Sunday, February 4, 2018

Why I am skeptical about the new Maya LiDAR results from NGS


New Tikal LiDAR map
I am skeptical about the hype surrounding the recent press release from the National Geographic Society about the new findings of LiDAR survey in the Maya region of northern Guatemala. I have no reason to question the quality of the LiDAR survey, or its potential usefulness for understanding aspects of ancient Maya society in this region. Rather, I question two aspects of the way these new findings have been portrayed, both in the NGS press release and in the journalism that has resulted from the find. (1) This is portrayed as revealing brand-new ideas, when in fact earlier LiDAR work had very similar results; and (2) The work is portrayed as a major scientific discovery, when in fact it is only the first step of a process, the end result of which will be (one hopes) some major scientific discoveries.


LiDAR is a relatively new airborne remote sensing technology that permits detailed mapping of the surface of the earth at a detailed scale. It is far superior to earlier forms of satellite or airplane mapping in that LiDAR can penetrate dense vegetation. It is ideal for the Maya lowlands, where the jungle vegetation hinders traditional mapping. Wherever it has been applied, in the Maya area, the result is the identification of many new houses and features of the built environment. (1) This is my first misgiving: the lack of acknowledgement that Mayanists have been working with LiDAR for more than seven years (Chase et al. 2012; Chase et al. 2011).

The NGS story has breathless quotes about how suddenly we know about many new features and structures on the Maya landscape. Well, that is what LiDAR does in the Maya lowlands. It finds many more features than archeologists knew about previously. If archaeologists are surprised about this, they just haven’t looked at the prior work, both in the Maya area (Brown et al. 2016; Chase et al. 2014b; Chase et al. 2014a; Chase et al. 2012; Chase et al. 2011; Chase et al. 2016; Chase 2016; Chase and Weishampel 2016; Ebert et al. 2016; Prufer et al. 2015; Von Schwerin et al. 2016; Yaeger et al. 2016), in other parts of Mesoamerica (Fisher and Leisz 2013; Rosenswig et al. 2015; Rosenswig et al. 2013), and particularly at Angkor in Cambodia (Evans et al. 2013; Hanus and Evans 2016).
 
Lidar-identified small reservoirs at Cacacol. Chase 2016.
One difficulty with LiDAR data is that while it is easy to see large structures like pyramids in the output data, small features such as houses or agricultural fields are more difficult to pick out. They often require a combination of intensive, time-consuming searching by eye, and sophisticated custom computer algorithms that can pinpoint such features automatically. For example, my student, Adrian Chase, analyzed LiDAR data to identify small residential-level reservoirs at the Maya city of Caracol (Chase 2016). In areas that had been mapped previously by traditional methods, Adrian’s algorithm identified 25 times the number of small reservoirs at the site! These did not stand out on the LiDAR landscape like dropped pins in Google-Maps. They had to be painstakingly identified.

As far as I can tell, the intensive phase of analysis has not yet been carried out (or is not reported in this press release). It is easy to use LiDAR to find a bunch of new features and make a pretty map. But the next two steps are more difficult. For the first step, the archaeologist has to analyze the data—staring at maps and applying algorithms—so that one can be confident that most of the relevant small features have been identified. The pretty color maps one sees in all the press accounts are not the only way to portray spatial data in LiDAR; often other visualization methods are more useful. Adrian was able to identify all those small reservoirs only because he did two things: he spent countless hours staring at the output, and he applied custom computer algorithms to the data to identify the features. There is no indication that archaeologists have carried out this intensive level of analysis of the new Guatemalan data.
 
3 LiDAR visualizations. Chase 2016
A second crucial step is to analyze the results quantitatively and spatially to construct population estimates and study the on-the-ground patterning in settlement data. The NGS article subtitle says there were “millions more people than previously thought.” The report has this quote:

“Most people had been comfortable with population estimates of around 5 million,” said [Francisco] Estrada-Belli, who directs a multi-disciplinary archaeological project at Holmul, Guatemala. “With this new data it’s no longer unreasonable to think that there were 10 to 15 million people there—including many living in low-lying, swampy areas that many of us had thought uninhabitable.”

It will take quite a bit of analysis to turn this quick preliminary suggestion into rigorous population estimates for settlements and regions. These additional steps—technical application of algorithms, lots of staring at screens, and then quantification and calculation—are only beginning for the Maya lowlands (Chase 2016; Chase and Weishampel 2016; Ebert et al. 2016), and there is no sign that they have been accomplished for the new Guatemalan LiDAR results.
 
LiDAR of central Caracol. Chase et al 2011.
So, what is my beef? The new results are just in, and the analysis is probably only starting. This is the normal process of science. (2) My second misgiving is the idea—promoted by NGS, by the people interviewed in the article, and by secondary articles in the media—that archaeological advances consist of discoveries in the field. Yes, the fieldwork is essential. But without an often lengthy period of analysis, one typically cannot know the meaning or importance of the finds.

There is a kind of archaeology where the main discovery is made in the field. If one is looking for the tomb of a king or noble, and one finds it, that may be the essential defining moment of discovery. But I pursue another kind of archaeology. I have spent my career on the archaeology of Aztec provincial households. When I dig up another house or trash midden, it seems pretty much the same as countless I and others have excavated. They are pretty boring, I have to admit. But once I have spent months or years studying the artifacts, quantifying them, sending off samples of technical analyses, only then do I make my discoveries. When I argue that this household was well-off and that one was poor, or when I argue that conquest by the Aztec empire had little effect on local people, these are my discoveries. They rely on extensive analyses of artifacts. I had no idea about these things at the time of excavation. I discuss this issue—what is the real moment of discovery?—in more detail in my recent book (Smith 2016).
 
Social interpretations at Yautepec were based on study of 1 million potsherds
When one focuses almost exclusively on the actual uncovering of a find during fieldwork (for an excavation), or on the initial pretty maps of a LiDAR surveybefore the hard work of analysis is done—one is distorting the scientific significance of our work. Will NGS have a big feature when the archaeologists involved actually publish a revised population estimate for northern Guatemala, or when they can quantify the amount of construction in rural vs urban areas?
We'll see.

A kind of archaeology based on extensive analysis
How can one spot a finding that seems spectacular but is actually a preliminary find, not yet analyzed, from a finding based on proper analysis and interpretation? Peer-review publication is the primary way to do this. The NGS piece was based entirely on interviews, not on a paper that has been peer-reviewed and accepted by a scientific journal.

Claims that LiDAR will revolutionize the study of Maya settlement and demography may very well be correct, but it is too soon to tell. The Guatemalan LiDAR has reached the stage of preliminary findings and pretty maps, but not the stage of solid architectural, demographic, and social findings. I look forward to the scientific results. I don’t care if they are an internet sensation; I’d rather see them published in a journal.

References


Brown, M. Kathryn, Jason Yaeger, and Bernadette Cap
2016 A Tale of Two Cities; LiDAR Survey and New Discoveries at Xunantunich. Research Reports in Belizean Archaeology 13: 51-60.

Chase, Adrian S. Z.
2016 Beyond Elite Control: Residential Reservoirs at Caracol, Belize. WIREsWater 3 (6): 763-797.

Chase, Adrian S. Z. and John F. Weishampel
2016 Water Capture and Agricultural Terracing at Caracol, Belize as Revealed through Lidar and GIS. Advances in Archaeological Practice 4 (3): 357-370.

Chase, Arlen F., Diane Z. Chase, Jaime J Awe, John F. Weishampel, Gyles Iannone, Holley Moyes, Jason Yaeger, and M. Kathryn Brown
2014a        The Use of LiDAR in Understanding the Ancient Maya Landscape. Advances in Archaeological Practice 2 (3): 208-221.

Chase, Arlen F., Diane Z. Chase, Jaime J. Awe, John F. Weishampel, Gyles Iannone, Holly Moyes, Jason Yaeger, Kathryn Brown, Ramesh L. Shrestha, William E. Carter, and Juan Fernandez Diaz
2014b        Ancient Maya Regional Settlement and Inter-Site Analysis: The 2013 West-Central Belize LiDAR Survey. Remote Sensing 6: 8671-8695.

Chase, Arlen F., Diane Z. Chase, Christopher T. Fisher, Stephen J. Leisz, and John F. Weishampel
2012 Geospatial revolution and remote sensing LiDAR in Mesoamerican archaeology. Proceedings of the National Academy of Sciences 109: 12916-12921.

Chase, Arlen F., Diane Z. Chase, John F. Weishampel, Jason B. Drake, Ramesh L. Shrestha, K. Clint Slatton, Jaime J. Awe, and William E. Carter
2011 Airborne LiDAR, Archaeology, and the Ancient Maya Landscape at Caracol, Belize. Journal of Archaeological Science 37: 387-398.

Chase, Arlen F., Kathryn Reese-Taylor, Juan C. Fernandez-Diaz, and Diane Z. Chase
2016 Progression and Issues in the Mesoamerican Geospatial Revolution: An Introduction. Advances in Archaeological Practice 4 (3): 219-231.

Ebert, Claire E., Julie A. Hoggarth, and Jaime J. Awe
2016 Integrating Quantitative Lidar Analysis and Settlement Survey in the Belize River Valley. Advances in Archaeological Practice 4 (3): 284-300.

Evans, Damian H., Roland J. Fletcher, Christophe Pottier, Jean-Baptiste Chevance, Dominique Soutif, Boun Suy Tan, Sokrithy Im, Darith Ea, Tina Tin, Samnang Kim, Christopher Cromarty, Stéphane De Greef, Kasper Hanus, Pierre Bâty, Robert Kuszinger, Ichita Shimoda, and Glenn Boornazian
2013 Uncovering archaeological landscapes at Angkor using lidar. Proceedings of the National Academy of Sciences 110: 12595-12600.

Fisher, Christopher T. and Stephen J. Leisz
2013 New Perspectives on Purapécha Urbanism through the Use of LiDAR at the Stie of Angamuco, Mexico. In A Primer on Space Archaeology: In Observance of the 40th Anniversary of the World Heritage Convention, edited by D.C. Comer, pp. 191-202. SpringerB riefs in Archaeology, vol. 5. Springer, New Yokr.

Hanus, Kasper and Damian Evans
2016 Imaging the Waters of Angkor: A Method for SemiAutomated Pond Extraction from LiDAR Data. Archaeological Prospection 23 (2): 87-94.

Prufer, Keith M., Amy E. Thompson, and Douglas J. Kennett
2015 Evaluating airborne LiDAR for detecting settlements and modified landscapes in disturbed tropical environments at Uxbenká, Belize. Journal of Archaeological Science 57: 1-13.

Rosenswig, Robert M., Ricardo López-Torrijos, and Caroline E. Antonelli
2015 Lidar data and the Izapa polity: new results and methodological issues from tropical Mesoamerica. Archaeological and Anthropological Sciences 7 (4): 487-504.

Rosenswig, Robert M., Ricardo López-Torrijos, Caroline E. Antonelli, and Rebecca Mendelsohn
2013 LiDAR Mapping and Surface Survey of the Izapa State in the Tropical Piedmont. Journal of Archaeological Science 40: 1493-1507.

Smith, Michael E.
2016 At Home with the Aztecs: An Archaeologist Uncovers their Domestic Life. Routledge, New York.

Von Schwerin, Jennifer, Heather Richards-Rissetto, Fabio Remondino, Maria Grazia Spera, Michael Auer, Nicolas Billen, Lukas Loos, Laura Stelson, and Markus Reindel
2016 Airborne LiDAR acquisition, post-processing and accuracy-checking for a 3D WebGIS of Copan, Honduras. Journal of Archaeological Science: Reports 5: 85-104.

Yaeger, Jason, M Kathryn Brown, and Bernadette Cap
2016 Locating and dating sites using Lidar survey in a mosaic landscape in Western Belize. Advances in Archaeological Practice 4 (3): 339-356.



18 comments:

Unknown said...

It's a little known fact that Copan, not Caracol, was the first Maya site where LiDAR was applied. True, this application wasn't on as large of a scale as all subsequent LiDAR studies, but technically Copan was first.

Michael E. Smith said...

@Unknown - It would be nice to see a history of the methods as used in Mesoamerica.

Jon Lohse said...

Yes, it would. I think the Copan work was pretty experimental, but they recorded most of the main plaza.

Debra Walker said...

Don't forget Campeche and Yucatan. The Yaxnohcah data from 2014 has also been published, and Ivan Sprajc's team has developed some great algorithms that improved our understanding, but ground-truthing is an essential component. Also Travis Stanton's data from Yaxuna has proved interesting.

Michael E. Smith said...

But I am still waiting for new analyses of demography or settlement to come from the LiDAR data.

Tim Pugh said...

I am glad you wrote this. Is this press release an advertisement for a National Geographic program rather than purely a presentation of research? If so, are there any conflicts of interest?

Michael E. Smith said...

@Tim - Good question! It looks like they do have a TV show this week (Tuesday) that features the LiDAR study. So maybe this is really an advertisement, not a news report.

Anonymous said...

The way NGS is presenting the findings of this research is worrisome. The documentary that will be showcased in the forthcoming days is entitled "Lost Treasures of the Maya". The misleading emphasis on archaeological remains as "treasures" is causing enormous harm in Central American archaeology. For the great public the idea of archaeology as a hunt for lost treasures gets entrenched and fixated. In the long term, the looting of sites and the illicit traffic of archaeological objects is a byproduct of this narrative of commodification of Maya archaeological sites and artifacts.

Miguel Covarrubias said...

You missed Hutson's mapping at Uci Cansahcab sacbe

Michael E. Smith said...

I didn't intend to provide a complete list of all Maya lidar studies. That should be done by someone who works on this topic. I merely grabbed a bunch of citations from my Endnote file.

Tim Pugh said...

From what I now hear, there will be a number of real publications on this data so there is some real news here. I look forward to reading these. I agree with anonymous (above), I think the Nat Geo "stuff" is doing a disservice and it also makes me worry about how such press or even funding impacts archaeological research. If we become entertainment, rather than research, archaeology would suffer. We need good documentaries, not attempts as "reality" programming that distort rather than report. Are there no producers that can make a good documentary that the public can appreciate?

Unknown said...

Here are a couple more just for fun
Fisher, Fernández-Diaz JC, Cohen AS, Neil Cruz O, Gonzáles AM, Leisz SJ, et al. (2016) Identifying Ancient Settlement Patterns through LiDAR in the Mosquitia Region of Honduras. PLoS ONE 11(8): e0159890. doi:10.1371/journal.pone.0159890

Fisher, A. Cohen, J. Fernandez-Diaz (2017) The Application of Airborne Mapping LiDAR for the Documentation of Ancient Cities and Regions. Quaternary International Volume 448, 20 August 2017, Pages 129-138. doi.org/10.1016/j.quaint.2016.08.050

Chris Fisher said...

Hey Mike, the unknown above was me, Chris Fisher, I can't seem to log under my name, did not mean to be anonymous - was just adding a couple more LiDAR citations.
Chris

Michael E. Smith said...

Hi, Chris (or should I call you "unknown"?). Thanks for the cites.

JCFD (Lidardealer) said...

A great blog post, for sure in science we have to approach every find/announcement with a mix of skepticism and open mind. Just want to provide 2 technical cents into the discussion. By my estimates there have been about 30 lidar survey projects in Mesoamerica, both reported and unreported. There have been papers on both the potential and limitations of the technique. While it is true that what has been tested so far indicates that lidar detection rates are high for large features a not so good for smaller features like domestic architecture and small agricultural/irrigation features it is important to note that 1) results from one survey/site should not be extrapolated elsewhere and 2) the lidar technology is quickly improving.

While we use the word lidar as a generic term, lidar data is not a generic uniform product and not all lidar data is created equal, even the metric we used to compare them (pulse or return density) may be misleading. How useful it is to detect a particular type of cultural feature it is a function of both the equipment (and how it is used) and perhaps most important the environment… Coastal Yucatan with extremely short vegetation is probably the worst environment… the border between Petén Guatemala and Campeche Mexico has proven to be the environment (tall tree, relatively little underbrush) that yields best results.

On the other hand the equipment has improved significantly in an obscure characteristic that we don’t talk that often called range resolution (Basically the ability to separate the ground from the vegetation above in the lidar signal). Just to give rough numbers, the first “experimental” survey of Copan in 2000 used a first generation of lidar sensor (only recorded first and last return) and had a range resolution of about 5 meters, for the Caracol survey in 2009 (and later in 2013) a second generation sensor was used with a range resolution of about 3 meters (recorded 4 returns although for archaeology the last is the only that matter). For this Pacunam Lidar Initiative (PLI) survey a third generation sensor was used that has a range resolution of about 1.5 meters. Besides the improvement in range resolution, for a comparable pulse density of a 2nd generation sensor, the 3rd gen sensor used in PLI illuminates more surface area and at 3 different look angles, so technically it is expected to be able to show more of those small elusive features after advanced image processing techniques have been applied. So I guess we will have to wait until more details come out of the science team to see how results really compare to previous ones.

Finally, I just want to point out that, for me the most important answer is not what we learn for a particular survey, call it Caracol 2009&2013, Copan 2015, Teotihuacan 2015, Tres Zapotes (2014) etc… the most important answer is what we learn when we put all of these together and what we learn when we incorporate the additional knowledge that comes from field work and community participation. And this is why blogposts that inspire this kind of discussion are great.

Dick Diehl said...

I am enjoying this discussion and could comment at length but will refrain. Groung_truthing, analsys, publication, responsible journalism and documentary making: all badly needed. See the BBC 4 series Time Team for exciting, professional public archaeology.

Michael E. Smith said...

@JCFD - Thanks for your comments. This is a relatively new application for archaeology, with lots of complexities. But the results can be spectacular!

Michael E. Smith said...

@Dick - Yes, indeed!