Submitted by Douglas C. Comer, Ph.D., DSC-RPG

Several notable archaeological discoveries have been made recently using radar data collected by satellites and space shuttles. Even the popular press has heralded the use of spaceborne radar in finding the "lost" city of Ubar in Arabia, Silk Route sites in the deserts of China, and large Angkorian water reservoirs in Cambodian jungles.

Such discoveries are made possible by the capacity of aerial radar to detect topographic variation (even surface roughness), to sense water and water saturated soils, and to penetrate cloud cover, leaf canopy, and very dry soil. For all anthropologists working at the interface of environment and culture, not just archaeologists, aerial radar holds great promise.

Aerial radar today employs a variety of wave band lengths and band polarizations that are appropriate to both discovering human occupation sites and classifying environments in ways that shed light on how and why occupations are patterned. Spaceborne radar in earlier years used only one band polarized in one way. Even so, this was enough to ascertain features dubbed "radar rivers," the beds of ancient streams now dry and filled with sand. Radar rivers detected in the Arabian Peninsula led archaeologists in 1991 to Ubar, a rich trading city from 2,800 BC to AD 300.

I have been working with National Aeronautic and Space Administration (NASA) space imaging/synthetic aperture (SIR-C/X/-SAR) radar data collected by the space shuttle Endeavour in 1994 of Petra, in Jordan. A Nabataean trading city that flowered from about 300 BC to AD 300, Petra is best known for its magnificent tombs, carved in the rose-colored sandstone canyon walls that surround it. Nonetheless, the Petra region provides a range of radar targets. Sites are large and small, surface and subterranean, and date from the pre-pottery Neolithic through Biblical, Byzantine, and Crusader times. SIR-C/X-SAR data were generated by polarized radar bands of three lengths that were transmitted and received both horizontally and vertically. In general, longer bands sense larger targets and can penetrate certain materials. Transmitting and receiving bands at the same polarization enhances penetration. Shorter bands detect smaller targets and very small surface irregularities and patterns. By assigning primary colors to specific band polarizations, up to three can be combined in the same image, each contributing its sort of information. I have used the data in the following ways:

1. Detecting landforms attractive to human occupation in the past.

At Petra, springs are strung along a bright ribbon, a geological fault, that runs vertically through radar imagery produced from combinations of long and short radar waves (see Fig. 1). That these springs are within the elevation range at which wild emmer wheat grows no doubt influenced the establishment of one of the earliest village sites in the world, Beidha, near the north end of this bright line. Just as surely, the combined attraction of the springs and sandstone canyon system just to the west led the Nabataeans to build their premier city, Petra, near the center of this image.

Topography as indicated by spaceborne radar tells us where to look for features important to the city of Petra. Slopes and streambeds were used to harvest a meager yearly rainfall through an intricate system of dams and cisterns. Rare expanses of flat ground are clues to the location of no longer extant structures.

2. Detecting landforms that were altered by human occupation in the past.

A curiously kidney shaped area was evident in most of the radar images of the Petra region. This proved to be the flattened top of a very high hill. Adjacent was a large structural complex, apparently Byzantine. As exemplified by this, human alteration of the landscape tends to produce geometric patterns of more regularity than do unaltered landforms.

Similarly, radar detection of sand-filled streambeds in the Taklamakan Desert led Derrold Holcomb to postulate the location of the Silk Route and associated structures in 1992. The regularity of some of the smallest of these now suggests to him that they might be canals.

3. Detecting archaeological sites and features directly.

One goal of my work at Petra was to detect archaeological sites and features directly. To do this, we "stacked" radar imagery over other informative higher resolution and larger scale imagery, such as LANDSAT and SPOT multispectral images, high resolution Russian satellite photographs, and large scale photographs obtained by airplanes and balloons. Very accurate coordinates for points visible in all imagery were obtained with a geographical positioning system (GPS). These points were used in computer programs to greatly reduce distortion in the images. They were then used to stack (or coregister) the images so that a point in any image would correspond to the same point in any other. Once coregistered, different images could be viewed in rapid succession or merged. Previously unrecorded sites and features were found in this way.

Also, coregistration and analysis indicated that image pixels (25 meter square) of a certain coloration, produced by specific mixtures of long and short radar waves, seem to be associated with open, angular subterranean chambers. We intend to statistically test this association, and to establish correlations between specific feature characteristics and pixel coloration.

The ability to detect such small archaeological features directly in radar images will be of benefit not only to research, but to site preservation as well. Pairs of radar images known to contain archaeological features can be used to ascertain changes in elevation on the order of millimeters: clues to erosion, looting, and changes in vegetation that threaten archaeological sites.

Radar data is increasingly varied and informative. Many more radar satellites will be launched in the near future. Hardware and software required for the analysis of aerial remote sensing data, including radar, are ever more affordable. Yet while training in aerial remote sensing for students in biological and geological sciences is commonplace, it is still unusual in anthropology. This is as curious as it is unfortunate. The combined appeal of the technique and the anthropological subject matter could be irresistible to many talented students.

Douglas C. Comer is office manager for the National Park Service, DSC-RPG Applied Archeology Center in Silver Spring, Maryland. Dr. Comer is a former Fulbright scholar in cultural resource management, and the author of Ritual Ground: Bent’s Old Fort, World Formation, and the Annexation of the Southwest (University of California Press, 1996), which examines how humanly contrived landscapes reflect and propagate cultural values and beliefs. He is also an adjunct professor at the University of Maryland, College Park. His work with SIR-C/X-SAR radar at Petra was funded by a grant from the J.M. Kaplan Fund, and was conducted in conjunction with the National Parks Foundation and in collaboration with Dr. Talal Akasheh, Dean of Research and Graduate Studies at The Hashemite University, Jordan. The National Aeronautics and Space Administration Jet Propulsion Laboratory (NASA/JPL) provided SIR-C/X-SAR radar data and preliminary image procession. ERDAS Corporation contributed software and technical support to the research.