cleanup. In this paper, we propose and test a new
model to represent the spatial distribution of UXO. Unlike existing DOD
models, the new model accounts for the tendency of UXO to cluster,
presumably around targets at which soldiers aimed during training. We
fit the cluster model to geographic data on UXO locations at two former
military installations and show that it describes key characteristics of
the data more accurately than the existing DOD model. We discuss how the
choice of a UXO spatial distribution model could affect important
decisions about cleaning up and reusing UXO-affected property. 

***

The authors compared a Poisson Cluster Model to the Complete Spatial
Randomness inherent in the Army Corps of Engineers' SiteStats/GridStats
and UXO Calculator tools, using field data from former Fort Ord,
California and the former Tobyhanna Artillery Range, Pennsylvania. They
concluded that the cluster model provided a better fit.

This has significant implications for UXO response. It may be necessary
to sample larger areas on ranges - than called for with the Corps'
current statistical methods - to come up with representative data.
Furthermore, the authors suggest, UXO investigations should begin by
"attempting to locate target areas."

The authors plan further research, leading toward the development of a
new quantitative risk model.

LS

-- 


Lenny Siegel
Director, Center for Public Environmental Oversight
c/o PSC, 278-A Hope St., Mountain View, CA 94041
Voice: 650/961-8918 or 650/969-1545
Fax: 650/961-8918
<lsiegel@cpeo.org>
http://www.cpeo.org
_______________________________________________
Military mailing list
Military@list.cpeo.org
http://www.cpeo.org/mailman/listinfo/military