Spanish archaeologists have recently made impressive strides in uncovering ancient underground irrigation systems called qanats, located in arid desert regions across the globe, with the help of artificial intelligence (AI).
This AI, specifically designed to analyze Cold War-era spy satellite imagery, has given researchers an unprecedented means of locating these hidden water channels.
By sifting through thousands of satellite photos taken during the Cold War, the AI system identifies the presence of qanats, which brought vital water from higher altitudes to sun-scorched plains below.
Unlike the more familiar above-ground Roman aqueducts, which transport water across impressive stone arches, qanats were created by digging channels deep underground. They are punctuated by a series of vertical shafts that served dual purposes: they allowed air to circulate within the channels and enabled maintenance access.
Because they were buried underground, qanats offered the advantage of shielding the water from the intense desert sun, reducing evaporation and ensuring a steady water supply. This method of water transport was critical for ancient civilizations living in desert areas, allowing them to sustain agriculture and communities in otherwise uninhabitable regions.
One of the most extensive networks of these ancient qanats lies in Iran, where they form part of the UNESCO-listed Great Wall of Gorgon, a significant cultural heritage site that received its designation in 2016.
However, qanats are not unique to Iran; these ancient water systems have been discovered across North Africa, Central Asia, South America, and even China. Their widespread presence across continents is a testament to the ingenuity and adaptability of early human societies in arid environments.
The project to identify qanats through AI is led by Hector Orengo and his team at the Catalan Institute of Classical Archaeology in Spain. Orengo explains that these structures are remarkable achievements in early engineering, allowing civilizations to thrive in places where survival would otherwise be impossible.
"These systems were extremely innovative," he said to New Scientist, adding, "They allowed people to live in areas where it would have been unthinkable before."
Many of the qanats discovered by AI date back as far as 3,000 years, yet they are challenging to locate because, unless one is standing directly near an access shaft, they remain virtually invisible.
To train the AI, Orengo and his team used historical satellite photos captured by the United States HEXAGON spy satellite program, which operated from 1959 to 1986. These Cold War-era satellites photographed extensive areas of the Middle East and North Africa, precisely the regions where qanats are most common.
By feeding the AI satellite images of known qanat sites, including the Rissani area of Morocco, the Maiwand region of Afghanistan, and Iran's arid Gorgan plain, researchers taught the system to recognize the subtle patterns that indicate the presence of qanats.
The AI's accuracy in identifying qanats has proven to be remarkably high. When it analyzed potential qanat locations, it accurately identified these ancient irrigation channels 88% of the time. Even more impressively, in 62% of the correctly identified cases, the AI could trace the entire length of the qanat.
This precision is invaluable for archaeologists, as it allows them to map out these channels without physically surveying each one, which would be time-consuming, costly, and sometimes dangerous.
With this level of accuracy, the AI represents a major breakthrough in the field of archaeology. Not only does it significantly reduce the time and expense involved in identifying qanats, but it also provides a safer alternative to on-the-ground surveying.
Field surveys are often challenging to conduct in remote desert locations or in areas with political instability. The AI now offers a method that is both more efficient and far safer, enabling archaeologists to locate and study these ancient water channels from afar.
Looking forward, this AI-powered approach is expected to help locate previously undiscovered qanats in other regions. By examining additional satellite imagery, it has the potential to reveal even more about how ancient societies managed to sustain their populations and agricultural activities in some of the harshest environments on Earth.