Researchers at Yamagata University in Japan used an artificial intelligence system and drone surveys to identify 303 previously unknown figurative geoglyphs etched into Peru’s Nazca desert, nearly doubling the total number of such figures on record. The discoveries, confirmed through targeted fieldwork completed in roughly six months, represent the largest single expansion of the Nazca geoglyph catalog since systematic study began decades ago. The findings raise immediate questions about how Peru’s heritage agencies will protect hundreds of newly documented sites spread across a fragile arid plateau already under pressure from agriculture and tourism.
How 303 new geoglyphs reshape what researchers know about Nazca
The AI system scanned high-resolution aerial imagery of the Nazca pampa and flagged 1,309 locations as promising candidates for previously unrecorded figurative geoglyphs, according to a university summary of the project. Field teams then checked about one-quarter of those candidates on the ground and by drone, confirming 303 as genuine figurative designs. That confirmation rate, achieved within six months of survey work, compressed a process that had taken earlier research campaigns years or even decades to accomplish at smaller scales.
The peer-reviewed study, published in the PNAS journal, reports that the 303 new figures nearly doubled the count of known figurative geoglyphs in the region. Before this campaign, the total sat in the low hundreds. Adding 303 at once shifts the statistical picture of how densely the Nazca people inscribed their desert and how many figures remain buried under centuries of windblown sediment and surface erosion.
Researchers now have a broader sample of motifs and sizes to work with. The newly documented designs include animals, plants, and human-like forms that match the stylistic conventions of previously known Nazca figures. Many are smaller than the iconic giant images that draw tourists, suggesting that large, spectacular figures may be only the most visible part of a much denser graphic landscape. The result is a more granular view of how imagery may have structured movement, ritual, and land use across the plateau.
One question the data invites but does not yet settle concerns placement patterns. The newly mapped figures appear across a wide swath of terrain rather than concentrated around the most famous line clusters visible from the air. Some researchers have speculated that geoglyph locations track subtle elevation gradients tied to seasonal fog-water collection zones, a hypothesis that would reframe the figures as markers of micro-climate resources rather than purely ceremonial or astronomical features. The PNAS study does not confirm or reject that interpretation, but the sheer number of new data points gives future analysts a far richer spatial dataset to test it against.
Yamagata’s AI workflow and what the field checks revealed
The detection pipeline combined machine learning trained on known geoglyph shapes with satellite and drone imagery covering broad sections of the Nazca plateau. By automating the initial scan, the team avoided the bottleneck of manual visual inspection across hundreds of square kilometers of desert. The AI flagged candidate sites, and researchers from Yamagata University then dispatched drones and ground crews to verify each one.
That two-stage process, automated detection followed by physical confirmation, is what allowed the team to cover so much ground in six months. Traditional survey methods rely on archaeologists walking transects or studying aerial photos frame by frame, a labor-intensive approach that had produced the existing catalog over many years. The AI system did not replace human judgment; it filtered the search space so that trained eyes could focus on the most likely targets, rejecting natural rock patterns, vehicle tracks, and modern disturbances that can mimic ancient lines in overhead images.
The 303 confirmed figures include animal, plant, and human forms consistent with the style of previously known Nazca geoglyphs. Their distribution across the pampa suggests the ancient Nazca civilization inscribed far more of the desert surface than earlier surveys had captured. Because only about one-quarter of the 1,309 candidates have been field-checked so far, the true total of undiscovered figures could be substantially higher. That gap between flagged candidates and confirmed sites is one of the most significant open questions the study leaves behind, and it highlights how much the final picture will depend on continued access, funding, and on-the-ground verification.
The study also underscores the limits of AI in archaeological contexts. While the model proved efficient at surfacing potential sites, the authors emphasize that context-soil disturbance, stratigraphy, associated artifacts-can only be assessed in person. False positives remain a concern, especially in areas where modern vehicle ruts or irrigation ditches cut across older surfaces. The team’s decision to publish only confirmed figures reflects an effort to balance rapid discovery with the discipline’s standards of evidence.
Protection gaps and unresolved questions after the Nazca expansion
Peru’s cultural heritage authorities now face a practical problem: hundreds of newly documented sites scattered across a desert that already stretches monitoring capacity thin. The Nazca Lines hold UNESCO World Heritage status, but that designation covers a defined zone, and many of the new geoglyphs sit in areas where expanding agricultural irrigation and informal road construction have already damaged surface features. Reporting in a British newspaper notes that Peruvian archaeological officials see the discoveries as evidence that many figures marked ritual pathways connecting sites rather than functioning as isolated monuments, a reading that would demand corridor-level protection rather than point-by-point fencing.
That pathway interpretation complicates management plans. Protecting individual figures from tire tracks or encroaching fields is already difficult; safeguarding long, continuous routes that may stretch for kilometers across mixed public and private land is harder still. If the new geoglyphs indeed form segments of broader ceremonial or processional networks, regulators may need to rethink how buffer zones are drawn and how infrastructure projects are screened before approval.
Several gaps in the public record limit independent evaluation of the results. The full candidate-detection dataset and exact model parameters remain unavailable beyond the PNAS summary. No field-verification logs or permit records from Peru’s Ministry of Culture have been released to corroborate the six-month timeline independently. Updated GIS coordinates for all 303 new figures have not yet appeared in any public archaeological registry, which means outside researchers cannot replicate the spatial analysis or test hypotheses about placement patterns tied to water resources or terrain.
The absence of direct statements from on-site drone operators or local community monitors also leaves a gap. Institutional summaries from Yamagata University describe the workflow and outcomes, but ground-level accounts of how communities near the Nazca pampa experienced the intensified survey season have not been widely circulated. That silence matters because local residents often serve as informal guardians of nearby geoglyphs and are among the first to notice damage from vehicles, land invasions, or new irrigation projects.
Looking ahead, the study’s authors argue that AI-assisted surveys can help heritage managers get ahead of threats by identifying vulnerable sites before they are disturbed. Yet without clear commitments from Peruvian authorities to expand patrols, update zoning rules, and involve local communities in monitoring, the rapid growth in the geoglyph catalog may outpace the state’s ability to protect what is being mapped. The Nazca plateau now appears more densely inscribed than many specialists had assumed. Whether those newly visible lines will survive the next wave of development may depend less on algorithms than on the political and financial choices made in the wake of this research.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
