The mountain range that determines Asia’s weather — and its wars

The canyon begins at the bend. For fifty kilometers, the Yarlung Tsangpo River falls two thousand meters through the eastern Himalayas — dropping faster than almost any comparable waterway on Earth, cutting the deepest river gorge on the planet before crossing into India and becoming the Brahmaputra. On Christmas Eve 2024, the Chinese government approved construction of the world’s largest planned hydropower project at this precise site: a dam with a projected capacity of 60 gigawatts and a cost of $137 billion, nearly three times the generating capacity of the Three Gorges Dam. Construction began July 19, 2025.

The site is fifty kilometers from a disputed border. Two nuclear powers fought over this frontier in 1962 and killed soldiers along it in 2020. Western media covered the dam announcement mostly as an energy story — China’s clean power ambitions, its carbon targets, its engineering credentials. That framing misses what the dam actually is. To understand what it is, you have to start not with the dam but with the plateau above it.

The Machine on the Roof of the World

At an average elevation exceeding 4,500 meters, the Tibetan Plateau covers 2.5 million square kilometers — roughly the area of Western Europe — making it the highest and largest plateau on Earth. These sound like geography-class facts: inert, forgettable, the kind of figures that vanish between reading and memory. But the elevation isn’t merely impressive. It’s the mechanism.

In summer, the plateau’s surface heats faster than the surrounding air at the same altitude because the atmosphere above it is thin and absorbs little solar radiation while the ground below absorbs much. The result is a vast low-pressure system suspended several kilometers above the surrounding terrain — effectively a furnace floating in the mid-troposphere. That pressure differential draws moist air northward from the Indian Ocean in a massive seasonal circulation: the South Asian monsoon. The plateau doesn’t sit there, elevated and scenic. It runs the machine.

The monsoon is not a weather event. It is the annual delivery of roughly 70 to 80 percent of South Asia’s total rainfall — a figure cited consistently in India Meteorological Department data and across India’s National Water Policy documents, and confirmed for the South Asian region as a whole by the World Meteorological Organization’s South Asian Climate Outlook Forum. Crops, groundwater recharge, drinking water for cities, river flow in the dry season: all of it is downstream of the monsoon, which is itself downstream of the plateau. Hundreds of millions of smallholder farmers across India, Pakistan, Bangladesh, and Nepal have no alternative water source when the monsoon fails or arrives late.

Consider the geometry. The Indian subcontinent and North Africa sit at nearly identical latitudes. North Africa is the Sahara. South Asia supports over a billion people in rain-fed agriculture. The difference is a plateau several kilometers above sea level to the north — one that deflects the jet stream, blocks cold Siberian air from moving south, and drives the pressure differential that pulls ocean moisture inland. Son et al. mapped the downstream monsoon consequences of this dynamical forcing in Geophysical Research Letters in 2020. The plateau doesn’t contribute to South Asian food security. It makes it possible.

The Third Pole

The plateau’s second function is storage. The Tibetan Plateau and its surrounding ranges — the Himalayas, Hindu Kush, Pamir, Karakoram — harbor glaciers covering more than 100,000 square kilometers, making this the largest repository of frozen freshwater outside the Arctic and Antarctic. Glaciologists and the World Meteorological Organization call it the Third Pole, a designation that is not promotional. It is a physical description of scale.

Ten major river systems originate here. Flowing east into China and Southeast Asia: the Yangtze, Asia’s longest river; the Yellow River; the Mekong, which passes through Myanmar, Laos, Thailand, Cambodia, and Vietnam before reaching the South China Sea; the Salween. Flowing south and west: the Brahmaputra, which enters India through Arunachal Pradesh and crosses Bangladesh to the Bay of Bengal; the Ganges, fed by Himalayan glaciers including the Gangotri; the Indus, which runs through India and Pakistan.

Direct river basin water dependency — the households and farms that actually drink from these rivers — runs to roughly 1.35 billion people, per a Council on Foreign Relations backgrounder on the Tibetan Plateau. The Tibet Policy Institute puts broader dependency, including indirect sourcing, at approximately two billion. Beyond that, three billion or more people live in agricultural zones where the monsoon determines whether crops grow. These are different claims: the first is about rivers, the second about rain.

The glaciers do something the monsoon doesn’t: they sustain river flow when rain has stopped. For the upper Indus basin in particular, glacier and snowmelt provides a disproportionate share of peak summer flows — considerably more than in the monsoon-dominant rivers to the east, and the reason Pakistan faces sharper exposure than India when the glacial contribution begins to fall. The upper Indus is more glacier-dependent because it sits furthest from the monsoon’s core reach; it depends on frozen storage the way the rest of South Asia depends on rain.

The glaciers are retreating. Roughly 82 percent are showing net mass loss. At 1.5 degrees of global warming, one third are projected to disappear by end of century; at 2 degrees, more than two thirds. The short-term effect is more meltwater as stored ice releases — rivers swell temporarily, masking the threat. The medium-term effect, as glaciers thin and ultimately vanish, is catastrophic reduction in dry-season flow. The plateau holds a finite bank of freshwater capital, and the balance is falling. Whoever controls the infrastructure on this plateau has a narrowing window to extract maximum advantage from it.

Sidebar A: The Peak Water Problem

Before glaciers disappear, they first release more water. This is called "peak water" — the moment at which accelerating melt produces maximum runoff before the glacier becomes too thin to sustain it. For some Himalayan glaciers, particularly in the western ranges, peak water has already passed or is imminent. For others, it lies decades ahead.

The practical consequence is paradoxical: rivers fed by retreating glaciers can be at historically high flow right now, even as the long-term trajectory bends toward severe shortage. For the Indus system — the most glacier-dependent of the major South Asian rivers, sitting furthest from the monsoon's reach — peak water arrival means a window of apparent abundance followed by a cliff. Pakistan, which depends on Indus water for one of the world's largest irrigation systems, faces the most acute exposure.

The leverage equation is therefore doubly asymmetric. Upstream infrastructure can shape flow in the short term. Glacial retreat will reshape it catastrophically in the medium term. Neither process is reversible on any timescale relevant to political planning. The clock runs on the leverage itself.

The Acquisition

All of this — the monsoon engine, the glacier bank, the river systems descending to ten downstream countries — was politically neutral until October 1950, when the People’s Liberation Army entered Tibet.

The Chinese government called it liberation. The government in Lhasa called it invasion. Tibet was formally incorporated into the People’s Republic under the Seventeen Point Agreement of 1951, signed under military duress; no serious historian outside China disputes this characterization. The primary strategic calculations were territorial — securing the PRC’s national boundary on the ethnic-nationalist terms the new state had defined for itself, establishing a buffer against India and the Western powers, and keeping Tibet out of American reach in the opening decade of the Cold War. There is no reliable evidence that Mao Zedong’s strategic calculus explicitly prioritized hydraulic resources.

It doesn’t matter. What China inherited in 1950 was the upstream position on rivers flowing into India, Pakistan, Bangladesh, Nepal, Bhutan, Myanmar, Thailand, Cambodia, Laos, and Vietnam. Ten downstream countries. No other territorial acquisition in the twentieth century — not Soviet expansion in Central Asia, not any colonial land grab — transferred hydraulic leverage over a comparable number of downstream populations simultaneously. The hydraulic windfall was extraordinary whether or not it was intended.

And nothing constrains how China uses it. China has no comprehensive water-sharing treaty with any downstream riparian state. The Mekong River Commission, established in 1995, has China as a dialogue partner but not a full member; China is not bound by its decisions or obligations. No equivalent governance body exists for the Brahmaputra. The Ganges Water Treaty between India and Bangladesh (1996) governs that bilateral relationship but is silent on Chinese upstream behavior. Brahma Chellaney, whose 2011 book “Water: Asia’s New Battleground” (Georgetown University Press) named this structural condition “hydro-hegemony,” observed that China is the only state in the world controlling the headwaters of rivers flowing to the greatest number of downstream countries simultaneously. That was in 2011. Nothing has changed except that China has built more dams.

The Architecture of Leverage

Structure is not strategy. Capability is not behavior. The question is whether China has actually exercised the leverage the plateau provides — and the answer, on two rivers and in two documented incidents, is yes.

Start with the Mekong. China has built twelve dams on its section of the mainstream — more than any other riparian state on a river of comparable international significance. Downstream countries — Myanmar, Laos, Thailand, Cambodia, Vietnam — collectively depend on the river for irrigation, fisheries, and drinking water for tens of millions of people. They have no mechanism to constrain Chinese dam operations.

In April 2020, researchers at Eyes on Earth, working under contract to the Stimson Center’s Southeast Asia program, published a satellite remote-sensing analysis of water levels and precipitation across the entire Mekong basin. Their finding: throughout most of 2019, rainfall in the Chinese portion of the catchment was normal to above average. Despite this, China’s dams were impounding more water than at any comparable period — specifically, holding back wet-season flows to maximize dry-season generation, when electricity prices run approximately three times higher. The result downstream was the worst drought in living memory. Fisheries collapsed. Crops failed. The study, authored by Brian Eyler and Courtney Weatherby, wasn’t modeling a risk or projecting a scenario. It was documenting a physical outcome from satellite data.

China denied that its operations constituted water deprivation and offered a data-sharing platform — the Lancang-Mekong Cooperation hydrological mechanism — that downstream analysts have consistently characterized as a transparency performance with no enforcement architecture. The Mekong River Commission acknowledged the evidence. It has no authority to act on it.

Then there’s the Brahmaputra. During the 2017 Doklam standoff — a 73-day military confrontation between Indian and Chinese forces over a disputed plateau in Bhutan — China stopped providing India with hydrological data for the Brahmaputra’s main Tibetan tributary. That data is essential every monsoon season: without it, India cannot anticipate flood surges bearing down on Assam and the northeastern states. China’s stated reason was upgrades to measurement stations in Tibet. At the same time, China continued sharing the same data with Bangladesh.

That comparison requires no interpretation. China held flood-warning information over India’s head during a military standoff and kept sharing it with the downstream neighbor that posed no threat to Chinese interests. The leverage was not deployed at scale. But it was not theoretical.

Sidebar B: The "Run-of-River" Question

China's official description of the Yarlung Tsangpo dam is a "run-of-river" hydropower scheme: water passes through turbines and returns to the river without large-scale reservoir impoundment. This framing is used to deflect concerns about weaponization — without a reservoir, the argument goes, China cannot impound water at meaningful scale.

The argument has limits. Run-of-river dams still affect flow timing and sediment loads, both of which downstream agriculture and flood-control systems depend on. Even without a large reservoir, a structure controlling a canyon choke point can modulate the timing and volume of flood pulses. Several of China's Mekong installations were also described as non-reservoir projects; the Eyes on Earth satellite analysis documented their real-world effects on downstream flow regardless.

As Yale Environment 360 and analysts writing in the Made in China Journal have noted, the "run-of-river" framing does not resolve the core concern: a dam at this location gives China structural control over the precise point where the Brahmaputra gathers maximum force before crossing into India. Whether it's operated with hostile intent is a separate question from whether it could be.

The Border Dispute, Reframed

The India-China border conflict is almost always narrated as territorial nationalism: two large states with overlapping claims, a humiliating war in 1962, a Line of Actual Control that neither side has formally defined, and periodic incidents where soldiers push into the same mountain valley. That narrative isn’t wrong. But it’s incomplete in a way that makes specific patterns of the dispute — why the eastern sector is so much more contested than the western, why certain valleys matter more than the maps suggest — harder to explain than they should be.

The border runs in three sectors. In the west: Aksai Chin, a high-altitude desert administered by China and claimed by India, which sits at the head of the Indus drainage basin and provides the land connection between Tibet and Xinjiang along the Karakoram Highway. China holds this primarily for interior connectivity reasons; the hydraulic argument is weakest here. In the east: Arunachal Pradesh, administered by India, claimed by China as “Zangnan” — South Tibet.

Arunachal Pradesh is the territory through which the Yarlung Tsangpo descends from the Tibetan Plateau into India. The river enters Arunachal after exiting Tibet; the canyon where the dam is being built is fifty kilometers from the McMahon Line — the boundary India recognizes and China does not. Control of the Indian-administered side of this border would give China territorial reach over the last segment of the Brahmaputra’s plateau descent before it enters the Assam plain. That’s not a peripheral territorial dispute. That’s a contest over the exit point of China’s most concentrated hydraulic leverage over India.

The 1962 war ended with China in control of Aksai Chin — which it needed for the Tibet-Xinjiang road — and India retaining Arunachal Pradesh. The hydraulic reading of the outcome is not a claim about what the 1962 campaign explicitly intended; the available evidence suggests the war was primarily about Aksai Chin, buffer territory, and military demonstration. But the outcome left China holding the upper Brahmaputra and India holding the territory through which the river crosses into Indian administrative control. China has contested India’s hold on the eastern sector continuously in the six decades since.

In June 2020, in the Galwan Valley in the western sector, twenty Indian soldiers and at least four Chinese soldiers died in hand-to-hand fighting — the first confirmed combat deaths on this border in 45 years. The immediate trigger was Indian road construction toward the Line of Actual Control. The hydraulic angle here is indirect at best; Galwan was about roads, LAC enforcement, and Chinese sensitivity to Indian military access in the region. Attributing it primarily to water would overstate the case.

But the structural argument doesn’t require individual incident-by-incident intent. China’s territorial claims in both sectors, taken together, form a perimeter that keeps the Tibetan Plateau — and its full drainage — inside Chinese administrative control. Whether any single military incident was explicitly motivated by water, the territorial outcome China is pursuing tracks the water geography with uncomfortable precision.

In October 2024, India and China reached a patrol agreement that ended a four-year military standoff at several LAC friction points. The agreement addressed where soldiers walk. It said nothing about water. In July 2025, construction began on the Yarlung Tsangpo dam. In mid-2025, India’s Central Electricity Authority announced a 6.4 trillion rupee ($77 billion) program to develop approximately 76 gigawatts of hydropower capacity from the Brahmaputra basin, with Arunachal Pradesh at its center — a program the state’s chief minister described explicitly as a “national security necessity.” The political conversation found a resolution. The physical conversation accelerated in both directions.

Sidebar C: The End of the River

Bangladesh sits at the confluence of the Brahmaputra and the Ganges, receiving what China and India, in their respective dam-building races, decide to send downstream. It has no water-sharing treaty with China. It was not consulted on the Yarlung Tsangpo dam. It has no seat in any India-China border negotiation. Bangladesh is not a participant in the upstream contest — only its consequence.

Bangladesh's northeastern agricultural region depends on Brahmaputra flooding for soil replenishment and irrigation. As India builds more dams in Arunachal Pradesh, it establishes its own upstream leverage over Bangladesh — leverage Bangladesh has no capacity to counter. This cascade is noted explicitly in the 2024 Springer Nature analysis of India-China water tensions: upstream actors secure their position before the downstream neighbor can establish counter-leverage, and the logic propagates until it runs out of river. Bangladesh is where the river runs out.

The War Already Underway

The phrase “water wars” has been a fixture of geopolitical forecasting since the 1990s, appearing in think-tank reports and UN warnings with sufficient regularity to have become, through repetition, nearly meaningless. It is consistently future-tense: coming decades, approaching crisis, potential flashpoint. The convenience of the future tense is that it never has to be right.

Something already happened.

In 2019, downstream Mekong countries suffered the worst drought on record. Satellite data confirmed that China’s Mekong dams were impounding more water than in any comparable period, despite normal to above-average rainfall on the Chinese side of the catchment. The economic incentive was clear: power prices run approximately three times higher in the dry season, so holding back wet-season flows and releasing them for dry-season generation maximizes revenue. The cost was borne entirely by Vietnam, Cambodia, Thailand, Laos, and Myanmar. No treaty was violated. No legal mechanism produced a remedy. The International Crisis Group’s 2024 report, “Dammed in the Mekong: Averting an Environmental Catastrophe,” documents the template in full: a shared river managed as a revenue asset for the upstream state, its downstream flows sacrificed to dry-season electricity prices, its fisheries and floodplains bearing costs that nobody upstream was ever asked to account for.

The Brahmaputra front is at an earlier stage, but the infrastructure is being installed. The Yarlung Tsangpo dam’s 60-gigawatt capacity represents a structural intervention at precisely the point where the river transitions from Tibetan plateau waterway to India’s border river. Its $137 billion price tag is roughly two and a half times India’s entire annual defense budget. Construction is projected to complete around 2033. India’s counter-program — the $77 billion northeast hydropower initiative framed explicitly as a national security response — is the downstream answer to the upstream bet. The race-logic is self-reinforcing: each upstream actor secures position before the downstream neighbor establishes counter-leverage, and the pressure propagates until it exhausts the system.

What does not exist anywhere in this architecture is a binding legal framework governing any of it.

The 1997 UN Watercourses Convention, which establishes principles of equitable utilization and the obligation not to cause significant harm to co-riparian states, has been ratified by 42 states as of 2026, verified against the UN Treaty Collection. Neither India nor China is among them. China voted against adoption in 1997; India abstained. The Indus Waters Treaty between India and Pakistan, brokered by the World Bank in 1960, remains the only successful negotiated upstream-downstream water framework in the region — and it functions because both parties faced genuine mutual vulnerability and accepted external mediation. No equivalent architecture exists for the Brahmaputra. No equivalent negotiation is underway.

The plateau generates almost no rainfall of its own. China drinks almost none of the water it releases. The rivers flow away from China toward the populations that need them. The state that least depends on these rivers controls all of them.

The Canyon, Revisited

The Yarlung Tsangpo dam is projected to be operational by approximately 2033. When the turbines turn, it will be the world’s largest hydropower installation, built at the point where the Brahmaputra is most kinetically powerful and China’s upstream position is most concentrated. What passes through it — how much, when, in what seasonal configuration — will be determined in Beijing.

A binding Brahmaputra governance framework comparable to what the Indus Waters Treaty produced would require China to accept constraints on infrastructure it has spent $137 billion to build. Chinese membership in a functional Brahmaputra commission would require Beijing to accept obligations it explicitly rejected in 1997. Neither of these things exists. Neither is being negotiated. The October 2024 patrol agreement addressed where soldiers stand. It did not address water.

From the canyon, the river descends into Arunachal Pradesh. The same drop that makes the dam so enormously powerful — two thousand meters in fifty kilometers, the most concentrated hydraulic energy in Asia — also makes the crossing point legible as something other than geography: the precise location where control of the plateau translates into control of what arrives downstream. What arrives downstream — how much, and when — will be determined by whoever holds the canyon. After 2033, that is not a geographic fact. It is a political one.

Avis de non-responsabilité de Gen AI

Certains contenus de cette page ont été générés et/ou édités à l'aide d'une IA générative.

Les médias

Zekai Zhu – Pexels

Principales sources et références

Son, S.-W. et al., “How Does the Tibetan Plateau Dynamically Affect Downstream Monsoon Precipitation?” Geophysical Research Letters, Wiley/AGU, 2020.

Ye, D.-Z. and Wu, G.-X., “The role of the heat source of the Tibetan Plateau in the general circulation,” Meteorology and Atmospheric Physics, Springer, 1998, vol. 67(1–4):181–198. DOI: 10.1007/BF01277509.

India Meteorological Department, annual monsoon statistics; Government of India, National Water Policy; World Meteorological Organization, South Asian Climate Outlook Forum publications.

World Meteorological Organization, Third Pole climate and cryosphere reporting; World Economic Forum, “Third Pole” explainer, December 2024.

Council on Foreign Relations, “Water Clouds on the Tibetan Plateau,” backgrounder.

Tibet Policy Institute, “Tibet: The Water Tower of Asia.”

Goldstein, Melvyn C., A History of Modern Tibet, 1913–1951: The Demise of the Lamaist State, University of California Press, 1989.

Shakya, Tsering, The Dragon in the Land of Snows: A History of Modern Tibet Since 1947, Pimlico, 1999.

Chellaney, Brahma, Water: Asia’s New Battleground, Georgetown University Press, 2011.

Mekong River Commission, official records on member and dialogue partner states.

Springer Nature, “Beyond the Border: Exploring the Complex Dynamics of Water Tensions Between India and China,” Discover Global Society, 2024. DOI: 10.1007/s44282-024-00089-x.

Eyler, Brian, and Weatherby, Courtney, “New Evidence: How China Turned Off the Tap on the Mekong River,” Eyes on Earth / Stimson Center, April 13, 2020. (Remote-sensing analysis confirming Chinese dam impoundment during 2019 downstream drought; reported by Foreign Policy, “Science Shows Chinese Dams Are Devastating the Mekong,” April 22, 2020, and Mongabay, “China held water back from drought-stricken Mekong countries, report says,” April 2020.)

International Crisis Group, “Dammed in the Mekong: Averting an Environmental Catastrophe,” Report No. 343, October 2024.

Business Standard, “China refuses to resume Brahmaputra data sharing suspended over Doklam row,” September 12, 2017.

War on the Rocks, “Did China Use Water as a Weapon in the Doklam Standoff?,” October 2017.

Dialogue Earth, “China resumes sharing Brahmaputra water flow data with India.”

Yale Environment 360, “China’s Mega Dam Project Poses Big Risks for Asia’s Grand Canyon”; Al Jazeera reporting on Yarlung Tsangpo dam; East Asia Forum, “India and China in deep water over Himalayan hydropower,” December 15, 2025.

Made in China Journal: “Infrastructure and State-Building: China’s Ambitions for the Lower Yarlung Tsangpo Project,” August 7, 2025; “The Technopolitics of China’s Yarlung Tsangpo Dam Project and the Paradox of Hydropower,” September 4, 2025.

International Crisis Group, “Thin Ice in the Himalayas: Handling the India-China Border Dispute,” Report No. 334, November 2023.

Indian government announcement, June 16, 2020 (20 Indian soldiers killed, Galwan Valley); Chinese government confirmation, February 2021 (4 PLA soldiers killed).

Chatham House, “How China–India Relations Will Shape Asia and the Global Order: The Evolution of the Border Dispute,” April 2025.

The Diplomat, “Chinese Dam on Transboundary River Raises Concern in Delhi,” January 2025. (Eastern sector and McMahon Line proximity.)

East Asia Forum, “India and China in deep water over Himalayan hydropower,” December 15, 2025; International Water Power, “India’s CEA outlines transmission master plan for 76GW of hydro and pumped storage in Brahmaputra Basin.”

UN Treaty Collection, treaties.un.org, verified May 12, 2026.

Indus Waters Treaty (1960), World Bank-brokered. Standard international water law record.