Prospect for Locating Ruins Using Spaceborne Imaging Radar

by Pamela Logan
Urumqi, Xinjiang - September 27, 1993


The southern Silk Road in China is a branch of the ancient trade route that connects China to civilizations of the west. This report is concerned with a segment of that route in southern Xinjiang Uighur Autonomous Region. It extends from Hotan in the west to Ruoqiang in the east, skirting the Taklimakan Desert on its southern fringes. In this area many ruins, some dating as far back as the Han Dynasty (ca. 2000 years old), have already been identified. Because of the gradual expansion of the desert, there are thought to be many more ruins lying undiscovered beneath the sands of the Taklimakan. The most likely location of such ruins is the strip of desert immediately north of the current "Silk Road" (a modern highway), an area where gradual desertification has forced many communities to be abandoned.

Until recently, searching for desert ruins was always an arduous, hit-or-miss undertaking, but the advent of Shuttle Imaging Radar (SIR) is revolutionizing the field. Imaging Radar is a synthetic aperture technique that produces radar-frequency "images" of the earth's surface. The radar technique has been found to be especially effective in locating desert ruins because it tends to highlight caravan trails and riverbeds (two ground features that invariably accompany human habitation), and because it can image objects buried under up to three meters of dry, fine sand. NASA's Shuttle Imaging Radar is operated from the Space Shuttle. So far it has been flown twice, and is credited with the discovery of the Lost City of Ubar, an ancient city located in the Arabian Desert.

In September 1993 a CERS team visited the southern Silk Road to learn about the conditions there and to assess the potential for the Shuttle Imaging Radar to locate undiscovered Silk Road ruins. The purpose was twofold: first, to learn about conditions so that future expeditions to potential ruins sites can be adequately equipped and prepared; second, to characterize the local topography and (in particular) to examine known ruins to serve as a reference for the Imaging Radar Shuttle mission to be flown in April, 1994. It is this second goal that forms the subject of this report.


The observations that follow are based on the following sites: Maligawat (Hotan County), Old Damagou (Chele County), Mazha (Yutan County), a fortress at Shezikebujang (Qiemo County), and Washisha (Ruoqiang County).

The structures can be classified into two types: wood and earth. Wooden ruins are found in younger sites (Qing dynasty). The remains of wooden structures usually consist of rows of closely- spaced sticks or branches stuck in the sand; these were once walls. It is possible that wooden beams or columns, up to about 8 inches in diameter, also remain. The sticks, which are typically 1-3 cm in diameter, that compose the ruins are usually broken off or buried so that no more than half a meter lies above ground level. The walls are arranged to form a rectangle which is limited in size by the length of available beams used in the dwelling roof (about 7 meters). Sometimes the walls are arranged in an irregular round shape--probably enclosures not dwellings but yard enclosures or livestock pens.

Earth structures are usually more durable than wooden ones; consequently older sites (Yuan dynasty and before) consist entirely of earth remains. Most of these are walls--typically very thick (probably up to 2 meters originally)--eroded so that no vertical surfaces remain, only irregular piles or knobs sticking up out of the ground. Examples of these were seen at Maligawat and the fortress in Qiemo. Sometimes the site is so deeply buried or so completely eroded that only an outline remains on the ground; for example, at Washisha it appears that the thin walls of a home survive, though the tops are only 10-20 cm above ground level. Two other kinds of earth remains were seen: at Maligawat huge blocks that were once Buddhist stupas have left irregular platforms up to 2 meters high; and at Washisha, a maze of irregularly-shaped platforms, all about 1 meter high, that perhaps are the result of earth compacted under feet of the city's Yuan-Dynasty inhabitants.


The ruins visited were all found in areas of dwindling water supply; most were surrounded by dead trees and a lot of dry wood. Some living vegetation still survives, but only because of underground water. The most common type of topographic feature is a round hill of sand with steep sides and bushes growing on the top, held together by a great tangle of roots. In every site visited, these mounds were taller than the ruins--up to 5 or 6 meters high and 10-15 meters in diameter. There is usually a fairly well-defined ground level on which little vegetation grows. On the ground, blowing sand undoubtedly changes the topography from season to season. In areas where mounds are not found or are widely spaced, the sand forms dunes; which in the sites visited were no more than one or two meters in amplitude.

The "soil" on all the sites visited consists almost completely of fine, dry sand; sometimes so fine that it could be called dust. On most of the sites, the sand contained few if any stones. (Stones tend to be more common in the south, where they have been washed down from the Kunlun Mtns, and are also abundant in river- beds). A marked feature of the ground around ruins sites, however, was the presence of pot-shards scattered over the ground. Apparently the shards tend to be exposed as sand lying over them is scoured away by the wind. A typical shard is about 10 square centimeters in area and 1-2 cm thick. They consist of red or black unglazed clay, usually are rough-textured and devoid of decoration.

The areas visited had few modern-day human inhabitants, but a few pastoral families still manage to eke out a living in this marginal region. These contemporary houses are composed of earth and wood--probably no different from dwellings used in ancient times. These herdsmen obtain their water from wells some 5 meters deep, and also build corrals of mud and brush for their livestock. The roads that serve these areas are little more than rough tracks through the sand.


The biggest obstacle to radar detection of ancient Silk Road ruins is the presence of ground features (the mounds) that are taller and larger than the objects being sought. It is not clear whether image quality will permit discerning man-made structures among these natural behemoths. Moreover, because ancient earth walls are so heavily eroded, there are virtually no flat surfaces or corners to present a distinct radar signature. An exception to this are ruins of the type found at Washisha, where the earthen structures are shaped like jigsaw puzzle pieces--broad and flat-- ruins of this kind ought to be easier to see. In the case of walls, the fact that they are markedly different in size from the mounds, and the fact that the pieces are arranged in rectangles, may help to distinguish them.

The remains of wooden walls, due to their short stature, should also be very difficult to detect. Nevertheless, because they are composed of different material than the surrounding sand, they should have a different reflective coefficient, and may prove to be better reflectors than the natural vegetation. This may give them a detectable radar signature, particularly where heavy beams or columns have survived.

Perhaps the best indicator of ancient ruins will prove not to be the walls, but the potshards scattered over the sand. Fine dry sand is a poor reflector of longer radar wavelengths (hence the radar's ability to image buried objects); however the presence of potshards will perhaps increase the surface reflectivity in a measurable way, causing areas of former habitation to appear "brighter" on the radar images. In stony areas, potshards will have little effect; but in most of the region of interest, stones are found mainly in riverbeds, which are themselves both indicative of civilization and easily distinguishable from it. Hence, unnaturally bright areas on the radar image which are not riverbeds may be sites of ruins.


Until the radar images ares actually collected and compared with ground observations, one can not predict with certainty how successfully Silk Road ruins can be detected from space. Nevertheless, the following tentative predictions can be made: = (1) The easiest types of ruins to detect will be upward-facing reflective surfaces (like those at Washisha), and large walls and pagodas that are far from any comparably-sized natural features (like at Maligawat).

(2) Ruins consisting only of heavily-eroded walls situated in the middle of a forest of mounds will be difficult to detect.

(3) Wooden remains may be detectable, depending on the difference in reflectivity between wood and sand, and the presence of nature vegetation close to the site that will tend to obscure it.

(4) Potshards scattered over the sand may distinguish what are otherwise difficult-to-identify ruins sites.

(5) Riverbeds ought to stand out very clearly in the radar images due to the heavy concentration of stones in them and the relative absence of stones in the surrounding sand.

Return to Silk Road Project home page