Archaeologists dig into digital data

• Archaeology for the past 150 years has been a "hit and miss practice," says Elaine Sullivan, an Egyptologist at the University of California, Santa Cruz.
• "These new technologies are allowing us to think more carefully about where we do hand excavation and where we can learn without excavating at all."

What's happening: Light detecting and ranging (lidar) measurements, virtual reality, 3D modeling and other technologies have become powerful tools for archaeologists.

• Aircraft-based lidar — which emits a laser pulse and measures the return time of light reflected from an object or feature of a landscape to determine the distance to it — can penetrate vegetation and give a precise 3D image of the ground below.
• But it has also revealed massive features concealed by their sheer size, even in more open land.

In a recent study, archaeologists analyzed lidar data and found nearly 500 new Mesoamerican sites across a large swath of southern Mexico.

• "Many sites were hidden in plain sight," says archaeologist Takeshi Inomata of the University of Arizona.
• The structures were likely built between 1050 B.C. and 400 B.C., and features of their remnants suggest the Olmec and the Maya civilizations interacted, Inomata and his colleagues report in the journal Nature Human Behaviour.

The intrigue: Mayans built a large structure — Aguada Fénix — between 1,000 and 800 B.C., just as they were transitioning to a more sedentary way of life, but "we don't see evidence of very much developed social inequality," Inomata says.

• The site — and a handful of others — is part of emerging evidence that suggests the construction of monuments, cultural centers and other features of civilization might not depend on societies developing the hierarchy and inequality that was long seen as an inevitable consequence of development.

What's next: Where there's data, there's usually AI close behind, and archaeologists are adapting algorithms to analyze lidar and other data from remote sensing.

• Leila Character, a graduate student at the University of Texas, Austin, is training a deep learning algorithm for object detection on more than 17,000 features of structures Inomata mapped at one archaeological site. She plans to then run the algorithm on unlabeled data from other sites to try to then find new sites, with confirmation in the field to follow.
• "You have to ground truth check your results," says Character, who is also using AI to spot shipwrecks from sonar and lidar imagery. "This is not a replacement for archaeological fieldwork."
• Mapping a large archaeological site can take years without lidar data and days or weeks if the data is available but has to be analyzed by a person. By combining remote sensing and machine learning, it can be done in a day, Character says.

But, but, but: Lidar data is expensive to collect — Inomata's came from a survey done by the Mexican government — compared to other techniques, though less costly than traditional archaeological digs. Some governments also don't allow planes to be flown over sites.

• And in these early days of AI archaeological analysis, there aren't standard operating procedures for how people are labeling data or what models are being used, Character says.

Artifacts are also getting the digital treatment.

• Elizabeth Thill, an archaeologist at Indiana University–Purdue University Indianapolis, is working on a project to create 3D scans of about 1,000 marble fragments of the Great Marble Map of Rome, which is thought to depict every ground floor room in ancient Rome.
• She and her colleagues are creating a VR experience where fragments can be viewed, turned around and set next to others to study them in ways they can't with the actual marble pieces — which range in size from a grapefruit to 150-pound pieces. They're also looking to use AI to piece together the fragments.

What to watch: How the field moves to preserve and standardize all this amassing data.

• It's not just information about structures and artifacts themselves, but also about the environment objects are excavated from that contains valuable information about connections between artifacts.
• These objects can never be excavated again, says Sullivan, who is using GIS and 3D modeling to recreate Saqqara, Egypt's oldest known pyramid built around 2630 B.C., and its surroundings.
• "In 150 years, are archaeologists going to be able to access and reuse this digital data?" Sullivan says. "If the answer is no, we have a real problem."