Archaeologists Using AI Discover 303 Unknown Geoglyphs In Peru’s Nazca Desert


With the help of an artificial intelligence (AI) system, a group of archaeologists has uncovered in just a few months almost as many geoglyphs in the Nazca Desert (Peru) as those found in all of the last century. The large number of new figures has allowed the researchers to differentiate between two main types, and to offer an explanation of the possible reasons or functions that led their creators to draw them on the ground more than 2,000 years ago.

The Nazca desert, with an area of about 1,900 square miles and an average altitude of 500 meters above sea level, has very special climatic conditions. It hardly ever rains, the hot air blocks the wind and the dry land has prevented the development of agriculture or livestock. Combined, all this has allowed a series of lines and figures, formed by stacking and aligning pebbles and stones, to be preserved for centuries. The first layer of soil is made up of a blanket of small reddish stones that, when lifted, reveal a second yellowish layer. This difference in color is the basis of the geoglyphs and is what was used to create them by the ancient Nazca civilization. Some are straight lines stretching several miles. Others are geometric shapes or rectilinear figures, also huge in size. The other major category includes the so-called relief-type geoglyphs, which are smaller. In the 1930s, Peruvian aviators discovered the first ones, and by the end of the century more than a hundred had been identified, such as the hummingbird, the frog and the whale. Since 2004, supported by high-resolution satellite images, Japanese archaeologists have discovered 318 more, almost all of them high-profile geoglyphs. The same team, led by Masato Sakai, a scientist from Yamagata University (Japan), has discovered 303 new geoglyphs in a single campaign, supported by artificial intelligence.

“It speeds up the discovery process,” Sakai admits, asked about the advantage that artificial intelligence brings. “The Nazca Pampa is a vast area covering more than 400 square kilometres and no exhaustive study has been carried out,” the Japanese scientist recalls. Only the northern part, where the large linear geoglyphs are concentrated, “has been studied relatively intensively,” he adds. But scattered throughout the rest of the desert are many relief-type figures that are smaller and that the passage of time has made more difficult to detect.

Convinced that there were many more, Sakai and his team contacted IBM’s artificial intelligence division, which is where the Watson system came from. They had high-resolution images obtained from airplanes or satellites of all of Nazca, but with a resolution of up to a few centimeters per pixel, the human eye would have needed years, if not decades, to analyze all the data. They left that job to the AI system. Although it was not easy to train its artificial vision (see comparison below) with so few previous images and so different from each other, the machine proposed 1,309 candidates. The figure came from a previous selection also made by the AI with 36 images for each candidate. With this selection, the researchers carried out a field expedition between September 2022 and February 2023. The result, as reported in the scientific journal PNAS, is 303 new geoglyphs added to this cultural heritage of humanity. All are relief-type geoglyphs.

On the left, what the artificial intelligence system sees. On the right, the geoglyph on the ground. The system identified 1,300 candidates for neoglyphs. A quarter of them turned out to be real.
On the left, what the artificial intelligence system sees. On the right, the geoglyph on the ground. The system identified 1,300 candidates for neoglyphs. A quarter of them turned out to be real.Nature/Instituto Universitario Yamagata de Nazca

The newly discovered shapes bring the total number found in Nazca to 50 line-type and 683 relief-type geoglyphs, some geometric and others forming figures. The large amount has allowed the authors of this work to detect patterns and differences. Almost all of the former (the monkey, the condor, the cactus…) represent wild animals or plants. However, among the latter, almost 82% show human elements or elements modified by humans (humanoids, domesticated animals such as llamas, and many decapitated heads, up to 33% of the total). “They are scenes of human sacrifice,” says Sakai.

The location of these severed heads and of almost all the other relief-type geoglyphs provides another clue, in this case about their function. The accumulation of data that has made this work possible brings to light a double connection. On the one hand, these relief-type forms are found a few meters from one of the many paths that cross the desert. They are not roads per se, but paths created by the passage of people until a path is created. According to the authors of the study, these creations were made to be seen by travelers. On the other hand, the large linear figures appear very close, also meters away, from one of the many straight lines that cut through the pampas. Here, according to Sakai, the symbolic value rules: “The line-type geoglyphs are drawn at the start and end points of the pilgrimage route to the Cahuachi ceremonial center. They were ceremonial spaces with shapes of animals and other figures. Meanwhile, the relief-type geoglyphs can be observed when walking along the paths.” Cahuachi was the seat of spiritual power of the Nazca culture between from around100 BC to 500 AD and, for the authors, the large forms could be ceremonial stops on the pilgrimage to or from there.

These explanations do not necessarily rule out, according to the authors, other possible functions that have been attributed to the Nazca lines and figures, such as being calendars, astronomical maps or even systems for capturing the little water that fell. What they are sure of is that artificial intelligence can revolutionize archaeology, as it has done with Nazca.

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition



Source link

About The Author

Scroll to Top