Weights of Evidence Modeling in Archeology, Rocky Mountain

Download Report

Transcript Weights of Evidence Modeling in Archeology, Rocky Mountain

Weights of Evidence
Modeling in Archeology,
Rocky Mountain
National Park
Dr. David M. Diggs and Dr. Robert H. Brunswig
School of Social Science
College of Humanities and Social Sciences
University of Northern Colorado
Greeley, Colorado
Project Goals
•
Identify Native American “Sacred Sites and
Features” in Rocky Mountain National Park, CO
(RMNP).
• Model the known occurrence of sacred
sites/features in RMNP using GIS AND other
sources of information.
• Use ArcSDM (Spatial Data Modeler) and the
Weights-of-Evidence technique to predict the
occurrence of sacred sites and features in RMNP.
Archeological Sites in RMNP
• Over 1,000 archeological sites
in RMNP.
• Approx. 29% of RMNP has
been surveyed.
• Over 400 prehistoric and early
historic sites.
• 31 sites and feature cluster sites
(183 features) were identified as
having well-established or
highly probable Native
American religious or ritual
components
System-wide Archeological Inventory
Program (SAIP) 1998-2002
source: http://www.nps.gov/romo/resources/history/prehistoric.html
In 2000, a long-term research program was initiated in the Park
and its surrounding mountain region to identify and study
archeological features thought associated with Native
American ceremonial practices and attempt to reconstruct
models of their hypothesized relationships with their ancient
cultural and environmental-topographic landscapes.
Sacred Sites
Specific locations with archeological evidence of
Native American religious practices are defined here
as constituting sacred sites.
Sacred circles for meditation,
prayer, ritual, and spirit contact
Prayer Circle on Glacial Erratic Boulder
Feature Lichen-Dated to AD 900
Vision quest features
Shrines/Altars:
Offering places for local
spirits
Model
variables
explored
Statistical significance
of variables tested with
Kolmogorov-Smirnov
(K-S) or chi-square
tests.
Not used in final model
Cost Distance from PreHistoric/Historic Trails
Sacred Landmarks Visibility
Weights-of-Evidence Method
• ArcSDM (Spatial Data Modeler) extension provides techniques to combine two
or more evidential themes for the generation of a response theme (Sawatzky etal.
2004). Includes weights of evidence, logistic regression, fuzzy logic and neural
network analysis capabilities.
• Weights of evidence was initially developed for mapping mineral potential
(Raines, Bonham-Carter, Kemp 2000). Also been applied to archeological site
prediction in California (see Hansen 2000 and Hansen et al. 2002).
• WOE evaluates the spatial distribution of known “events” relative to multiand/or binary map patterns.
• Calculates weights of spatial association (W+ and W-) between an Evidence
Layer and a Training Point Set.
• The weights (W+ and W-) can be used to reclassify and generalize an evidence
layer.
• Evidential themes can then be combined to create a response theme. This
includes a posterior probability—the chances that we will find similar
features/sites in each cell of the grid theme (in our case the chances that we
would find sacred sites).
Final Weights Table
W+ (positive weight):
0 to .5
Mildly Predictive
.5 to 1.0 Moderately Predictive
1.0 to 2.0 Strongly Predictive
> 2.0
Extremely Predictive
Agterber and Cheng Test showed variables were independent of one another. This is a
required condition of the WOE method.
Posterior Probability of finding
sacred Features (with trails)
within .5 sq. km.
Posterior Probability of finding
sacred Features (without trails)
within .5 sq. km.
So what does this mean?
Relative certainty of the Posterior Probability.
Recommended survey sites in the
Never Summer Range, RMNP
Recommended future
survey sites in RMNP
Conclusions*
• WOE may overestimate probability of an event.
• Posterior Probability #s: think in terms of relative favorability,
not exact probabilistic maps.
• Weights-of-Evidence (WOE) can be a useful tool for
exploration of data, and ideas. The generalization of maps into
binary maps can involve both objective and subjective
methods.
• WOE was an important contributing factor in the selection of
future survey sites. HOWEVER, other sources of knowledge
were equally valuable.
* ArcSDM / Spatial Data Modeler 9.2 version available as ArcScript at:
http://arcscripts.esri.com
9.3 version at: http://www.ige.unicamp.br/sdm/default_e.htm
Acknowledgments. The authors would like to acknowledge funding support for this project by Rocky
Mountain National Park (the National Park Service) and the University of Northern Colorado's Sponsored
Programs Office. We would also like to particularly acknowledge the very important contributions to this
project by Dr. Bill Butler, Rocky Mountain National Park archeologist, and our consultants from the
Northern Ute, Southern Ute, Ute Mountain, and Northern Arapaho tribes. We would also like to recognize
our energetic field crew, Cody Dalpra, Louise Elinoff, Tom Lux and Dr. Steve Cassells, our lichenometry
expert.
Author Information
Dr. David M. Diggs
Campus Box 115
Geography Program
College of Humanities and
Social Sciences
Univ. of Northern Colorado
Greeley, CO 80639
ph.970-351-1113
Fax 970-351-2890
[email protected]
Dr. Robert Brunswig
Director, School of Social
Science
College of Humanities and
Social Sciences
Univ. of Northern Colorado
Greeley, CO 80639
ph.970-351-2138
Fax 970-351-1527
[email protected]
References
Brunswig, Robert H. 2003. Archeological, Ethnographic, and Historic Investigations of the 5LR7095 Rock Feature Complex Site, Rocky Mountain
National Park, North Central Colorado: Report to Rocky Mountain National Park, National Park Service, Estes Park, Colorado. Department of
Anthropology, University of Northern Colorado, Greeley.
_____. 2005. Prehistoric, Protohistoric, and Early Historic Native American Archeology of Rocky Mountain National Park: Volume 1-Final Report of
Systemwide Archaeological Inventory Program Investigations by the University of Northern Colorado (1998-2002). Prepared for the National Park
Service; Rocky Mountain National Park. Anthropology Program, University of Northern Colorado, Greeley.
Brunswig, Robert, David M. Diggs, and Christine Montgomery. 2009. Native American Lives and Sacred Landscapes in Rocky Mountain National
Park. Prepared Under: Cooperative Agreement: 1443-CA-1248-007. Anthropology Program, University of Northern Colorado, Greeley.
Brunswig, Robert, Sally McBeth, and Louise Elinoff. 2009. Re-Enfranchising Native Peoples in the Southern Rocky Mountains: Integrated
Contributions of Archaeological and Ethnographic Studies on Federal Lands. In Post-Colonial Perspectives in Archaeology, edited by Peter Bikoulis, D.
Lacroix, and M. Pueramaki-Brown, pages 55-69. Chacmool Archaeological Association, Calgary, Canada.
Bonham-Cart, Graeme F. 1994. Geographic Information Systems for Geoscientists. Tarrytown, New York, Pergamon Press, Elsevier Sciences, Inc.
Butler, William. 2007. Personal communication.
Diggs, David and R. H. Brunswig. 2006. Modeling Native American Sacred Sites in Rocky Mountain National Park. In Proceedings of the TwentySixth Annual ESRI International User Conference (web publication at http://gis.esri.com/library/userconf/proc06/papers/abstracts/a2243.html) ESRI,
San Diego.
Duncan, Clifford H. and J. A. Goss. 2000. Consultancy Report on Traditional Ute Sites in Rocky Mountain National Park (August 22, 23, and 24
2000). Report on file, Rocky Mountain National Park, Estes Park, Colorado.
Hansen, David T. 2000. Describing GIS Applications: Spatial Statistics and Weights of Evidence Extension to ArcView in the Analysis of the
Distribution of Archaeology Sites in the Landscape. Proceedings of the 20th Annual ESRI International Users Conference. San Diego.
Hansen, David T., G. J. West, B. Simpson, and P.Welch. 2002. Modeling Spatial Uncertainty in the Analysis of Archeological Site Distribution.
Proceedings of the 22nd Annual ESRI International Users Conference. San Diego.
Lux, Thomas. 2005. Ancient Trails at Rocky Mountain National Park. M.A. Thesis, Denver, Colorado, Department of Anthropology, University of
Denver.
Mihalasky, Mark J. 2001. Mineral Potential Modeling of Gold and Silver Mineralization in the Nevada Great Basin: A GIS-Based Analysis Using
Weights of Evidence. Open–File Report: 01-291. U.S. Dept. of Interior, U.S Geological Survey, Washington, D.C.
Raines, Gary L. 2009a. Spatial Data Modeler. ArcScript available at:
http://arcscripts.esri.com/details.asp?dbid=15341 Last accessed 4/6/2009.
Raines, Gary L. 2009b. Spatial Data Modeler for ArcGIS 9.3 ArcScript available at:
http://www.ige.unicamp.br/sdm/default_e.htm Last accessed 4/10/2009
Raines, Gary L., G.F. Bonham-Carter, and L. Kemp. 2000. Predictive Probabilistic Modeling Using ArcView GIS. ArcUser. April-June: 45-48.
Rohe, Christopher M. 2003a. Final Report on the Development of Predictive Models for Rocky Mountain National Park: Report to Rocky Mountain
National Park. Department of Anthropology, University of Northern Colorado, Greeley.
_____. 2003b. Reading the Landscape: A Location Model for Prehistoric Sites in Rocky Mountain National Park. M.A. Thesis. University of
Arkansas, Department of Anthropology, Fayetteville.
_____. 2004. Use of the Simple Weighting Method in Modeling Prehistoric Site Locations in Rocky Mountain National Park. In Ancient and Historic
Lifeways of North America’s Rocky Mountains: Proceedings of the 2003 Rocky Mountain Anthropological Conference, edited by Robert H. Brunswig
and William B. Butler, pp. 425-452. Department of Anthropology, University of Northern Colorado, Greeley.
Sawatzky, D.L., G.L. Raines, G.F. Bonham-Carter, and C.G. Looney, C.G. 2004 ARCSDM3.1: ArcMAP extension for spatial data modeling using
weights of evidence, logistic regression, fuzzy logic and neural network analysis. http://www.ige.unicamp.br/sdm/ArcSDM31/default_e.htm