Archaeology from the Air

Juan Carlos Fernandez-Diaz. American Scientist. Volume 104, Issue 1. Jan/Feb 2016.

Driven by dreams of fame and wealth beyond imagination, the Spanish conquistadors of the 16th century were willing to face great hardships and to risk their lives in a search for the “cities of gold” common to the mythology of the aboriginal peoples of Central and South America. Although they easily defeated the local inhabitants, they themselves were defeated by the rugged terrain and dense rain forest in the interior of the country. The conquistadors never found the Ciudad Blanca (White City) of Honduras, said to contain fabulous treasures. Just cutting a trail through the lush vegetation of the rain forest is onerous enough. To do so while enduring constant attack from an endless variety of insects- and also remaining continually on the watch for jaguars, tapirs, and poisonous fer-de-lance snakes-was enough to dissuade even the most enthusiastic treasure hunters from exploring much of the region on foot.

Five centuries after the arrival of Europeans in the New World, large extents of the area known as La Mosquitia, or the Mosquito Coast, still remain uncharted-a mystery that continues to intrigue not only modern-day explorers but biologists, zoologists, geologists, and archaeologists alike.

Developing the Right Technology

No longer seeking to plunder but to examine and understand, modern day explorers have fared slightly better than their predecessors. Working with the renowned archaeologist Alfred Vincent Kidder, the famed aviator Charles Lindbergh enjoyed some success with airborne archaeological exploration in the sparsely vegetated drylands of the southwestern United States. He also explored areas in Mexico and Central America. When he reportedly succeeded in filming a “lost Mayan site” in July 1929, a New York Times editorial praised him as a “pioneer of aerial archaeology.”

By contrast, areas covered with rain forests have been far less easy to explore, because everything beneath the forest canopy remains veiled in shadows even when the Sun is nearly overhead. The overgrowth of tropical vegetation also provides a natural camouflage that makes it nearly impossible to discern the ruins of even the largest human constructions, whether these are pyramids, plazas, ball courts, or roadways-all of which are common to the ancient cities of Central America. It would take more than airborne photographic exploration to reveal what lay beneath the rain forests of the world.

Beginning in 1957, the Space Age led to the development of powerful new remote sensing technologies. NASA launched a succession of satellites equipped with ultraviolet, infrared, and visible sensors designed to image the surface of Earth in ever more detail. Dedicated satellites, as well as the space shuttle, carried microwave and laser altimeters that created lattice works of profiles across the surfaces of the oceans, polar ice caps, and continents. NASA scientists and academic researchers pored over the observations, trying to see through the dense rain forests of Central and South America, Southeast Asia, and Equatorial Africa. Tantalizing as some of the images derived from the space-based sensors were, their spatial resolution (usually tens of meters) was simply not good enough to reveal archaeological ruins hidden in the rain forests.

The Launch of Airborne LiDAR

By the early 1990s the Global Positioning System (GPS) was reaching full operation, and inertial measurement units (IMUs), developed for the guidance of intercontinental ballistic missiles, were becoming available to both government agencies and private companies. These units are composed of triads (three sensors mounted at 90 degrees to one another) of optical gyroscopes and integrating accelerometers. The IMUs made it possible to determine changes in the orientation (roll, pitch, yaw) and accelerations, along and across (horizontally and vertically) track, of an aircraft. Precise optical scanners, highspeed chip-based electronics, and compact lasers with pulse rates of a few thousand pulses per second and pulse lengths of 10 to 15 nanoseconds could be assembled into light detection and ranging (LiDAR) units. By combining the LiDAR data with those from the GPS and IMU, it is possible to determine the north, east, and up coordinates of millions of points spread over hundreds of square kilometers of terrain in a matter of hours rather than the weeks or months required for classical mapping methods.

Three-dimensional plots of the unequally spaced reflections from vegetation, structures, ground, and water tend to be amorphous and are referred to as point clouds. The accuracy of each point in the LiDAR point clouds depends on the flying height and speed of the aircraft, flight dynamics associated with atmospheric winds and turbulence, and systematic and random errors arising from imperfections in the mechanical, optical, and electronic components of the unit.

The early development of airborne LiDAR was not driven by any intent to use it for archaeology, or even a consideration that it might one day be used to produce high-resolution images of the ground in areas covered with dense rain forests. With laser pulse rates of only a few thousand per second and laser footprints a few decimeters in diameter, the first airborne LiDAR systems illuminated just a small percentage of the surface of the terrain in a single pass. But even the earliest LiDAR units could collect sufficiently dense returns from the ground to produce topographic maps suitable for many engineering applications faster and at less cost than ground surveys, aerial photogrammetry, or space-based observations. Managers of such agencies as state departments of transportation and environmental protection initially were skeptical of the new technology, but the results of test projects were undeniable: Airborne LiDAR enabled them to reduce the time and cost of acquiring the information they needed to build new highways, measure beach erosion, delineate flood zones, and quantify damage from forest fires and other natural disasters.

As the engineering applications of airborne LiDAR grew, researchers in the earth sciences community began to take note of the detailed maps, shaded relief images, and other products derived from the observations, and to think about scientific questions that they might be able to answer using the technology. LiDAR units were expensive, however (about $1 million each in the mid-1990s), and even if they could find the funds to purchase a system, busy earth scientists did not want to take the time to learn how to collect and process the observations themselves. Contracting with commercial mapping companies, which were just beginning to learn the technology, was not the answer they were seeking. To gain a competitive edge, most mapping companies developed their own procedures and computer software, which they refused to fully disclose, even to academic researchers. What the scientific community needed was a reliable source of high-quality LiDAR observations and products derived from them, collected and processed for their specific research goals by experts with whom they could communicate, or perhaps even collaborate.

In 2003 the National Science Foundation (NSF) established the National Center for Airborne Laser Mapping (NCALM) for the purpose of collecting airborne LiDAR observations of research quality-data of sufficient accuracy and density to meet the needs of the specific research application, which is generally getting more stringent with time-for projects funded by NSF through peer review of proposals. Two other important missions of NCALM are to support the advancement of airborne LiDAR technology and to educate and graduate college students to fill the needs of government agencies, academic institutions, and private companies.

Many of the projects proposed by the scientific community shared a need for high-resolution images (down to one meter or less) of the bare earth-that is, the surface of the ground stripped of vegetation, be it brush only a few meters in height, or the tall redwood forests of California. Developing the capability of seeing (so to speak) through vegetation became a high priority for NCALM. Obtaining LiDAR units with much higher laser pulse rates was part of the answer, but other factors also came into play, particularly the flying height and speed of the aircraft.

It may seem obvious that the aircraft (typically a small prop plane) should fly slowly and at low altitude to collect the highest quality LiDAR observations. The lower the aircraft flies-between 400 and 600 meters above local ground level, for most scientific observations-the stronger the return signals are. At a fixed rate of pulses per second, the more slowly the aircraft flies the more points per square meter the coverage contains. However, the laser is not always operated at its highest possible rate, because the energy per pulse typically decreases at higher pulse rates; some applications, including the mapping of areas covered with dense vegetation, may require the highest possible energy per pulse to make the ground returns strong enough to be detected by the LiDAR sensor.

Perhaps a less obvious, but even more important, reason for flying low and slow is to make it possible for a pulse of laser light to follow virtually the same path through vegetation on its way to the ground and on its return to the aircraft. If the aircraft flies too fast, the reflected light will have to take a different path through the vegetation to return to the sensor, and the chance of there being a second opening in the vegetation in just the right direction at the required instant of time is, for practical purposes, nearly zero. It is this need for thousands of pairs of simultaneous pathways through the foliage-one path for the light to reach a point on the ground, and another path for the reflected light to reach the camera, at the time of each shutter opening-that makes simple aerial photography nearly useless for obtaining an image of the ground in forested areas.

There is another difficulty that must be overcome in using airborne LiDAR to see through vegetation. In terrain covered with dense vegetation the great majority (perhaps 99 percent or more) of the laser pulses encounter vegetation before reaching the ground and produce reflections strong enough to trigger the sensor. Fortunately, the vegetation commonly intersects only a fraction of the laser beam, and sufficient light continues to the ground to produce a reflection that is strong enough to be detected. But it is important that the sensor and associated timing electronics recover quickly-or, in other words, that they have a short dead time-after receiving a reflection; otherwise, returns from vegetation near the ground will cause the loss of many returns from the ground itself. Modern airborne LiDAR units generally have sensors with dead times of only a few nanoseconds, with electronics capable of recording several discrete returns per laser pulse. Some units actually digitize and record the full waveform of the reflected light for each laser pulse.

New Maps for Archaeological Sites

By about 2006, technology had advanced to the point that it was possible for commercial companies to build lasers suitable for airborne LiDAR units with pulse rates well over a hundred thousand pulses per second, and the spatial resolution of bare-earth digital elevation models and images derived from airborne LiDAR observations rapidly improved to the level of a single meter, even in areas covered with moderate to heavy vegetation. Archaeologists working in temperate climates began to use airborne LiDAR to map ancient ruins, with excellent results. But those working at sites in more tropical regions were reluctant to risk their funds and time on the technology, fearing the laser would not be able to penetrate the denser vegetation well enough to produce images useful for their studies.

Finally, in 2009 a group of American archaeologists led by Arlen Chase of the University of Central Florida obtained funds to use airborne LiDAR to map the ancient city of Caracol, Belize. Members of the group had been visiting the site on foot to map and excavate ruins for nearly three decades. They had documented the remains of hundreds of buildings, roadways, and agricultural terraces, but suspected that the limits of the ancient city might extend well beyond the area documented. It was hard, time-consuming work cutting lines-of-site through the thick vegetation to do ground surveys. And the latticework of surface profiles derived from such surveys left large gaps in the coverage that might well contain important, but as yet undetected, archaeological features.

Researchers from NCALM and the University of Florida agreed to map an area of some 200 square kilometers, reaching well beyond the known extent of Caracol. Large overlaps of adjacent swaths were used to obtain a higher point density (nominally 20 points per square meter), and enable analysts to identify erroneous points, as well as remove systematic offsets in the aircraft trajectories. Much to their surprise, the analysts discovered clumps of points scattered throughout the project area, which appeared to be well below (some as much as several meters below) the surrounding ground level. A closer look revealed that many of these clumps contained points from two or more swaths, including those collected with the aircraft flying in opposite directions. Clearly they were not caused by some strange malfunction of the LiDAR unit-they marked openings to subterranean features of some sort. Perhaps wells or cisterns?

The archaeologists immediately dismissed the likelihood of wells or cisterns, suggesting instead that the LiDAR observations had more likely captured the openings of caves common in the karst topography. Using coordinates derived from the LiDAR observations, archaeologists at the site were able to walk to a number of previously unknown cave openings hidden in the tropical vegetation, some located just tens of meters from pathways they had been using for years. A more focused search of the LiDAR observations identified possible openings to more than 50 caves-a bonus discovery due entirely to the LiDAR observations. Most of the caves will require climbing gear to enter, and it will take years to explore them all, but the early results are promising. One of the first caves to be explored was found to have at least four “rooms,” presumably formed by erosion as water flowed through the cave, and ancient artifacts discovered there indicate the enclosure was once used in some way by people living in the area.

When members of the archaeological team received the digital elevation models and geodetic images of Caracol, they were amazed by the number of previously unknown buildings, walls, and causeways they could immediately detect. Perhaps even more surprising were the extensive areas of agricultural terraces, some differing from adjacent terraces by as little as 30 centimeters in elevation, which the LiDAR observations captured in vivid detail. The Caracol, Belize, project proved that airborne LiDAR observations could be used to rapidly map ancient archaeological ruins with previously unachievable spatial resolution, even in areas covered with dense tropical vegetation. Chase and his colleagues declared this the most significant new technology to be introduced to archaeology since radiocarbon dating. Together, the two technologies enable archaeologists not only to reliably date ancient settlements, but also to begin to understand the scale and spatial relations among them.

Mapping the Unseen

The prospect of finding a previously unknown archaeological site holds great allure for amateur archaeologists as well as their academic counterparts. In 2011 Steve Elkins, a professional videographer, proposed enlisting researchers at NCALM and the University of Houston to map an extensive area in the Mosquitia region of Honduras. Elkins wanted to search for archaeological ruins, but the project he suggested was very different from the one carried out in Belize.

For one thing, the rain forest of La Mosquitia was even denser than that of Belize; but a more significant challenge was the obscure nature of this quest. Whereas archaeologists had been exploring Caracol for decades and knew what they were looking for, the portion of La Mosquitia that Elkins wanted to explore had not been visited for centuries (to the best of his knowledge) by anyone, let alone archaeologists. No one knew whether there had once been cities in the area-and if so, who had built them, when, and using what materials. It might very well turn out that nothing would be found. The project carried a sizable financial risk for Elkins and his colleagues, as well as the possibility of an embarrassing failure for the UH-NCALM researchers. If, however, it succeeded in locating sites that had previously been impossible to find, this project could prove transformative to archaeological exploration in tropical regions around the world. The opportunity was too good to pass up.

Elkins selected four areas to be mapped, and in early May 2012 the UH-NCALM field team began collecting and processing the airborne LiDAR observations covering three of the areas. The first archaeological features (ruins of buildings, a plaza, an earthen pyramid, and agricultural terraces) were identified in an image just a few days later. As the observations continued to flow in, it became clear that there had once been two sizable settlements in the region as well as a scattering of more isolated structures that were most likely singlefamily dwellings. Without the maps created from the airborne LiDAR observations, ground parties could have spent decades exploring the region, perhaps stumbling upon the ruins of isolated buildings or artifacts such as pottery, but they still would have had little idea of the extent and contents of the site.

In February 2015 an exploratory party of more than 30 archaeologists, geodesists, videographers, former members of the British Special Air Service (SAS) forces, and Honduras military personnel was flown by helicopter into the area. The Honduran contingent were assigned to provide security in case of a chance encounter with criminals engaged in narcotics trafficking and to protect any artifacts that might be discovered (they are still guarding the area to this day). The ex-SAS officers, were there to prepare landing and camp sites and to share their jungle survival skills with the scientists and filming crew.

The primary goal of this expedition was to locate and verify a number of the features displayed in the LiDAR images, including the large plaza surrounded by public buildings, and the adjoining earthen pyramid. The team had hand-held units containing highresolution maps and images derived from the LiDAR observations, and GPS receivers, which enabled them to make their way directly to selected features-literally, close enough to reach out and touch them. At the base of the earthen pyramid, immediately, next to the plaza, the team discovered a cache of more than 50 stone carvings, partially buried in the ground. The artifacts, most likely religious offerings that appeared to have been undisturbed in recent times, were left untouched, but the exposed portions were photographed and then imaged with a ground-based LiDAR unit. These observations document the exact locations of each of the artifacts, to the millimeter level, as they were discovered. Later scans could be made to detect and document any changes or missing objects.

Remote, But Not Risk-Free

Archaeological surveying by air sounds almost sanitary when compared to prospecting through the rain forest, but even LiDAR is not without its risks. One inherent risk is that the region offers no safe place to land a plane in the event of engine failure; for this reason, it’s advisable to use a twinengine aircraft. Piloting the aircraft is tedious and demanding. Passing back and forth over a limited area for hours at a time, trying to stay on a preplanned route while minimizing roll, pitch, yaw, and changes in altitude, could cause even the most experienced pilot to make a dangerous mental error. The first beach mapping project conducted by University of Florida researchers ended prematurely when the pilot mistakenly retracted the landing gear just before touching down to refuel. No one was injured, but both the aircraft and the LiDAR unit suffered extensive damage.

At ground level, the rain forests remain dangerous today. Members of the Honduras exploratory party quickly learned that the animals in the area had little fear of humans: Visitors to their camp included a tapir, a wild boar, and several fer-de-lance snakes (one of which, lurking under a hammock used for sleeping, fortunately was discovered and killed before it attacked anyone). It was impossible, however, to avoid being bitten by insects, and one of every two members of the exploration party contracted leishmaniasis, a parasitic disease carried by insects that proved resistant to conventional treatment back in the United States. Under the care of doctors at the National Institutes of Health, these patients are responding well to treatment.

The Honduras project proved the transformative nature of airborne LiDAR for prospective archaeology: In a matter of days it can provide detailed information about the extent, content, and connections among archaeological ruins scattered over hundreds of square kilometers, even in the world’s most dense rain forests. And airborne LiDAR technology is still evolving rapidly: In 2014 the University of Houston funded the development of a next-generation airborne LiDAR unit that transmits 300,000 pulses per second at each of three different wavelengths, for a total of 900,000 pulses per second. The instrument was immediately sent to Antarctica to map dry valleys. The observations collected will be compared to observations made by NASA in 2001-2002 to detect changes in the terrain and glaciers in the dry valleys. Of particular interest is the detection of small surface features (thermokarsts) caused by the melting of subsurface ice due to the general warming trend.

On its return from Antarctica, the new LiDAR sensor was used to map archaeological ruins in Guatemala. By using multiple lasers of different colors, this new unit not only increases the point density of its coverage but opens the possibility of mapping the bottoms of shallow lakes, streams, and coastal waters, as well as identifying differences in the materials and texture of reflecting surfaces. At the same time the applications of airborne LiDAR are growing ever more diverse, from research designed to test geomorphic transport laws, map channel incision, explain hydrologic processes, define vegetation patterns and explore their controls, document snow accumulation, quantify floodplain sedimentation, document and discover active faulting, map landslides, predict fire hazards, characterize impact craters, quantify lava flow mechanics, document sediment transport and erosion, monitor beach erosion, and map archaeological ruins. With the publication during the past decade of hundreds of papers containing new scientific findings derived from the analysis of airborne LiDAR observations collected by NCALM, the prospects for LiDAR continue to expand.