Beneath Angkor's foliage

By Damian Evans

Using a radical new imaging technique, Sydney archaeologists have uncovered a network of unknown temples and an entire city beneath the foliage surrounding Angkor in Cambodia.
Image of Angkor Wat

Credit: Michael Coe

For more than one and a half centuries, explorers and scientists alike have relied on the machete to clear the dense vegetation that obscures much of the remains of the great medieval civilisation of Angkor, which flourished across mainland Southeast Asia from the ninth to 15th centuries AD. Until the past year, that is.

In July 2012 Roland Fletcher, a Professor in the Department of Archaeology, and I sat in an air-conditioned office at the University’s Robert Christie Research Centre in Cambodia, having just taken delivery of the results of a landmark aerial survey, which deployed an airborne laser scanner strapped to the side of a helicopter. The whole project had been something of a gamble: several years of planning and a quarter of a million dollars invested in a technology that had never been used for archaeology anywhere in Asia.

The laser scanner, or ‘lidar’, had the potential to see through a dense forest canopy and reveal traces of the civilisation remaining on the forest floor, but no-one was entirely sure that it would work, or that there would be anything much to see even if it did. As a map derived from four billion laser measurements of Angkor slowly unfolded on the screen of a high-performance computer, however, we watched in awe as entire cities were revealed for the first time beneath the jungle of northwest Cambodia, and realised that the age of the machete had just drawn to a decisive close.

We are currently fundraising for a second mission to take a first look at a couple of other temple complexes in the region where, it is suspected, entire cities also lie undiscovered.

The great monuments of Angkor, along with their inscriptions and artworks, have long fascinated scholars, and over the past century a huge body of work has been produced on these aspects of Khmer civilisation, mostly by French scholars. In the 1990s, as decades of conflict in Cambodia finally wound to an end, the arrival of modern archaeological methods sparked a renewed interest in Angkor, but with an entirely different focus.

For the first time, scholars began seriously to address broader questions about the context of the temples: Who exactly were the people who built them, where did they live, how were they so successful in the unforgiving environment of monsoon Asia, and perhaps most importantly, what happened to them?

The problem for archaeologists had always been that houses of stone were reserved for the gods, and that the stuff of everyday life was mostly non-durable material – houses made of thatch, and even royal palaces made of wood – that has long since disappeared. However, in 1992 a French archaeologist, Professor Christophe Pottier, noted that traces of the remains of ponds, occupation mounds, village shrines, roadways and canals could still be discerned from the air. It was the fabric of the urban and agricultural network of greater Angkor.

Image of Sydney researchers at Angkor Wat

Professor Roland Fletcher (left) and Postdoctoral Research Associate Damian Evans at Angkor in Cambodia. Credit: Michael Coe

Pottier began to use aerial photos to sketch those traces by hand onto paper maps from the 1950s, setting in motion a process that would lead eventually to the lidar survey two decades later. A meeting between Pottier and Fletcher in 1998 laid the groundwork for the University of Sydney’s involvement in Angkor. The technical skills of the Archaeological Computing Laboratory (now Arts eResearch) were brought to bear on Pottier’s maps, which were converted into an enormous digital mapping database and, as Pottier jokes, “allowed me to enter honourably into the 21st century”.

For the next decade and a half, professors Fletcher, Pottier and I continued the mapping work, eventually expanding the inventory of temples in the Greater Angkor area from around 350 to 1250, and discovering that Angkor was actually the largest integrated settlement complex of the pre-industrial world, a sprawling conurbation comparable in size to modern-day Sydney.

The discovery was remarkable, but there were no ‘eureka moments’, just a gradual accumulation of more and more pieces of the puzzle over the course of many years of painstaking analysis and difficult fieldwork. Most troubling of all was the fact that vast swathes of the new archaeological map consisted of nothing more than white space, where forest cover prevented conventional technologies like satellites and airborne radar from identifying any traces of the civilisation that might remain etched into the surface of the landscape. Question marks loomed over issues like the completeness of the work, and the accuracy of the conclusions drawn.

Changing the game

The very few archaeologists in the world who specialise in remote sensing have long been looking forward to the ‘next big thing’ on the horizon, as new technologies, often developed initially for military applications, slowly transformed into relatively inexpensive commercial applications.

In the 1990s it was radar, in which the University was a world leader, using it for archaeological research (my honours and PhD theses at Sydney both revolved around using it at Angkor); in the late 1990s and early 2000s it was very high-resolution conventional imagery, of the kind we commonly see these days in Google Earth; in the last few years, it has been lidar (see accompanying article).

In 2005, Roland and I began to hatch a plan to use lidar’s capacity to ‘see through’ the forest to fill in those blank spaces on the map. The problem was that no-one had ever used lidar for archaeology anywhere in Asia before. It was an untested technology in this specific context, and many researchers working in Cambodia were sceptical that the instrument could see through the dense vegetation that surrounds the temples and deliver worthwhile results.

No-one was prepared to commit the six-figure sum the mission required. Encouraged by a successful application of the technology in a similar context at a Mayan site in Belize, however, and given momentum by seed money from National Geographic, we spent years essentially going door-to-door seeking the participation of the various other international teams working at Angkor.

Image of Angkor Wat forest cover

Angkor Wat forest cover; above Angkor Wat minus forest cover

Eventually, by 2011, enough small contributions had been raised to make the project viable. Convincing the Cambodian authorities of the merits of the idea was a different matter entirely: the permissions process took six months, involved several different ministries, required an unprecedented exemption from the no-fly zone above Angkor Wat, and went all the way up to prime ministerial level.

In the end, eight different teams representing seven different countries committed support, in what would turn out to be the broadest research cooperation ever achieved in Cambodian archaeology, and the largest archaeological lidar acquisition ever undertaken anywhere in the world.

In April 2012 the Indonesian branch of a Canadian survey company, PT McElhanney, was contracted to import the necessary equipment and undertake the lidar survey. A red helicopter spent nearly a week systematically crisscrossing the airspace just 800 metres over the World Heritage Site of Angkor, as well as two other remote and forested locations. The vast amount of data produced had to be hand-carried to Cambodia from McElhanney’s labs in Jakarta in July.

Once the complex process of interpreting the data was completed, the excitement of that first day has turned out to be entirely justified. The imagery reveals that archaeological features are almost ubiquitous beneath the forest cover, and that the ceremonial centre of Angkor, with its great temples shrouded in jungle, was surrounded by a formally planned, grid-like network of roads. A French team had spent many years surveying one part of that ‘downtown’ area on the ground, using machetes and hand-operated survey levels; the lidar mission covered that same area in 45 minutes of flying time, and produced a map with greater precision and accuracy.

At least half a dozen previously undocumented temples have been uncovered in the immediate vicinity of Angkor Wat, which more than two million tourists visit every year, along with a previously unknown urban layout within the very confines of the temple’s moat. In the remote Kulen mountains to the north of Angkor, where dense forest and extensive mine fields have traditionally frustrated mapping efforts, an entire urban layout has emerged from beneath the vegetation. It corresponds to a previously undiscovered city referred to in thousand-year-old inscriptions as Mahendraparvata.

The results are “a total game-changer” for archaeology in the region. Just recently, a group of prominent Mayan scholars likened the emergence of lidar to the advent of carbon dating.

The newly discovered cities clearly extend beyond the limited lidar coverage that has so far been achieved, and we are currently fundraising for a second lidar mission to extend that coverage and take a first look at a couple of other temple complexes in the region where, we suspect, entire cities also lie undiscovered on the forest floor.

Our job will be significantly easier this time around: thanks to the first results, the initial scepticism surrounding the method has largely evaporated, and has now given way to enthusiasm. The results, according to Professor Fletcher, are “a total game-changer” for archaeology in the region.

Just recently, a group of prominent Mayan scholars published an article in which they likened the emergence of lidar to the advent of carbon dating in terms of its importance for the study of tropical forest civilisations. They, too, plan additional, more extensive lidar acquisitions. A collaborative network of researchers interested in the archaeological applications of lidar is rapidly emerging, not only in Southeast Asia and Mesoamerica but in other regions as well, bound by a common interest in the ability of lidar to provide extraordinary insights into interactions between humans and their environment in the distant past.

There are also implications for site preservation and heritage management, as illegal logging, corporate land concessions and rapid urban expansion threaten the delicate traces of the Angkorian civilisation that remain inscribed into the landscape around the temples. These are urgent issues, and UNESCO is at the forefront of efforts to extend the lidar coverage in Cambodia in order to fully understand the spatial extent of the remnant cultural heritage.

This dry season, staff and volunteers from University of Sydney teams have been working in the field with Cambodian colleagues to confirm and document new sites so that they can be protected. The thrill of discovery is only partially offset by the difficulty of the task, which often involves trekking deep into the dense forests of northwest Cambodia. As it turns out, the machete is not quite redundant around here just yet.

Damian Evans is a Postdoctoral Research Associate in Professor Fletcher’s Angkor Research Program.


Lidar - The radar with laser

Airborne laser scanning revolves around a technology called light detection and ranging (Lidar), which is analogous to the more familiar technology of radar. Instead of radio waves, however, a Lidar instrument sends out a laser pulse, which is then reflected back to the instrument by any object or surface that the laser beam encounters.

Koh Ker 3D view of forest lifted off landscape

Koh Ker 3D view of forest lifted off landscape. Credi: Michael Coe

The instrument measures the time that it took for the reflection to return, which it uses to calculate the distance between the instrument and the reflective surface: the longer the time, the more distant the surface is. Given enough of these measurements taken looking down from an aircraft, incredibly detailed three-dimensional models of the landscape can be created.

The Lidar used in the Angkor mission emitted about one million laser pulses every four seconds, acquiring an enormous amount of data in a 600-metre-wide swathe along the flight path of the helicopter. The laser light does not actually penetrate or ‘see through’ vegetation. For archaeological applications, the idea is to bombard every square metre of the landscape with so many laser beams that at least a handful of the laser pulses are likely to find tiny gaps in the foliage, hit the ground, and reflect back to the sensor.