[UPDATE: Please check the comments for Damian Evans’ friendly critiques of my comments]
I first visited Angkor in 2005 as a tourist (following a couple months of survey in southern Cambodia). If you’ve visited Angkor perhaps you, like me, stared out from your tuk-tuk as you breezed pass the thick trees in Angkor Thom and wondered “I wonder what is going on in there?” You might also have recognized that at its height, Angkor and its many temples would not be the vacant ceremonial center that it appears today, but a bustling city and so you might have also wondered “Where were all these people living?”
Our 2015 excavation of a house mound within the Angkor Wat temple enclosure.
I did not know then that a French archaeologist, Jacque Gaucher, had spent the better part of 20 years hacking through the thick forests of Angkor Thom, mapping the mounds and features there and identifying the city and its ancient grid system. He has produced a detailed map of that city, but the decades it took to put together meant that to create similar maps of the rest of Angkor would take several lifetimes.
Walking around the forested parts of Angkor, you can tell that these areas are not empty spaces. You can spot ceramics on the surface and you can see the undulations on the ground that indicate the presence of mounds and depressions (some of which may be ponds).
In 2010 (before we had LIDAR data) my Greater Angkor Project colleagues tried to map some of the mounds and depressions we saw within the Angkor Wat enclosure. You can see from the top image below (taken from our recent Antiquity publication) that my colleagues had figured out a rough pattern. However, due to the dense tree cover it was extremely difficult to see the broader grid system within the enclosure. Nor were we aware that this orthogonally organized mound-depression pattern extended outside the eastern moat of Angkor Wat, to an area we have referred to as the External Eastern Enclosure (it is bounded by a rectangular earthwork).
From here.
This, my friends, is the magic of LIDAR.* What would have taken many people many years of survey work (ground and aerial) and archaeological mapping can instead be achieved in just a few minutes of survey with LIDAR. And with this we can clearly see the detailed landscape modifications surrounding temples, connecting urban spaces and waterways, and even buried sandstone quarries. We can see how much the Angkorians (and we are learning, the pre-Angkorians) extensively planned and managed their landscapes.
However, LIDAR is not the culmination of our research on Angkor but a new beginning point. With these data we now have clearer picture of, for example, where people are living. (This in particular has been the inspiration for my research to more intensively excavate and investigate house mounds). The LIDAR data has provided a detailed map and with this, new questions arise and has the potential to launch many new research projects. It is a tool to help is focus on more specific and more productive research questions, but it is not the answer in itself.
In fact, there is quite a lot that Lidar cannot do. For example:
-Lidar cannot tell us that everything we find is an archaeological site. For this, we need to – boots on the ground – visit and verify that the site is archaeological. This is called ground-truthing. This is part of some of the ongoing work by my colleagues associated with the Cambodian Archaeological Lidar Initiative.
-Lidar cannot tell us what all the features are. There are quite a few enigmatic sites (geometric shapes that may be gardens and the “mound fields” to name two), which must be investigated with on the ground archaeological work in order to determine their nature and function.
-Lidar cannot tell us how old a feature is. For this we need to go out and visit a site specifically – perhaps even undertaking an excavation in order to get datable material. Although some materials on the surface (diagnostic ceramics, for example) can give us an idea of how old something is.
-Lidar cannot necessarily tell us how these features relate to one another. Lidar picks up modifications to the landscape, but because many of these features may have been constructed at different times, we need archaeological research to understand how they are connected (or not) to one another.
-Lidar cannot tell us what is under the ground. For this, we need archaeological investigations or other remote sensing techniques like ground-penetrating radar.
The Lidar data is magic, but knowledge is produced when combining this with scientific, archaeological research.
In the next post, I’ll discuss some of the finds in the forthcoming Lidar publication in a bit more detail.
*I was not directly involved in this project, but have benefitted from the results and work with several members of CALI.
BarCamp Phnom Penh
Alison in Cambodia 
Hi Alison, thanks for this interesting and thought-provoking article. A nice counterpoint to the festival of stupidity that’s appeared in most of the mainstream media over the last couple of days in relation to these finds. I’ve been thinking for a while about some of the issues you raise, and respectfully I’d like to quibble with you on a few points:
“Lidar cannot tell us that everything we find is an archaeological site. For this, we need to – boots on the ground – visit and verify that the site is archaeological. This is called ground-truthing.”
Let me play devil’s advocate here and suggest that ground verification of features identified in remote sensing data can in many cases be highly over-rated, and often reflects the discipline’s traditional preoccupation with ground-based methods rather than a sensible strategy for follow-up from a remote sensing campaign. This is especially true for features identified in lidar imagery, as the instrument is often picking up highly structured and patterned topography that is barely visible on the ground, even more so when covered with vegetation.
Having a finite amount of resources, and being safety-conscious, I honestly question the value of paying a team of people to trudge through a forest full of landmines, snakes and people with guns for several hours to arrive at some spot and conclude “Oh look, a lump on the ground, exactly like in the lidar”. And then slog it all the way back home again. The lidar is mostly picking up sites that are unlooted and uncultivated (the topographic variation is so subtle that even looters, who are much more skilled at finding archaeological sites than archaeologists are, haven’t recognised them as sites) so the ground is undisturbed and there’s no ceramics or anything else to see on the surface.
The whole thing therefore becomes a kind of pointless exercise in confirming that the lidar is super-accurate, which we knew already. We are now covering such wide areas with lidar, as in thousands of sq km, and picking up so many sites that we have to consider whether comprehensive ground verification isn’t just wasting precious time and resources that could be devoted to targeted excavations that would be yielding results that are actually really useful.
“Lidar cannot tell us how old a feature is.”
I don’t know that this is entirely true either, and I’m always surprised to hear statements from archaeologists along the lines of “aerial imagery cannot tell us about chronology”. Imagine if this were strictly true: we wouldn’t even be able to tell prehistoric stuff from stuff that was built yesterday; an ancient road would be indistinguishable from an autobahn; and we would need to be excavating absolutely everything to find the age of it. So straight away we realise that there is something seriously wrong with this assertion.
Relative spatial relationships between features can tell us a lot about chronology, and things identified in the lidar often have very precise logical, functional and operational relationships with features that are very well-dated – and from which we can therefore infer dates for other stuff. For example, a canal identified in the lidar that carries water from the moat of a temple built in 1000 AD, goes along for a bit and is then destroyed by a temple built on top of it in 1010 AD tells us that the canal was almost certainly operational for that specific decade.
There is also in many (most?) cases a quite well-known and well-dated evolution in the spatial structure of things, which is why we know that a moated circular site dates from the Iron Age; that temple cities with nice formal square grids date from the early 12th century; that temple cities with messier grids date from the late 12th century onward, and so on. Sometimes the dating of the evolution of landscape structures can be very precise indeed: we know exactly what the spatial structure of Khmer Rouge field systems looks like, for example, which narrows it down to a 4 year period, and is very distinct from both post-KR and ancient field systems. In fact, dating things from above is the very foundation of the discipline: military aviators in WWI a century ago noticed patterns on the ground that could only be seen from above, and which very clearly dated to specific stages of antiquity (Roman sites, barrows, Iron Age fortifications, tells, etc) and out of their continued curiosity after the war aerial archaeology was born.
“Lidar cannot tell us what is under the ground.”
This too is highly debatable. Cultural deposits in Cambodia are so shallow, and structures are therefore buried so close to the surface, that things under the ground are often represented in patterning of the dirt that covers them. This is especially true when the buried features have a very distinctive spatial patterning. The quarry fields at Beng Mealea in the paper are a good example of this, but we have also identified buried Buddhist Terraces near Angkor Thom based on the distinctive topographic relief they’ve produced on the forest floor. This is actually true of pretty much all kinds of aerial imagery and is again a kind of fundamental principle of the discipline of aerial archaeology: people have identified buried Roman villas, for example, by looking at the patterns in crop marks alone.
“[Lidar] is a tool to help is focus on more specific and more productive research questions, but it is not the answer in itself… knowledge is produced when combining this with scientific, archaeological research.”
Perhaps, but it’s important to recognise that analytical work on the imagery is an equally important form of knowledge production than ground work. Sometimes the production of very important knowledge using lidar imagery is astonishingly simple: if you have a research question like “I want to know how big this medieval city under the jungle used to be, because the extent to which they deforested the landscape to build it is important for reason X” then you can have an important research outcome in 5 minutes of simple visual analysis.
To take another example, we’re currently working with engineers on complex modelling of hydrological, climatological, structural, volumetric and topographical data to work out why the main dam at the 10th century Angkorian capital of Koh Ker failed. It’s work that (aside from the environmental data) almost entirely derives from and revolves around desk-based analysis of the lidar, and is likely to tell us much more about the trajectory of growth and decline of the city than ground work ever would.
I know this is an archaeological heresy, but think it’s both possible and quite common to overstate the usefulness of ground work in archaeology, and of excavation in particular, which seems more often than not to produce more questions than answers, or highly ambiguous data from stratigraphies and so on that can be interpreted in a multitude of different ways (usually in the way that conforms nicely to the archaeologist’s preconceived notions). And yet excavation remains the gold standard for archaeology, as you point out. Apropos of which:
“There are quite a few enigmatic sites (geometric shapes that may be gardens and the “mound fields” to name two), which must be investigated with on the ground archaeological work in order to determine their nature and function.”
These features have been the subject of intensive ground research including multiple excavations over many years by several different archaeological teams, and we are still no closer to understanding what they are than when we first saw them in the lidar. Just saying 😉
Hi Damian – Thanks for your response. I think your critiques of my assessments are quite well-founded and definitely show my bias towards excavation. As a “dirt” archaeologist, I will say perhaps we’re in disagreement there about the utility and effectiveness of the method. Perhaps we should just agree that both approaches have their uses and are quite often complementary. 🙂 I will also say that with Rachna, I’ve been working on those square spirals and if we answer them it will be through data gathered in excavation. I have pollen/phytolith samples that are getting processed (should clarify if things were growing nearby) and a geoarchaeologist working on site formation processes. This is a drawback to dirt-based archaeology is the TIME (and $) all this analysis takes.
I suppose my point re: LIDAR is not a dating tool is that I’ve gotten a few questions like “Can LIDAR tell us how old those mounds in Angkor Wat are?” In that case, I think people wonder if they technique can help with direct dating. I should have clarified (as you eloquently did) that by considering these features as part of a broader landscape you can relatively date them. My impression is that people wonder if you can get absolute dates, which you can’t, as LIDAR is not an absolute dating technique.
Thanks for correcting my misconceptions!
Never mind, I’m just stirring the pot a little. I should clarify also that I’m talking about excavations on the Angkor period in particular, in relation to urban and agricultural elements, where messy stratigraphies and vague ceramic sequences seem to leave an awful lot of room for doubt. One thing I notice is that the ratio of peer-reviewed publications to such kinds of excavations is wildly disproportionate: there’s been a terrific amount of digging, and hardly any writing. I can think of excavations completed decades ago at Angkor, by scholars still active in the field, that have never seen the light of day. People who dig prehistoric burials on the other hand, or temple sites, seem to be pretty diligent about publishing.
Also I take your point about relative vs. absolute dating, but again I’m not sure that the distinction is quite as neat as many of our colleagues would have us believe. To take the example of the occupation mounds that you’ve been digging at Angkor Wat: dating some organic material from an excavation as a proxy for inferring the date of a feature is supposed to be the gold standard, but you will get a date back that will look something like, let’s say, 1100 AD plus or minus 20 years. From the lidar on the other hand we can use the dates of the temples as a proxy for inferring that the feature dates precisely from the reign of Suryavarman II, considering that it is a component of an extremely distinctive patterning of space that appears universally, and exclusively, in the temples of his reign. Of course, excavation has the potential to add loads more detail in relation to other stuff, but – contrary to prevailing wisdom in archaeology – in terms of dates I would much prefer “relative” dating from remote sensing over “absolute” dating methods in a wide range of circumstances, for example in my canal analogy and the KR field wall example above.
Ah, that is because you have yet to see my BEAUTIFUL sequence of C14 dates from AWT (2010/2013/2015 excavations) which adds some nuance to the use of the enclosure space. Too early to talk about it on the blog, but will be writing it up for publication ASAP.
Haha, OK, I look forward to seeing it in print, and will be happy to be proven wrong. 🙂