(To view an enlarged image, click on the small icons below)
Site name: Army City
Then...
And now:
Nearest town: Ogden; State/country:
Kansas; Type of site: World War I commercial complex associated with
Camp Funston (now Fort Riley); Date of site: 1917-1921; Sponsor:
Strategic Environmental Research and Development Program (SERDP), Department
of Defense; Collaborators: Center for Advanced Spatial Technologies,
NASA Marshall Space Flight Center, US Army Corps of Engineers Construction
Engineering Research Laboratory, Geoscan Research USA, Fort Riley.
Project Background:
This research project aims to demonstrate the effectiveness of a suite of non-invasive
remote sensing methods combined with new analytical tools to allow recovery
of detailed information about subsurface archaeological content. Such an approach
can significantly reduce costs and the large investments of time and labor associated
with traditional archaeological surveys and excavations. The project mixes high-resolution
ground-based geophysical (magnetometry, resistivity, conductivity, magnetic
susceptibility, ground penetrating radar), aerial (thermal, panchromatic, multi-spectral),
and space-based (panchromatic, multi-spectral) sensor data. GIS-based data fusion
methods that associate contextual with multiple-layer statistical patterns enhance
results by enabling superior visualization of the complex, high-dimensional
data sets. In order to assess capabilities under variable conditions a range
of prehistoric and historic archaeological sites in a variety of environments
from throughout the country will ultimately be examined. Remotely sensed data
will be acquired at DoD installations at Fort Bliss (Texas), Fort Benning (Georgia,
Alabama), and Fort Riley (Kansas), and at the DoE Savannah River Site (Georgia).
Existing site data from two additional locations, Whistling Elk (South Dakota),
and Mt. Comfort (Arkansas), will complement information obtained at these sites.
This presentation gives initial and very preliminary results obtained
during the summer of 2002 at Army City, Kansas, a World War I commercial complex
established to provide entertainment and other services to troops at Camp Funston
(now part of Fort Riley). The town, created in 1917 and abandoned in 1921 after
it burned, presently rests under a hay field with few surface indications of
its presence (bottom photo, above). Part of the town, at its height, is illustrated
in the top photograph (see "Images
of Kansas Cities and Towns" and "A
Kansas Portrait" for additional views of this site, and "Life
at Camp Funston" for interesting views of that installation). A News
Release from the Fort Riley Public Affairs Office gives further details
of the summer fieldwork.
In
1995-6 most of the entire town was geophysically surveyed by Geoscan Research
using electrical resistance methods. The figure at left shows their results,
with good indications of many features of the town as well as a paleo-channel
to the northeast now invisible on the surface (see the NADAG
project page and the Archaeo-Physics,
LLC website for additional details about this work). The locus of our current
1.6 ha study area is also indicated. It is also highlighted in red in the photograph
at the top of the page.
Data from five geophysical methods (listed above), most of which respond to
different dimensions of subsurface physical properties, are presented below,
with many similarities and differences apparent. An ultimate goal of the project
will be to employ the data together, simultaneously, utilizing a variety of
new data fusion approaches that ultimately will indicate better detail and new
insights about this important site. The image above the historical photograph
is an illustration of a simple image fusion (see also below).
Thermal, high resolution panchromatic and multispectral data from aerial and
space-based platforms (the QuickBird satellite) will soon be fused with the
geophysical data to yield an even clearer understanding of the subsurface content
and organization of Army City. Vegetation patterns indicated on the surface
during fieldwork (see Resistivity, below) suggest great promise from
aerial methods.
A detailed mapping of all surface-visible features and indications of subsurface
changes was undertaken throughout the study area. This graphic indicates that
a wealth of evidence about the subsurface was detectable at the surface with
the human eye. These data will ultimately be compared with the various forms
of remote sensing evidence. GIS technology, together with new "data fusion:
software, will ultimately allow thorough analysis, combination, and understanding
of these data.
A comprehensive website devoted to this project will later be established.
(top)
Geophysical Results:
Type of survey: Resistivity; Instrument: Geoscan Research RM-15
with MPX-15 multiplexer; Prospection depth: 0.5 m; Sampling interval:
0.5 x 1.0 m; Area surveyed: 16,000 m2 (1.6 ha); Date of
survey: 7/02.
Comments: The resistance data (left icon) clearly reveal individual walls
and rooms of a large structure in the northwest, historically called the "Hippodrome."
Some of these walls were visible as vegetation markings seen on the surface
(middle icon). The principal northwest-southeast trending street, known as "Washington
Street," is readily seen as are a number of other structures, including
the "Orpheum Theater" in the north-central portion of the study block.
Of large interest is the faint outline of a rectangular structure seen in the
south-central portion of the study block that is indicated as a series of unconnected
dots represent building footers. These features are revealed despite the somewhat
coarse sampling interval employed (relative to the size of these features),
indicating the robust contrast they offered compared to the background data.
Surprisingly, these anomalies are also clearly indicated by vegetation marks
in the surface (right icon). They can also be discerned in the GPR data, below,
but not in the other imagery. The image behind the right icon reveals what is
hidden within saturated areas in the raw data, after application of a high-pass
filter. The speckled nature of these data is largely due to the extremely dry
conditions that prevailed at the time of the survey, which caused large variations
in probe contact resistance as the RM-15 was moved about the site. Future work
will focus on reducing this defect. Many of the indicated structures are illustrated
in the photograph at the top of the page, including the Hippodrome, the Orpheum,
and the southern rectangular structure (labeled 37-38).
(top)
Type of survey: Electromagnetic Conductivity; Instrument: Geonics,
Ltd. EM-38, quadrature phase; Prospection depth: 1.5 m, with peak sensitivity
at 0.4 m; Sampling interval: 0.5 x 1.0 m; Area surveyed:
16,000 m2 (1.6 ha); Date of survey: 7/02.
Comments: While soil resistivity and conductivity are the theoretical
inverses of each other, the results of these surveys include important differences
stemming from the very different instrumentation employed to obtain each data
set and variations in the soil volumes evaluated. The raw conductivity data
(left icon) largely illustrate, at present, highly conductive buried pipes that
lie beneath many of the buildings. Some building outlines are indicated, however,
including a different visualization of the south-central rectangular structure.
Initial processing of these data employed despiking to remove some of the low-level
metallic noise, a high-pass filter for trend removal, and a low-pass filter
for further noise supression (right icon). A number of subtle details not apparent
in the raw data are enhanced and it becomes possible to visualize features within
saturated areas around the many pipes.
(top)
Type of survey: Magnetic gradiometry; Instrument: Geoscan Research
FM-36 fluxgate gradiometer; Prospection depth: up to 1.5 m; Sampling
interval: 0.25 x 1.0 m; Area surveyed: 16,000 m2
(1.6 ha); Date of survey: 7/02.
Comments: The magnetic gradiometry data (left icon) markedly illustrate
the distributions of iron and steel artifacts, including iron in concrete foundations,
individual iron or steel artifacts like nails, nuts and bolts, etc., and buried
iron pipes. Some of the iron pipes illustrate a pattern of alternating positive
and negative values, indicating strings of magnetic dipoles that probably correspond
with pipe joints. A comparison against the conductivity image (above) suggests
that some pipes may not be of ferrous metal. Magnetometry is particularly sensitive
to fired materials; some of the response near the buildings may therefore point
to bricks or possibly burned earth from the 1921 fire. A neighborhood variance
balancing filter was employed to reveal greater detail in areas of low magnetic
activity and reduce high dynamic ranges elsewhere; the result was also subjected
to a low-pass filter (middle icon). The photo (right icon) shows one of several
iron (sewer?) pipe holes that intrude at the surface.
(top)
Type of survey: Magnetic Susceptibility; Instrument: Geonics,
Ltd. EM-38, in-phase component; Prospection depth: about 0.5 m; Sampling
interval: 0.5 x 1.0 m; Area surveyed: 16,000 m2 (1.6 ha);
Date of survey: 7/02.
Comments: These data provide many new insights about the buried structures
at Army City, possibly zones of brick debris, intensive burning, refuse dispersal,
topsoil mounding or extraction (left icon). The raw magnetic susceptibility
map does not illustrate the extreme anomalies caused by many of the numerous
metal pipes that are seen in the other data sets. Initial processing of these
data employed despiking, a high-pass filter to eliminate large zones of high
measurement, and a low-pass filter for further noise supression (right icon).
A number of subtle details not apparent in the raw data are enhanced and it
becomes possible to visualize the many features that exist within the saturated
areas of high magnetic susceptibility.
(top)
Type of survey: Ground penetrating radar; Instrument: Geophysical
Survey Systems, Inc., SIR-2000 with 400 MHz transducer & survey wheel; Prospection
depth: about 1.5 m; Sampling interval: 0.5 x 0.05 m; Area surveyed:
16,000 m2 (1.6 ha); Date of survey: 7/02.
Comments: GPR was a big question at this site for it lies in conductive
clay/silt soils in the Kansas River bottomlands that have frequently been subjected
to flooding. Given here (left icon) is a preliminary look at the GPR data based
on a composite of time slices resulting from 8 individual sub-blocks that were
necessary to survey the entire area. All the slices are taken from the 5-15
nS range (TWTT; to perhaps .75 m depth). The slices are derived from the raw,
unprocessed profiles, although background removal was performed. Seams are readily
apparent between the 8 blocks of data as are stripes and other defects resulting
from extreme temperatures and other circumstances (these defects will ultimately
be corrected). Interestingly, a light rainstorm may actually have clarified
results along the central area by exaggerating contrasts between building materials
(e.g., concrete) and the surrounding soils. Whatever the case, this initial
look suggests great promise in these data. In particular, many details in the
building and non-building areas are indicated that do not appear to occur in
the initial data sets from the other instruments (above). Further data processing
using a variety of common GPR transforms, gain balancing, and more detailed
time-slicing will surely reveal many important insights, and at multiple depths.
Preliminary Data Fusions:(top)
A variety of multi-layer mathematical, statistical, contextual, and other imaging
methods will ultimately be employed to fuse the information content of these
and the yet-to-be-gathered aerial and space-based data. The following presents
some initial views of fusion using relatively simple methods applied to the
unprocessed data. Stay tuned for better and much more sophisticated results!
Fusion through computer graphics
A simple visual illustration is provided here utilizing Adobe
Photoshop software. Beginning with the resistance data as a base (left
icon, left image), with opaqueness set at about 40%, the magnetic susceptibility
image was superimposed, with opaqueness at about the same level (left icon,
2nd image). Next, the magnetic gradiometry was superimposed (left icon, 3rd
image), and finally the soil conductivity data were added (left icon, right
image), both with similar levels of opaqueness. Each input image was color coded
as in the above maps, resulting in the 4-image composite (middle icon). The
middle fused image is also portrayed above the historical photograph at the
top of the page. Finally, an animation of this process is given by the third
icon.
Fusion through principal components analysis & color compositing
The five principal geophysical data sets were subjected to a principal components
analysis that revealed common underlying dimensions of relationship. Three of
the components gave clear interpretation: about 96% of the variance of one was
derived solely from magnetometry, about 77% of the variance in another was equally
derived from GPR-resistivity, and about 72% of the variance in a third was equally
split between the conductivity-magnetic susceptibility data. These three components
were respectively assigned the primary colors of blue, green, and red. The image
at the left portrays the composite of the three color primaries.
(top)
Back
to "LIBRARY OF GEOPHYSICAL IMAGERY" home page
(Last update: 10/12/02)