¿ìè¶ÌÊÓƵ

The human eye is no longer good enough for the discriminating archaeologist.
Where you or I might see only an ordinary field of tall grass and trees,
an archaeologist knows that beneath it may lie a Palaeolithic village, undisturbed
for 30 centuries. But the eye detects only the narrow band of visible light
that lies between 0.4 and 0.7 micrometres, so even our archaeologist is
unlikely to see signs of the buried village.

The evidence is still there, however: it lies in those parts of the
electromagnetic spectrum that are invisible to the eye. For decades, aircraft
and satellites have taken images of the Earth in these non-visible wavelengths.
Although only able to distinguish among vegetation, soil and water, such
images have given geologists clues to where minerals are buried and helped
foresters to track down logged areas of rainforests. In recent years, archaeologists
have also learned to read these images, the result of techniques known collectively
as remote sensing.

This technology relies on quirks of soil temperature and vegetation.
Subterranean structures affect the moisture and temperature of the overlying
soil, which in turn cause subtle changes in the temperature, moisture and
chlorophyll content of the plants on the surface. Tilled soil, enriched
with phosphorous or nitrogenous fertilisers, also stands out against the
background soil. Differences such as these can point to archaeological features
which would otherwise be invisible. Tom Sever, NASA’s only archaeologist,
is an unabashed advocate of remote sensing technology. ‘Everything in the
Universe has an emissivity,’ he explains. ‘If your finger is at 98 degrees
F and your ham sandwich too, they may be the same temperature but they emit
energy at different rates. A sensor can detect that.’

When, in 1982, Sever convinced NASA to test some of its newer remote-sensing
instruments at an archaeological site, many archaeologists were sceptical.
They had expected great things when NASA launched the first Landsat satellite
in 1971. As it turned out, these satellite images were a disappointment;
they were much too coarse for archaeological needs. ‘A lot of money and
effort was spent that didn’t work out,’ recalls Sever. But Sever believed
that remote-sensing equipment similar to that on satellites would work from
aircraft. At the time, archaeologists had already tried aerial photography
in the visible and infrared range, with some success. Satellite pictures
would have several additional advantages: archaeologists could buy them,
removing the need to hire a plane; and the images cover larger areas.

Sever’s first detailed experiments began in April 1982, when two remote-sensing
instruments, a thematic mapper simulator (TMS) and a thermal infrared multispectral
scanner (TIMS), were flown in a small jet over Chaco Canyon in northwestern
New Mexico. Cut into the sandstone of a high, dry plateau, Chaco Canyon
is one of North America’s oldest sites of human habitation. Signs of settlement
by nomads appear at about 5500 BC but the site blossomed after 1000 AD under
the Anasazi, agriculturalists who built multistorey buildings and a complicated
network of roads.

Although many features had already been discovered by archaeologists,
Sever believed he could uncover ones they had missed. It was also a good
place to test the mettle of his new instruments at an altitude that could
give him images with a resolution of a few metres. The first instrument,
TMS, recorded emissions in seven bands, spanning the visible to the beginning
of the infrared (near infrared), through the far infrared to the farthest
portion of the infrared, known as the thermal infrared. TIMS focused on
only six small bands within the thermal infrared region.

Sever’s results changed archaeology forever. Buried prehistoric walls,
buildings, agricultural fields and roadways stood out as if daubed in paint.
Images taken in the thermal infrared region were especially revealing. These
reflect slight differences in the temperature of objects, even to a few
degrees. In arid environments, the surface of the earth is a mosaic of temperatures;
sand, soil and rock not only retain heat differently, but they emit heat
at different rates.

The texture of the soil also affects its temperature, so loose soil
over buried walls or old fields is easy to spot. Irrigation ditches filled
with sediment, for example, give themselves away because they hold more
moisture and less heat than the surrounding earth. Even the undulations
and edges of an ancient roadway create a recognisable temperature pattern.

Since the Chaco Canyon experiment, Sever and others working with him
have developed a more sophisticated approach to the use of remote-sensing
images – including the latest satellite images – in archaeology. For example,
research in Latin America over the last four years has taught them to use
different parts of the spectrum in humid and dry climates. In humid conditions
the temperature of vegetation varies little from that of the atmosphere,
and any evidence of human activity tends to be obscured because the picture
is not so clear. Instead of recording temperature in the thermal infrared,
archaeologists take images in the near infrared to detect variations in
the ‘exuberance’, or relative vigour, of plants, as measured by their level
of moisture and chlorophyll. Remote sensors apply the colour red to the
signals, so the more ‘exuberant’ the vegetation, the brighter red it appears
in the images. More exuberant vegetation can point to natural features such
as a stream bed. But it is often a clue to settlements or fortifications,
which are characterised by disturbed soil. This tends to hold moisture differently
from the surrounding earth, encouraging the growth of more exuberant vegetation.

In addition to airborne scanners, satellite images have now become a
common part of archaeology’s tool kit. The currently orbiting Landsat satellite
has a resolution of 30 metres, while France’s SPOT satellite gives 20-metre
views. These are good enough to provide thermal and vegetative informationfor
large features such as roads or major fortifications. And archaeologists
are hiring aircraft and even planning tethered balloons to carry scanners.

Remote sensing for archaeological purposes is still novel enough, however,
that expeditions using the technique are usually considered experimental.
Skill is required in choosing the correct bandwidths and instruments: local
conditions such as soil types, vegetation, weather, and what sort of archaeological
features are sought all determine which region of the spectrum to use. Since
his first experiments, Sever has found that observations in different bandwidths
identified some objects but not others, and that combinations of bandwidths
sometimes worked better than single views. On the ground, the flux of heat
from different features varies over time; by 10 pm, for example, prehistoric
roads had reached thermal equilibrium with their surroundings and become
indistinguishable. Sever has begun a project to create a standard reference
that matches bandwidths with different archaeological features. To do this,
he and his colleagues in the US are collecting the experiences of archaeologists
who have compared remote imagery to what they’ve found on the ground.

Payson Sheets, an archaeologist at the University of Colorado in Boulder,
solved a mystery in Costa Rica in this way, when he and Sever looked at
an infrared image of the Tilaran region in the country’s northwestern corner.
‘What surprised us (in the photo) was an odd, twisting line,’ says Sheets.
It started near an ancient cemetery (dated between 500 and 1200 AD), ran
downhill, took a bend around a buried rock pile, split into two at a stream,
unified on the other side, then skirted another rock pile before petering
out.

When they excavated part of the line, they discovered it was a buried
footpath, lying between 1 and 5 metres under volcanic ash, and at least
1000 years old. It had begun as compacted earth only one-third of a metre
wide, but erosion of the path’s borders had widened it to almost 5 metres
at some points. Successive inundations of fine ash, or tephra, from the
nearby Arenal volcano, which has erupted four times since 175 AD, buried
and preserved it. The path appeared as dark red lines on the infrared images.
‘That’s because the pasture grasses, with their roots penetrating 1 to 1.5
metres, grow better along the paths,’ explains Sheets.

Successive overflights in the area using TIMS found the same path and
many others, but by another means. The buried paths are cooler than surrounding
areas because their soil is less compact and does not hold the heat of the
day as long. The team was also able to locate the paths with airborne radar,
which offers the advantage of penetrating the 50-metre high tree canopy
in Costa Rica’s forested areas.

Sheets and Sever are currently uncovering a network of paths connecting
villages with other archaeologically interesting sites. Many paths run from
different villages to communal cemeteries. Burial stones beside the paths,
and other paths leading from the cemeteries to streams and rubbish dumps,
suggest that people spent lengthy periods at cemeteries, perhaps for ritualistic
communing with ancestors, says Sever.

Remote sensing is even giving archaeologists a new insight into well-documented
countries such as Greece. Satellite images from Landsat and photos taken
from the space shuttle have revealed new quarries, citadels and other features
of early Hellenic civilisation. A team of archaeologists and remote sensing
experts from the University of Minnesota have been going over well-trodden
ground in western Peloponnesos, where civilisations flourished in the middle
and late Bronze Ages (between 1700 and 1200 BC) and again during the classical
period (between 800 and 300 BC).

Despite scouring the surface for pottery fragments, the group had missed
two important classical sites that were later discovered by chance – ruins
of Prasidakion, hidden under dense brush, and ruins of a temple at Periovolia.
So the team equipped themselves with microcomputers and software to interpret
images from Landsat before setting to work in the vicinity of Pylos, a prosperous
city during the Bronze Age.

One of the archaeologists from Minnesota, Frederick Cooper, was able
to combine with remote sensing the skill he calls phytoarchaeology – the
study of plants as clues to archaeological sites. ‘In rocky areas,’ Cooper
explains, ‘when the entrance to a tomb was cut, it was usually backfilled
with soil. From then on, that site produces richer vegetation.’ This vegetation
has a different spectral emission from the surrounding vegetation, so it
can be detected by remote sensing.

In 1990 the team applied this combination to locate previously unknown
sites between Pylos and Olympia to the north, using the clue that the scrub
oak known as kermes in Greece prefers limestone soil: ancient fortifications
or quarries are rich with limestone. In satellite images of their study
area, they saw unusual colours and patterns around known ruins of citadels.
On the ground, they discovered that the scrub oak was indeed responsible
for the noticeable redness of the images – the oak had grown over the old
circular walls, while some peculiarity of the soil over the sites made the
sites emit noticeably in the blue and green bandwidths of the visual region.
After studying the satellite images in more detail, the team returned last
year.

Near the town of Vrina they discovered previously unknown citadels,
the oak growing over their buried walls, and abandoned quarries. They also
found ancient clay beds that would have provided raw material for terracotta
pottery, bricks and roof tiles. ‘We are not sure why it works,’ archaeologist
Joseph Alchermes says of the distinctive satellite images. ‘It could be
a combination of masonry remains and the vegetation that likes limestone
²õ´Ç¾±±ô.’

Landsat images, with a resolution of 30 metres to a pixel, are usually
too coarse to enable researchers to pick out the geometric details that
can indicate hid-den buildings . As the Minnesota team has proved, however,
anomalies in vegetation and soil can be clear enough clues. Sometimes, even
the coarser Landsat images in the visible part of the spectrum are good
enough. Cooper and his colleagues spotted abandoned irrigation channels
near the town of Andravida from such images. The channels appear dry to
observers on the ground, but they still held enough moisture for remote
sensors to register. Again, a knowledge of local conditions was essential:
water emits characteristically in the blue region of the spectrum, but water
vapour in the atmosphere normally washes out these views. Greece’s dry climate
enabled them to use the method there. When the team returns to the area
next year, Cooper expects remote sensing to reveal hundreds of hilltop fortifications
never discovered because of the rugged terrain.

Remote sensing’s reach can also be measured in metres. Anna Roosevelt
at Columbia University in New York has adopted a suite of ground-based remote-sensing
techniques for her work in northeastern Brazil. Roosevelt’s team used her
techniques at the Ilha de Marajo, a flood plain near the mouth of the Amazon
River, where she was looking for the remains of mound villages that appear
to date back to at least AD 400. Roosevelt takes the Marajoaran settlement
of perhaps 100 000 people as evidence that concentrated, agricultural societies
can survive, even prosper, in Amazonian forests, even those that flood seasonally.

‘Most archaeologists just start digging,’ Roosevelt explains. ‘It might
take 10 years to do a big site. Why not survey it geophysically in two weeks?’
Roosevelt hired geophysicists and started mapping sites at Marajo in 1988
using resistivity instruments. In this technique, four electrodes are placed
into the surface soil, about 15 to 20 centimetres deep and about 5 metres
apart, and a charge is passed between them. Variations in the conductivity
of the soil identify strata to a depth similar to the distance between the
electrodes. Artefacts beneath the surface, such as fired brick hearths,
show up because they are denser, and so more conductive than surrounding
soil. Loosely packed earthworks conduct less well than packed soil. Anything
that collects moisture, such as a buried structure, is likely to be more
conductive.

Roosevelt also used seismography – the technique of sending shock waves
through the earth and recording how, and how fast, they bounce back to the
surface. As well as helping to map the strata, seismographs also detected
voids in the earth at Marajo that contained burial urns (a fact well known
to looters, who perform their own crude seismography by stamping their feet
and listening for a hollow reply).

Proton magnetometers also detected the fired clay urns as magnetic anomalies,
as well as hearths that led excavators to two buried houses. Roosevelt and
her colleagues eventually excavated 45 sites based on geophysical evidence
available almost instantaneously on portable computers connected to the
instruments. She is now applying the technology at Santarem and other locations
in the northeastern Amazon.

Information such as Roosevelt’s can be converted into useful images
by computer software called geographic information systems (GIS). Developed
in the mid-1980s, it has recently become popular with archaeologists because
it can translate data from maps or remote sensing instruments into images
that can be manipulated on a computer screen. An archaeologist like Fred
Limp at the Arkansas Archaeological Survey in the US believes he can recreate
history with it.

In the 16th century, Hernando de Soto explored what is now Florida in
search of the New World’s gold. He left a sketchy record of his travels,
but it was enough to match a few places and dates. Limp used GIS to plot
the contours and slopes of the hills and valleys and the courses of rivers,
lakes and bogs that would have influenced the route taken by a traveller
on foot or horseback. The computerised result is the optimum route that
de Soto logically would have taken, and as such is a guide to possible artefacts.
As it turned out, written records suggest that de Soto did not stick to
that optimum route but turned toward some low, less passable mountains.
‘He came from Peru, where he had been a conquistador, and he expected gold
in the mountains, not in the river valley,’ explains Limp. Nevertheless,
GIS not only improves site-mapping but has revolutionised what Limp calls
‘non-site’ archaeology, or the study of how people used the landscape .

Archaeologists will probably never give up their trowels, but few are
likely to resist technology that reduces the number of empty holes they
will dig. Sever predicts that new thermal sensors flown in balloons will
soon produce images with resolutions measured in centimetres. Meanwhile,
NASA is designing a multi-billion dollar series of new satellites, known
as the Earth Observing System, to be launched in the late 1990s. Their goal
is to read even more closely the script of the Earth’s surface. Sever and
others in the sensing camp are confident that new discoveries await them
in the fine print.

Christopher Joyce writes about natural sciences and the environment
from Annapolis, Maryland.

Douglas Fuller is a graduate fellow at the University of Maryland and
a research associate at NASA’s Goddard Space Flight Center

Mark Noel manages the archaeological consultancy Geoquest Associates
and teaches archaeology at the University of Durham.

* * *

1: DIGGING OUT THE PAST BY SATELLITE

Remote sensing for archaeologists begins with sensors called radiometers,
which measure energy in the visible, infrared or microwave portions of the
electromagnetic spectrum, between 0.4 and about 14 micrometres. They consist
of arrays of tiny detectors, each composed of different inorganic compounds
sensitive to specific wavelengths. A prism or a fine grating (or, in the
case of microwaves, an antenna) filters the photons of radiation into bands
of different wavelength, which are then directed by optical devices such
as scanning mirrors to the appropriate detectors. As a photon strikes a
detector, its energy is absorbed and converted to an electronic signal.

The size of the detectors and the altitude of the satellite determine
the sensor’s resolution, or ‘instantaneous field of view’. In essence, the
field of view is the smallest area on the ground that the instrument can
identify. The smaller each detector is, the smaller area of the ground,
or ‘ground resolution element’ or ‘pixel’, that it records. The smaller
and more numerous the pixels, the greater the spatial detail in the image.

The Thematic Mapper (TM), a radio-meter aboard one of the Landsat satellites,
epitomises the current generation of space-borne multispectral scanners.
Launched in 1984, the TM contains arrays of silicon detectors to measure
radiation in the visible range, with wavelengths between 0.4 and 0.7 micrometres,
and in the near-infrared range, with wavelengths between 0.7 and 1.1 micrometres.
It has indium antimonide detectors to sense infrared energy with wavelengths
between 1.1 and 3.0 micrometres, and mercury cadmium telluride ones to sense
infrared energy at the other (thermal) end of the band, with wavelengths
of 3.0 to 14 micrometres.

The resolution of the TM is 30 square metres, making it among the best
non-military sources of high resolution images. Its weakness, however, is
that it views only seven spectral bands at various points across the spectrum,
rather than viewing continuously from visible to thermal infrared wavelengths.
In a sense, this is like seeing only seven hues in an oil painting; emissions
or reflected light from objects outside these bands are not picked up.

Archaeologists can expect the available technology to improve, however.
Engineers are designing new orbiting imagers, known as spectrometers. These
incorporate many more and slightly different detectors which, unlike radiometers
like the TM, can resolve incident energy into several hundred adjacent wave
bands, each a few nanometres wide. Also, spectrometers can distinguish subtle
variations in the level of energy within a band, something radiometers cannot
do.

As a result, spectrometers paint a more complete spectral image of an
object, so observers can detect objects in more detail, and create a more
reliable ‘signature’ for different types of object. There is a price to
pay, however; the narrower the spectral band, the less the energy available
to generate a signal. So the signal is more likely to be lost in the electronic
‘background noise’ of the instrument.

Douglas Fuller

* * *

2: FROM TROWEL AND SPADE TO MOUSE AND MEGABYTES

Without even lifting a spade, archaeologists can now ‘excavate’ a site
by tracing subtle variations in the Earth’s magnetic field, or in the electrical
and thermal properties of the soil. This type of geophysical archaeology
is already yielding a rich bounty at Lanchester Roman Fort about 7 miles
west of Durham.

The Lanchester camp covers about 20 000 square metres of an elevated,
windswept site on Dere Street, a main Roman road about 50 kilometres south
of Hadrian’s Wall. It was one day’s march for a Roman legionnaire to the
neighbouring strongholds of Ebchester and Binchester. From trial excavations
in the 1930s, and occasional discoveries of inscriptions, archaeologists
have dated the fort to the reign of Emperor Antoninus Pius, around the middle
of the 2nd century, or earlier. They also know that the fort was empty for
a short time in the late 2nd to the early 3rd century, before being rebuilt
in stone during the reign of Emperor Gordianus III. The site was finally
abandoned in the mid or late 4th century, but the ramparts survived to a
height of almost 4 metres until they were plundered for building materials
in the 18th and 19th centuries.

Today, the fort is a featureless pasture encircled by scattered relics
of Roman masonry, but it remains an important historic site. Archaeologists
aim to document it with the help of a geophysical programme which I set
up at Lanchester in the summer of 1990, with the help of a Roman military
archaeologist, John Casey and students from the University of Durham.

Geophysical techniques have several advantages over traditional methods.
Geophysical surveys are cheap, fast, and leave the evidence intact for future
generations to study. One alternative to digging that our team at Lanchester
might have considered was to inject weak alternating currents into the ground.
We could have measured its electrical resistivity to build up a picture
of the contrasting levels of soil moisture associated with buried walls,
pits and ditches. Instead, we chose to do a geomagnetic survey. In the presence
of the Earth’s magnetic field, buried features are weakly magnetised, to
an extent that depends on a property called the magnetic susceptibility.
This is the basis of a faster surveying technique, which was possible at
Lanchester because there was a large enough difference between the magnetic
susceptibilities of the local building sandstones and soils. Also, magnetic
responses do not change seasonally, unlike electrical resistivities which
depend on variations in rainfall.

Magnetic responses from buried objects are affected by daily fluctuations
in the Earth’s magnetic field, however. To get round this problem we used
an instrument known as a gradiometer. This compensates automatically for
the variation by recording the Earth’s magnetic field with two sensors placed
50 centimetres vertically apart and subtracting them electronically.

The team surveyed the Lanchester fort by first dividing the site into
a mesh of 137 rectangular units, measuring 20 metres by 10 metres. By deliberately
aligning the mesh at an angle to the grid on which the fort had been laid
out, we ensured that important features, such as walls, did not lie along
the seams of the mesh, where they would be more difficult to interpret.

Geophysicists measure features with reference to the regional average
for uniform ground containing no buried structures. For uniform ground they
would not be able to detect any variation in the Earth’s magnetic field.
But because the features are there, variations or anomalies appear in these
readings. Archaeologists can relate them directly to walls, ditches and
other buried structures and can make them ‘appear’ by transforming the data
into a map.

At Lanchester we took measurements at half-metre intervals – small enough
to ensure that there was enough information to work out what was buried,
but large enough not to be overwhelmed with data. After a survey that lasted
through the grim winter months of 1990/91, the geophysical team were finally
blown off the site clutching a floppy disc bearing 93 552 measurements.
These data were then passed through an image-processing program that blended
the seams between grid units, reduced the range of readings to reveal detail
within strong anomalies and from this created a map with 33 shades of grey.

The result is the most detailed geophysical image of a British Roman
fort. From a study of the anomalies, they have pinpointed the fort’s major
structural elements, such as roads, buildings and hearths and compared them
with similar camps on the Roman frontier. The headquarters building or principia,
with its set of offices, cross hall and courtyard, is similar to the headquarters
at Chesters, near Hexham in Northumberland. To the north of the principia
is a pair of granaries with buttressed walls; to the south, the remains
of the commandant’s house. Between the headquarters and the east gate are
a set of eight barrack blocks with contuburnia. Contuburnia were two-room
units for eight soldiers, one for sleeping and one for cooking. They were
built in blocks of 10 with a larger unit at one end for the centurion.

Several strong magnetic anomalies probably signify hearths in the barracks,
and there is evidence of the remains of several turrets and corner towers
built into the ramparts. A soil-filled aqueduct appears as an anomaly that
curves through the western part of the fort, marking an earlier series of
barracks, or possibly stables. Fainter features on the image suggest that
earlier buildings and ditches are preserved beneath those providing the
strongest responses. These may date back to a fort of turf and timber that
probably first stood on this site.

Resistivity tomography and radar imag-ing are currently adding a third
dimension to the survey at Lanchester. Results of this latest work will
be published later this year.

Mark Noel