Karez Irrigation System: How Underground Water Sustained Silk Road Oases

Drive out of Turpan towards the mountains and you will see them before you understand them: lines of low earth mounds marching across the gravel, each ringing a dark hole in the ground. They are the visible tops of the karez irrigation system, a network of hand-dug wells and tunnels that carries groundwater from the foot of the mountains to the fields of a basin that sees almost no rain. Understanding how that water moves, and where the idea came from, explains a good deal about why oasis towns could exist on the Silk Road at all. The karez is also a stop on Day 12 of our small group Silk Road journey from Xi’an to Tashkent, on the morning the group leaves Turpan for Urumqi; our overview of the history-led Silk Road tour from Xi’an to Tashkent sets that day within the whole route.

An oasis in a basin without rain

Turpan sits in a deep fault basin on the southern side of the eastern Tian Shan. The floor of the Turpan Depression drops to about 154 metres below sea level at Ayding Lake, and the basin is ringed by mountains. A 2018 report in the New York Times put average rainfall at about half an inch, roughly 13 millimetres, a year, and the tour page notes that summer temperatures around the Flaming Mountains can reach 50 °C (122 °F) or higher. On those figures, farming by rain alone is impossible.

What Turpan does have is water underground. Snow and ice on the Tian Shan melt each summer, and the meltwater runs down into the great fans of gravel and sand that spread out from the foot of the range. Much of it sinks into those fans and moves slowly downhill beneath the surface, towards the lowest part of the basin. A river crossing that gravel in the open would lose much of its flow to evaporation and seepage long before it reached a field. The challenge for any oasis community here was not finding water, but bringing it to the surface where it was needed without losing it to the sun on the way.

How the karez irrigation system works

A karez solves that problem with gravity and patience. Diggers begin at the foot of the slope where the groundwater lies and sink a vertical well down into the water-bearing gravel. From the fields far downhill they then dig a tunnel back up towards that source, cut at a slope only slightly gentler than the slope of the ground above. Because the tunnel falls more gently than the land, it emerges at the surface some distance down the fan, delivering water by gravity alone. No pumps, wheels or animals are needed: once the channel is open, the water flows day and night.

The tunnel cannot be dug from its two ends alone. Along its line, workers sink a series of vertical shafts, which let them haul out spoil, let air into the workings and give access for cleaning and repair once the channel is running. The excavated earth piled around each shaft opening produces the rows of mounds that are the karez’s signature on the landscape. In the Iranian qanats from which the karez takes its name, these shafts are typically a few tens of metres apart, and the deepest, at the head of the system, can go down well over a hundred metres.

China’s submission of the Turpan karez to UNESCO’s Tentative List in 2008 describes four parts: the vertical wells, the underground channel that forms the main body, an open channel where the water reaches the surface, and a small reservoir or pond from which it is shared out. The point where the tunnel opens to daylight is called the “dragon mouth”. The same submission counts more than 1,100 karez in the Turpan area, spread across Turpan city and the counties of Shanshan and Toksun, with a combined channel length running to several thousand kilometres.

Keeping the water underground is the whole point. A channel in tunnel loses far less to evaporation than an open canal in Turpan’s heat, and it is protected from the wind-blown sand that would choke a surface ditch. The cost is labour: shafts collapse, silt builds up and the tunnel floor must be cleaned, so a karez survives only as long as a community is willing to maintain it.

Is a karez the same as a qanat?

Essentially, yes: they are regional names for the same kind of gently sloping, shaft-dug tunnel that taps groundwater and delivers it by gravity. “Karez” comes from the Persian kārīz, one of the Iranian names for a qanat, and the same technique is called a falaj in Oman and a foggara in Algeria. Local practice differs in depth, length and how the water is shared, but the engineering principle is the same.

Where did the karez come from?

The technique is old, and Iran is usually given as its homeland. Encyclopaedia Britannica dates the development of the qanat to about 2,500 to 3,000 years ago in Iran, and in 2016 UNESCO inscribed eleven Iranian qanats on the World Heritage List as “The Persian Qanat”. Under the Achaemenid Persian empire the technique travelled with imperial power. The Greek historian Polybius records that the Persians granted the use of land for five generations to anyone who brought water to ground that had none, and qanats at the Kharga Oasis in Egypt’s Western Desert are dated to the 5th century BC, when Egypt was under Persian rule.

How and when the technique reached Turpan is a genuine debate, and it is worth knowing the main positions before you stand beside a karez yourself. Writing in the journal The Silk Road in 2010, the researcher Arnaud Bertrand summarised three theories in Chinese scholarship. The first holds that the karez was imported from Persia across Central Asia. The second argues that it was developed locally in Turpan. The third, associated with the early 20th-century scholar Wang Guowei, traces it to China’s own heartland: the Shiji, Sima Qian’s history of the Han, describes the Longshou Canal dug under Emperor Wu (r. 141–87 BC) in Shaanxi, where workers sank a line of wells and linked them underground to carry water beneath a hill. China’s 2008 UNESCO submission follows this line, crediting Han Chinese with bringing the well-sinking method to Xinjiang.

The documentary evidence for Turpan itself points later. Bertrand, drawing on the historian Éric Trombert’s study of the written sources, concludes that the karez was introduced to Turpan no earlier than the 17th and 18th centuries. The Qing statesman Lin Zexu, exiled to Xinjiang in the 1840s, is widely credited with promoting karez building there, and China’s UNESCO submission names him in that role. The careful position, then, is that the principle is ancient and Persian in origin, that a related shaft-and-tunnel technique was known in Han China, and that the dense network around Turpan seen today was built mostly in the last three centuries.

Oases, caravans and trade: what underground water made possible

If the karez network is relatively late, how did ancient Turpan survive? The answer is the same principle on a simpler scale. Bertrand describes how the oasis capital of Gaochang was fed by a main canal, the Manshui, which brought water into the heart of the oasis and fed a ditch around the city walls, while the cliff-top city of Jiaohe relied on around 300 wells, most of them dated from the 5th century to the end of the Tang, and dug to at least 20 to 25 metres. Whether by canal, well or karez, the oasis depended on capturing water that came down from the mountains and keeping it out of the sun. The story of those two cities, the kingdoms that ruled from them and the caves they paid for is told in our guide to Turpan’s oasis archaeology and the northern Silk Road.

That water is what turned a desert basin into a waystation. The Silk Road was never one road but a chain of places where caravans could rest, water their animals, buy food and trade. On the route north of the Taklamakan, the oases strung along the foot of the Tian Shan were those places, and Turpan was a major one. Our overview of the Silk Road through China, from Xi’an to Kashgar follows that chain of oases stop by stop, and our wider guide to the history and legacy of the Silk Road places it within the whole network.

Water also shaped what Turpan produced. The oasis is known for grapes, and the green valleys cut into the foot of the Flaming Mountains, watered by streams and channels, stand out against the bare red rock around them. Grapes are dried into raisins in ventilated mud-brick drying houses, whose lattice walls let the hot, dry wind do the work. A light, durable, high-value crop like that suited a trading town: raisins travel well, and a reliable harvest gave the oasis something to sell as well as feed itself.

Seeing the karez on the Silk Road journey

The tour reaches Turpan on Day 10, transferring by coach from Dunhuang to the station and taking a high-speed train north-west into Xinjiang. Day 11 is spent on the oasis’s past: the Flaming Mountains, the Bezeklik Thousand Buddha Caves, the ruins of Gaochang, the Astana Tombs and, after lunch, Ayding Lake, now dried to a salt rime. On the morning of Day 12, before the coach transfer to Urumqi after lunch, the group visits Grape Valley and the Sugong Minaret, a mud-brick tower the tour page dates to the 1770s, and views the karez. The tour page describes the system as a series of vertically dug wells linked by underground canals, collecting water from the base of the Tian Shan and the Flaming Mountains.

Seen in that order, the visit makes sense of everything before it. The ruined cities of Day 11 only existed because someone could feed them, and the vines of Grape Valley on the morning of Day 12 show what water brought down from the mountains still grows. From Urumqi the route continues by air to Kashgar and then over the Torugart Pass into Kyrgyzstan and on to Samarkand, Bukhara and Tashkent.

How physically demanding are the Turpan days?

They are among the hotter days of the tour rather than the hardest walking. The tour is rated Level 2, Moderate, which on the tour page means being able to carry your own luggage, climb and descend stairs, and take part in three to five hours of physical activity a day, the equivalent of walking up to 8 kilometres on uneven ground. In Turpan, the open archaeological sites have little shade and temperatures can reach 50 °C (122 °F), so a hat, sun protection and plenty of water matter more than stamina.

Are karez still used in Turpan?

Some are, but far fewer than in the mid-20th century. The 2018 New York Times report cited government figures of about 200 working karez, down from nearly 1,800 in the 1950s. A 2006 study by researchers from Xinjiang Normal University and the Max Planck Institute for Chemistry linked the decline to population growth, expanding farmland and motor pumps since the 1950s, and recorded groundwater falling by 20 to 30 metres from the early 1980s.

What the karez reveals about the Silk Road

The karez irrigation system is easy to overlook beside the cave paintings and ruined cities of Turpan, yet it answers the most basic question a traveller in the Taklamakan region can ask: how did anyone live here at all? The answer lies under your feet, in a technique carried across Asia from the Iranian plateau, matched by older Chinese engineering, and rebuilt by generations of oasis farmers who kept the tunnels clean. Water made the oasis, the oasis made the waystation, and the waystation made the road.

That is the logic our Silk Road journey follows. Travelling as a small group of up to 16 travellers with an Odyssey Program Leader, you see the karez on the same morning as the vines it waters, a day after the ruined cities it once helped to feed, and then continue west along the chain of oases towards Kashgar and Central Asia.