Peru’s Nasca Lines have intrigued archaeologists since they were first studied nearly a century ago. Scratched on the surface of the ground between 500 B.C. and A.D. 500, these ｂ geometric figures depict living creatures, stylized plants, and imaginary beings, as well as geometric figures several kilometers long.
A total of 300 geometric figures have been identified thus far, though archaeologists remain unsure as to their exact purpose. For some, they represent astronomical observations, while others think they may hold coded messages to communicate with deities of the time. Others still have theorized everything from aliens to supernatural forces.
A key difficulty for scientists is the fact that some of these etchings are 370 meters long, which makes studying them at ground-level very difficult. But studying them from above is not much easier – the area to survey is extremely vast and the natural erosion of the surroundings mean the lines and patterns are not always clearly visible.
Using AI to overcome human limitations
A team of researchers from Japan’s Yamagata University has used deep-learning to increase its observational and analytical capabilities – effectively improving the way it identifies, interprets, and preserves the geoglyphs.
Working with IBM Research, the team uses a cloud platform to stitch together massive amounts of geospatial data, including lidar, drone, and satellite imagery and geographical surveys, to create high-fidelity maps of search areas.
Neural networks are then trained to recognize data patterns of known shapes to uncover new ones. This exercise proved harder than expected, as each of the 100 biomorphic geoglyphs identified so far is fairly unique – making it difficult for the AI algorithm to learn what to look for when tracking down hidden shapes.
According to Professor Masato Sakai, leader of the project, the team had to “specifically build techniques in the deep-learning framework to learn and distinguish between these different patterns and sizes of the geoglyphs.”
Breaking new ground faster than before
Despite the obstacles, the exercise paid off. During the testing phase, the AI discovered a new design – a small humanoid figure that had been missed in previously collected data. The discovery took only two months, as opposed to the several years that previous methods had required.
Professor Sakai notes that it is a “major achievement to find a new geoglyph in an area that was often investigated.”
These exciting breakthroughs are just the start of more work that can now be done using AI.
“This technology and these efforts are expected to promote understanding of all the Nasca Lines as a whole, and accelerate research and awareness towards activities to protect this World Heritage Site,” write the researchers.
From discovery to preservation
The use of AI to quickly and precisely identify geoglyphs is important to not only understand the vast extent and complexity of the Nasca Lines, but also to ensure they can be preserved for future generations.
“The expansion of urban areas in the Nasca region risks damaging the lines,” write the researchers. “There is an urgent need to gain an accurate understanding of the geoglyphs’ location and distribution so that they can be adequately protected.”
Declared a UNESCO World Heritage site in 1994, the Nasca Lines risk being destroyed if they are quickly and precisely mapped out.
“The most important point is not the discovery itself,” says Professor Sakai. Indeed, “gaining a detailed understanding on the worldview of the people who made and used these geoglyphs” is one side of the equation; the other is the fact that making them more visible, should ensure they are “protected as important cultural heritages.”
In this context, the AI serves as a tool to communicate across generations; communication between the people who created the geoglyphs and us finding them now, and communication between those working to preserve them today and future generations who will be able to learn about them.