Notes and images of a talk given at the GPR in Archaeology Conference, Nara, Japan, Feb. 16, 2001.  You can click on the images to make them larger.

We have all come a long way from the days we were using these types of raw data profiles.  This is a noisy, un-filtered profile that was produced from an old thermal printer as the data were being acquired.

Today most of us are collecting data using digital systems that can filter the data, while removing background noise, and readily converting time to depth.

I will now present a number of case studies that my students and I have completed in different parts of the world.  This is a photo from Bluff, Utah in the American Southwest.  A shallow depression was thought to contain some type of archaeological features, perhaps a "great kiva", which is a type of semi-subterranean building used for ceremonial purposes.

Our reflection profiles showed a "bowl shaped" depression, filled with homogeneous material, and enclosed by a wall or some other type of enclosure.

To make three-dimensional maps of this feature, the data were sliced into layers about 25 centimeters thick.  In these slices the highest amplitude reflections are colored red and yellow, while the homogeneous material that produces little reflection of radar energy is blue.  The outside wall of the kiva is circular and can be seen in the second slice from 47-68 cm.  It is partially collapsed.

This was easily confirmed by excavations, and the outside wall was confirmed at the depth mapped in the slices.

But even more exciting were the deeper slices that appeared to show a square feature within the wall of the kiva. 

Nothing like this had been found before in large kivas, and the excavation archaeologists were at first reluctant to dig deeper to see if it was "real" or not.

But after much digging, the interior wall was confirmed.  Its function is not known.  But most interestingly, it is not visible on the two-dimensional radar profiles, and was only noticed by studying the subtle amplitude changes in the deeper slices.

In another area of the American Southwest the typical method of finding deeply buried archaeological features is by digging trenches.

This method is claimed to be "non destructive" because the teeth are taken off of the mechanical shovel!

The maps produced by this kind of mapping look something like this:

Of course the archaeologists have no idea what is between the trenches, or what the extent of the features cut by the trenches is.  So in an attempt to prove that GPR could not only find buried features, but do it non-destructively, a grid of data was collected in an area where excavations were to take place.  The "raw", un-processed profiles were extremely "noisy", and did not appear to contain any usable reflections at all.

These data were then filtered to a narrow frequency and the resulting digital data were enhanced to bring out the remaining subtle reflections.  When this was accomplished, features like this concave upward pit house floor were visible.  This is the same data set as shown above.  Look at the difference!

Then the data were sliced to find the amplitude anomalies at the depth where the floor of the pit houses were expected.  They are shown in red and yellow in the map below.  As you can see, many anomalies appeared.

But when they were compared to each of the individual profiles (shown as the horizontal slashes on the map below), there is sometimes a correlation, but sometimes not.   Many of the amplitude anomalies are probably caused by geological changes in the soil matrix, and are not archaeological at all.

The area was then totally excavated, after first putting in trenches (the vertical stripes).  Almost all of the excavated features were predicted by the GPR maps, but a combination of amplitude analysis (the slice maps) and manual interpretation was necessary.  Here is an aerial photo of the excavations after completion.   The pit house edges are visible as the deeper excavations

In Peru, South America, Moche Period (about A.D. 300) kiln sites have never been excavated. 

In a small gully in the Chicama Valley numerous surface pottery scatters indicated ceramic production might have occurred here.

GPR data were collected using a 500 MHz antenna.

In addition a cesium magnetometer was used to find areas that had been burned (hopefully the kilns).

The reflection amplitude map from  the GPR (top map) was then compared to the magnetic anomalies (lower map).  Where they were coincident, kilns were found by excavation.  The GPR reflection anomalies without a magnetic signature were found to be floor of other buildings (ramadas or sun shades) that had not been burned.  This study shows the importance of combining geophysical techniques.

In Washington state, USA, it was known from historic maps that a shed for making barrels was located in this historic fort built by the Hudson Bay Company in the 1830s.  A grid of 500 MHz data was collected over the site.

Standard amplitude slice maps showed a very confusing picture:

In order to resolve the features that might be vertical (the vertical log foundations of the old wooden structure) a series of 100 slices were produced, and then rendered in three-dimensions.  In this view you can see many of the post molds of this foundation, that are not visible in the standard slices due to clutter.

At a test facility on the University of Illinois campus in the USA, a number of archaeological features were constructed, and then buried, compacted and planted in grass.   A 50 x 50 meter GPR grid was collected over the known features to determine what could be imaged.

The reflection data were very poor quality!  The soils are very rich in clay, and it had rained for a week prior to the survey.

An amplitude slice map of the grid did, however, show a number of features:

The large anomaly in the middle is where two pigs were buried to simulate human burials.  These data were then re-processed, and very distinct anomalies were found where the pigs were known to be buried.

When compared to the known location of the pigs, the correlation is quite good, indicating that even very poor quality data has the ability to produce good images of the subsurface when processed.  The large red anomaly on the left was later discovered to be a rodent burrow that had been filled with water just a few days before the survey by a gardener!  The small green anomaly on the upper left was an 8 inch long nail that was left in the excavations by mistake!

More subtle soil features are sometimes very difficult to map using GPR.   We decided to test whether this garden on the University of Denver campus was still visible 20 years after it was dug up and covered by sod.

GPR data were collected over where the garden was thought to be located.

After much data processing, a subtle reflection was visible on some profiles.   Could this be the remains of the garden?

Amplitude slice-maps showed many features, one of which is the linear anomaly to the south that is a sidewalk, and some others that are electrical conduits.  The circular anomaly in the northwest corner of the third slice is the reflection seen in the above profile and is in about the right place for the garden.

An auger was used to collect two cores, one within the anomaly, and one outside.  The one inside is on the left.

It was found that the white zone in the left core was a layer of imported soil that was used as fill, and the dark soil below is the buried garden.  In this case, very subtle changes in soil character are visible using GPR.

The location of graves is always of interest.  Here is a profile across 8 known graves, marked by headstones.

But often caskets or human remains are not as readily visible as in the above profile, but can only be seen as subtle changes in soils across the burial shaft.   When this is the case, amplitude slice-maps can locate these changes, as seen in red and yellow anomalies in the map below from a Mormon pioneer cemetery in Utah, USA.

The American Southwest has numerous adobe sites, where the clay of the structures has "melted" over the years, and is often encased in more clay.   These can be very challenging environments for GPR.  This is the San Marcos site in New Mexico, where the remains of Spanish Colonial church are being excavated.   The individual adobe bricks can be seen in the foreground:

A stratigraphic cross section through the church shows many layers and vertical features.  GPR data were collected parallel to the stratigraphic profile and over much of the rest of the site.

Many profiles show bowl shaped reflections, which are "adobe melt" that have filled up the partially collapsed walls that remained standing.

Amplitude slice-maps also show these melt layers (the circular anomaly) within the well defined angular walls:

A profile parallel to the stratigraphic section above is shown here.  It is very confusing!

But when the known layers are placed on top of the profile, the features are visible. 

G

click here to return to Home Page