NASA Algorithm Reveals Ancient Images from Satellite Data

NASA’s first‑ever satellite‑image algorithm, originally designed to detect subtle changes in Earth’s surface for climate monitoring, has been adapted by archaeologists to uncover hidden relics beneath the earth’s crust. The machine‑learning model, which uses convolutional neural networks to parse high‑resolution satellite data, is now fed with ground‑penetrating radar and aerial photographs of archaeological sites. By training the system on known ruins, researchers can now identify patterns that signal buried walls, roads, or even entire cities that would otherwise remain invisible to the naked eye.

The adaptation has already yielded promising results. In a recent field study in the Sahel region of Mali, the algorithm flagged a series of linear anomalies that corresponded to the remains of a pre‑colonial trading route. Similar successes have been reported in the American Southwest, where the software helped map the extent of an ancient Puebloan settlement. The technology not only speeds up site surveys but also reduces the need for invasive excavation, preserving fragile contexts for future generations.

Experts see this cross‑disciplinary application as a significant step forward for both planetary science and archaeology. By leveraging tools developed for monitoring our planet’s changing environment, scientists can now peer back into human history with unprecedented clarity. As the algorithm continues to refine its predictive accuracy, it promises to unlock more of the world’s hidden past while safeguarding the integrity of archaeological sites.