Executive Summary

The Tibetan Plateau is the single most important source of fresh water in Asia. The Indus, Sutlej, Karnali, Gandaki, Koshi, Brahmaputra, Salween, Mekong, Yangtze and Yellow rivers all begin in its glaciers, snowfields, springs and high wetlands, and together they sustain well over a billion people across South, Southeast and East Asia. Since the People’s Republic of China took control of Tibet in 1950, this landscape has been treated less as a shared ecological commons and more as a frontier to be secured, engineered and extracted. Over the past two decades the pace of that transformation has accelerated sharply. Mainstream dams on the Yarlung Tsangpo, a mega hydropower project at the Great Bend in Medog, highways and railways driven through unstable valleys, fortified border settlements, expanding mining, the forced sedentarisation of nomads, and even large-scale attempts to modify the weather over the Plateau have all been pursued by Beijing with little transparency and almost no consultation with downstream countries.

This report examines those activities against the background of the catastrophic flood of 26 August 2026, which tore through the Lhende Khola, Bhote Koshi and Trishuli corridors on the Nepal-Tibet border. By the end of September, Nepal Police were reporting 1,452 confirmed dead and 5,132 people listed as missing, while on the Chinese side Xinhua has held Tibet’s toll at 43 dead with 519 missing since 6 September. Hydropower plants, the Rasuwagadhi and Gyirong border crossings, dozens of bridges and long stretches of highway were wiped out, and Nepal’s Finance Minister estimated reconstruction costs at US$4 to 5 billion, close to a tenth of the national economy.

The scientific consensus now places the trigger of the 2026 collapse on the Nepali flank of Langtang Lirung, and this report does not claim that any single Chinese project caused it. The disaster matters for a different reason. It came just thirteen months after the July 2025 flood that did originate in Tibet, from a supraglacial lake on the Purepu Glacier in Gyirong County, and it fits a long pattern of transboundary cryosphere disasters descending from the Plateau. More importantly, it exposed how little the downstream states know about what is happening upstream, how badly exposed critical infrastructure has been sited in the flood path, and how Beijing continues to treat hydrological and environmental information as a strategic asset rather than a humanitarian obligation.

The central argument of this report is that Chinese policy on the Plateau is steadily raising the ecological risk carried by Tibetans and by the people of Nepal, India, Bangladesh and Bhutan, while denying them the information and voice they need to protect themselves. The report closes with a set of practical recommendations for downstream governments, regional bodies and the wider international community.

1. Introduction: The Water Tower Under Management

The Plateau is often called the Third Pole because it holds the largest store of ice outside the polar regions. It is also warming faster than most of the planet. Assessments of the Hindu Kush Himalaya have found that glaciers in the region disappeared 65 percent faster in the 2010s than in the the preceding decade, and that on current emissions pathways as much as 80 percent of present glacier volume could be gone by the end of the century. Even if warming is held below 2°C, a loss of roughly a third to a half of the ice is projected. More recent observations show that ice loss rates across the region have roughly doubled since 2000.

Climate change is a global problem, and China alone cannot be blamed for it. But how a state manages a warming mountain system is very much a national choice. A government that recognised the fragility of the Plateau would slow down, map its hazards openly, share its data freely, and involve both local Tibetan communities and downstream neighbours in decisions about large projects. Beijing has done almost the opposite. It has treated the Plateau as an engine for national energy supply, a buffer for territorial security and a laboratory for top-down environmental engineering. Tibetans have had little say in any of this, and countries downstream have had even less.

The consequences of this approach are visible at three levels. On the Plateau itself, rivers, wetlands, grasslands and pastoral communities have been reshaped by dams, drainage, mining and resettlement. Along the Himalayan border, construction for trade and military access has pushed roads, ports and settlements into narrow valleys exposed to floods and slope failures. And downstream, in Nepal, India and Bangladesh, people living along glacier-fed rivers face growing hazards with less and less warning.

2. The August 2026 Flood: What Happened

At about 8:37 in the morning on 26 August 2026, a very large slab of fractured rock and glacier ice broke away from the north face of Langtang Lirung, a peak of over 7,200 metres close to the border. The mass fell roughly 1,200 metres into the valley of the Lhende Khola, the stream that marks the boundary between Nepal’s Rasuwa district and Gyirong County in Tibet. It shattered into a mixture of pulverised rock, ice and meltwater, picked up loose sediment from a channel that had already been stripped and widened by the previous year’s flood, and became a debris flow of extraordinary speed. The energy released was strong enough to register on seismometers in the United States and Germany as the equivalent of a magnitude 5 earthquake, and early reports did in fact describe it as a quake.

The surge smashed into the confluence at Rasuwagadhi and Gyirong Port, then roared down the Bhote Koshi and Trishuli through Timure, Syabrubesi and Betrawati. At Galchhi, about 100 kilometres downstream, the river rose around nine metres within half an hour. Bodies were later recovered as far away as Chitwan and Nawalparasi, which tells its own story about the reach of the flood. Nepal’s disaster authority classified two districts, Rasuwa and Nuwakot, as severely affected and put the number of people impacted at roughly 1.6 million.

The damage to infrastructure was immense. Around 40 kilometres of road in Rasuwa were completely destroyed, about 13 kilometres of the Prithvi Highway were buried in mud and debris, and 41 bridges were lost. Preliminary assessments identified damage to fourteen hydropower and solar projects with a combined capacity of about 748 MW—part of it operational and part still under construction— including Rasuwagadhi, Chilime, Upper Trishuli 1, Sanjen, Langtang Khola, Trishuli and Devighat. Hundreds of hydropower workers lost contact in the first days, and rescuers were still trying to reach men trapped in flooded tunnels days later. More than 500 vehicles were reported to be standing in the customs area at Rasuwagadhi when the flood struck. The victims included villagers, soldiers and police, customs and immigration staff, truck drivers and traders, and a large number of pilgrims travelling to or from Kailash Mansarovar, many of them Indian nationals.

The warning failure was near-total for the upper corridor. The head of Nepal’s disaster authority said people had not received even five minutes of lead time, adding that Nepal has no weather stations in its high mountain areas because of the lack of mechanisms for sharing data across the border. SMS alerts went out to hundreds of thousands of people further downstream only after Timure was already under water.

Scientists are careful about the trigger, and this report follows that caution. The US Geological Survey places the failure inside Langtang National Park, and ICIMOD has stated that although some early reports described the flood as originating in Tibet because the Lhende’s headwaters lie in Gyirong County, the trigger was on the Nepali side. The honest position, then, is that Beijing did not cause this particular collapse. But three facts remain inescapable. The corridor it struck had already been hit thirteen months earlier by a flood that did come out of Tibet. The most exposed infrastructure at the confluence, including Gyirong Port itself, had been built on exactly the flat ground where debris flows slow down and dump their load. And the disaster laid bare how thin cross-border monitoring and information exchange remain, years after the risks were known.

3. A Pattern, Not an Accident

The Bhote Koshi corridor has a long record of floods descending from Tibet. On 11 July 1981, the outburst of Cirenmaco (Zhangzangbo) lake north of the border devastated the valley, killing around 200 people, destroying about 27 kilometres of road and 12 bridges, damaging the Sunkoshi hydropower system and cutting trade between Nepal and Tibet. On 5 July 2016, another outburst originating in Tibet entered the Bhote Koshi. The source lake released only about 110,000 cubic metres of water, but by entraining sediment the flow grew to an estimated peak of around 2,400 cubic metres per second some 30 kilometres downstream, damaging houses, bridges, the Arniko Highway and the intake of the Upper Bhote Koshi hydropower project.

Then came 8 July 2025. A lake that had formed on the surface of the Purepu Glacier in Gyirong County, roughly 30 to 36 kilometres north of the border, drained suddenly and sent a flood down the Lhende and Bhote Koshi. It destroyed the Nepal-China Friendship Bridge at Rasuwagadhi, wrecked the customs yard, swept away vehicles, damaged hydropower plants and left dozens dead or missing. Trade through the crossing was interrupted for close to six months, and the closure contributed to a loss of more than Rs 22 billion in import revenue for Nepal. There was little useful local rainfall signal and no meaningful advance warning from the Tibetan side. Independent researchers had already flagged the Gyirong and Rasuwa watershed as the trans-Himalayan basin with the highest average glacial lake outburst risk along the Nepal-China border.

The same pattern is visible across the Plateau’s rim. In 2000, a landslide dam on the Yigong Tsangpo, a tributary of the Yarlung Tsangpo just above the great gorge, gave way and sent a flood peaking at around 125,000 cubic metres per second down into Arunachal Pradesh and Assam, where it killed dozens of people and left tens of thousands homeless. India received no meaningful warning. In the same year, a flash flood out of Tibet raced down the Sutlej into Himachal Pradesh, damaging roads, bridges and the under-construction Nathpa Jhakri project. In 2004 and 2005, the Parechu lake that formed in Tibet above the Sutlej became a source of acute anxiety in India, and information flowed only after sustained diplomatic pressure. In 2018, an ice and rock avalanche at Sedongpu blocked the Yarlung Tsangpo for more than two days. In 2016, two massive glacier collapses at Aru in western Tibet buried pastures and killed herders.

These should no longer be treated as a series of unrelated natural accidents. They form a consistent pattern of cryosphere hazards in steep, rapidly warming headwaters, and in most cases the hazard begins in territory under Chinese control while the worst of the damage falls downstream. That asymmetry is the core of the problem this report addresses.

4. The Engineering of Tibet’s Rivers

4.1 The Yarlung Tsangpo cascade

For decades, Beijing insisted that it had no plans to dam the mainstream of the Yarlung Tsangpo. Working through state-owned enterprises, it began doing precisely that in the early 2010s. The Gezhouba Group built the Zangmu Dam in the Gyatsa gorge between Lhokha and Nyingtri, and it began operating in 2015 as the first of eight dams planned for that stretch. By the end of 2023, Jiexu and Jiacha had also been completed, Dagu was under construction, and four more were on the drawing board. Across the Yarlung Tsangpo and its major tributaries, 27 dams have been built, with further projects planned. One study has estimated that five recently built dams on the river contributed to a reduction of about 8 to 11 percent in the mean annual discharge of the Brahmaputra into the Bay of Bengal. Much of the concrete for these structures comes from sand dredged out of the very rivers being blocked.

Beijing’s standard reply is that these are run-of-river projects that store little water and therefore cannot harm downstream users. That argument conveniently ignores what dams do to sediment, to fish migration, to water temperature, and to the timing of flows, all of which matter enormously to farming and fishing communities further down. It also ignores the fact that downstream states have no independent way of verifying how these structures are operated.

4.2 Medog: the largest dam on earth at the most dangerous bend

The most alarming project is the one now under construction at the Great Bend of the Yarlung Tsangpo in Medog County, the last Chinese-administered county before the river enters Arunachal Pradesh. Here the river falls from about 3,500 metres to about 500 metres within some 250 kilometres, cutting through the deepest gorge on earth between Namcha Barwa and Gyala Peri. The plan relies on diverting water through tunnels bored beneath Namcha Barwa, a mountain sacred to Tibetans, the Pemakopa and the Adi. Construction formally began in July 2025 with an announced investment of around 1.2 trillion yuan, making it the most expensive infrastructure project in the world, with a generating capacity expected to be roughly three times that of the Three Gorges Dam.

The site sits at the junction of tectonic plates in one of the most seismically active zones on the planet. The 1950 Assam and Tibet earthquake, one of the strongest ever recorded on land, struck this very region. Unlike the dams higher up on the Plateau, Medog lies at a lower altitude that receives heavy monsoon rainfall, so its operation could have a much greater influence on the river’s flow into India. The gorge is also prone to landslide dams and outburst floods, as the Yigong and Sedongpu events showed. A reservoir and tunnel complex placed in this setting raises the risk of cascading failure in a way that should worry every government downstream.

What makes this worse is the secrecy. India and Bangladesh, as downstream nations, still lack comprehensive access to the project’s geological assessments, seismic risk models, dam-break scenarios, emergency response plans, reservoir operating rules, environmental impact evaluations, sediment studies and real-time hydrological data. A project of this size and risk profile, built on a transboundary river without prior consultation, sets a dangerous precedent for the whole region.

4.3 The eastern rivers and the Dri Chu cascade

The same logic has been applied to the rivers of eastern Tibet. Under the Western Development programme, cascades were planned down the Dri Chu (upper Yangtze), the Yalong, the Dadu, and on the Mekong. The Dri Chu cascade, including Wudongde, Baihetan, Xiluodu and Xiangjiaba, is now the largest hydroelectric generating system in the world, producing about twice the power of the Three Gorges, with a further 21 smaller dams planned upstream. China’s eleven dams on the upper Mekong have already altered the river’s flood pulse and sediment load, and communities in Laos, Cambodia, Thailand and Vietnam have complained of low flows in drought years. Chinese engineers point to fish-friendly turbines, fish lifts and water outlet towers designed to reduce the shock of cold water releases, but there is no definitive evidence that these technical fixes work. Dams still fragment habitats and isolate populations of native fish.

Across China the state has built over 6,500 large dams and more than 95,000 small ones. Widespread and poorly coordinated development has caused extensive environmental damage everywhere, but the effects have been particularly severe on the Plateau, where geological, climatic and ecological systems are finely balanced and where local communities have almost no avenue to object. When Tibetans in Derge protested in early 2024 against the Kamtok dam on the Dri Chu, which threatened to submerge villages and historic monasteries, the response was mass detentions rather than dialogue.

5. Roads, Railways, Ports and Border Villages

The August 2026 disaster has drawn attention to a second category of risk: construction in narrow, glaciated border valleys. Over the last two decades, Beijing has pushed an enormous programme of highways, railways, airports, logistics hubs and fortified settlements toward its Himalayan frontier. The Qinghai-Tibet railway, the Lhasa-to-Nyingchi line, the Sichuan-Tibet railway, upgraded highways and hundreds of new so called xiaokang or well-off border villages have transformed the southern rim of the Plateau. Much of this is dual-use infrastructure serving military logistics as much as civilian needs. The broader effect has been a race by rival states to take possession of the mountains through roads, dams and garrisons, against a backdrop of melting ice that this very contest tends to accelerate.

Road cutting, tunnelling, quarrying and heavy traffic in steep valleys destabilise slopes, load unstable debris into river channels and place people and assets exactly where floods and landslides are most likely to strike. Serious analyses of the 2026 disaster have been careful to say there is no conclusive evidence that any particular Chinese project caused the glacier collapse, and this report accepts that. But the same analyses make the point that intense construction in fragile, high-altitude zones compounds natural vulnerability, and that development models designed for stable terrain are dangerously unsuited to rapidly warming mountains.

Gyirong is the clearest example. Beijing developed Gyirong Port as its main land gateway to Nepal and as the anchor of a planned railway to Kathmandu under the Belt and Road Initiative. The port, the bridges, the customs facilities and the truck parks were built on the flat ground at the confluence of the Lhende and the Kyirong channel. This is the first relatively level stretch of riverbed below a stream that drops from over 5,000 metres to around 1,800 metres, which is precisely the place where a debris flow slows, spreads and deposits its heaviest load. On both sides of the border, the most valuable assets were placed squarely in the flood path. The destruction of 2025 should have forced a rethink. Instead, a Chinese-built Bailey bridge restored traffic by January 2026, and the same ground was busy with vehicles and cargo when the 2026 surge arrived.

Within Nepal, a group of journalists went further and publicly accused Chinese construction around Gyirong town of deliberately altering the river and destabilising glaciated slopes. Such claims are not established by the evidence and should be treated with care. Their appearance, however, reflects a deep and growing distrust in Nepal about what is being built across the border and why so little is disclosed about it.

6. Grasslands, Nomads and the Politics of “Ecological Migration”

6.1 Moving people in the name of the environment

One of the most far-reaching Chinese interventions on the Plateau has been the sedentarisation of Tibetan nomads. Officially, this is carried out under the banner of shengtai yimin, or ecological migration, and tuimu huancao, which means retiring livestock to restore grasslands. In practice, herders have been pressed to sell or slaughter their animals, fence their pastures and move into concrete settlements along roads. Around 2.5 million people across China were resettled under the label of ecological migration between 1980 and 2010, and official plans have spoken of another 10 million considered in need of it.

The Sanjiangyuan region in Qinghai, at the headwaters of the Yangtze, Yellow and Mekong, shows how this works on the ground. Mandatory ecological migration began there in 2005, more than a decade before the national park plan was made public. About 90 percent of those resettled were ethnic Tibetans. The officially recorded population of Sanjiangyuan fell from 223,000 in 2007 to 64,600 in 2015, and the national park plan envisages further, unspecified reductions until a so-called steady state is reached in 2035. Sanjiangyuan, at 47,500 square miles, is larger than the other nine parks of the 2017 national park plan combined.

6.2 Bad social policy and bad ecology

The official justification is that grazing has degraded the grasslands. Yet, there is mounting evidence that removing herders is itself damaging to the environment, and that dividing pastures into fenced household plots produces worse ecological outcomes than traditional grazing across larger, shared ranges. The Plateau’s grasslands co-evolved with grazing animals over thousands of years, and grazing plays a role in buffering them against the effects of climate change. Many of the claims of success for these programmes rest on satellite imagery that is a poor guide to the actual quality and type of vegetation. Ecological migration thus combines social injustice with ineffective ecological policy, while presenting itself as climate adaptation.

The social costs are severe. Resettlement towns suffer from unemployment, lack of resources and poorly planned construction. Former nomads find it hard to enter wage-labour markets built around Han Chinese norms, and the loss of culture has been accompanied by disease, alcoholism and a breakdown of community life. Local township officials often privately recognise that the programme is ill-advised but have little political space to push back. Provincial officials, for their part, tend to see themselves as benevolent modernisers bringing new houses to supposedly backward herders, and dismiss Tibetan grievances as attempts to extract more compensation. The shared, though unstated, aim of these policies is the pacification of frontier regions and the assimilation of ethnic minorities.

Resettlement has also concentrated populations along roads in conspicuously built-up settlements, often presented as poverty alleviation and environmental protection. The spread of cement and asphalt across the Plateau, now tracked by Chinese planners themselves as a measure of urbanisation, is changing surface hydrology and local microclimates in ways that have barely been studied. Tibetan cities such as Xining have been rebuilt to match urban models imported from inner China with little room for local culture, a process that has been aptly described as a civilising machine.

6.3 Draining the wetlands

The Plateau’s wetlands act as giant sponges that store water, regulate flows and sustain river baseflow through the dry season. They have been badly damaged by state-driven development. The Ruoergai (Zoige) marshes at the headwaters of the Yellow River were drained through ditches and channels from the 1960s onward to expand livestock production. Over the past sixty years, the swamp has shrunk by more than 52 percent, and serious desertification has spread along the banks of the upper Yellow River and its tributaries. Chinese scientists working in the Yellow River source region have themselves acknowledged that many of its ecological problems have been caused by human activities such as dredging, drainage channels, highway construction and mining, which have dried out plateau marshes and riparian meadows and released large quantities of wind-blown sand. Roughly 2,500 square kilometres of desertified land now exists in the Yellow River source area alone. The irony is hard to miss: the same state that blames nomads for grassland degradation has itself drained the wetlands that held the grasslands together.

7. Mining and Extraction

The Plateau holds rich deposits of copper, gold, lithium, chromium, and other minerals, and extraction has grown rapidly alongside the new roads and railways. The Gyama (Jiama) polymetallic mine near Lhasa has been linked by local residents to contamination of the river and livestock deaths, and in March 2013 a massive landslide at the mine’s waste area buried 83 workers. In 2016, residents of Lhagang in Minyak saw dead fish and yak carcasses floating in the Lichu River downstream of a lithium mine and staged rare public protests. Similar complaints have surfaced around sacred mountains and rivers in many parts of Tibet, and they have routinely been met with detentions, surveillance and pressure on those who speak out.

Mining matters beyond its immediate footprint. Waste dumps and excavations in steep terrain add to the pool of loose material available to debris flows. Tailings and chemical runoff enter rivers that flow into other countries. And because independent environmental monitoring and journalism are effectively impossible in Tibet, neither Tibetans nor downstream populations can know the true scale of contamination. The global push for lithium and other critical minerals is likely to intensify pressure on the Plateau in the coming years, and there is little sign that Beijing intends to open these activities to scrutiny.

8. Engineering the Sky

Perhaps the most striking example of Beijing’s willingness to reshape the Plateau is the Tianhe or Sky River project. Proposed by Tsinghua University in 2016 and backed by the China Aerospace Science and Technology Corporation and the Qinghai government, it involves a network of solid fuel combustion chambers that release silver iodide particles into the atmosphere to seed clouds. Some 500 experimental chambers have been reported in Tibet, Qinghai and Xinjiang, with plans for tens of thousands more, guided by weather satellites monitoring winds over the Indian Ocean. The stated goal is to increase precipitation on a scale equal to about 7 percent of China’s water consumption.

The moisture this system aims to capture is carried north by the South Asian monsoon. Sceptics have pointed out that intercepting that moisture over Tibet could reduce rainfall further along its path, with potentially serious effects for South Asia. No downstream government has been consulted, no independent assessment has been published, and there is no international mechanism through which India, Nepal, Bhutan, or Bangladesh could raise objections. Altering the weather over the roof of the world is, by any reasonable standard, a matter of regional concern. Beijing treats it as a domestic engineering choice.

9. Information as a Strategic Asset

9.1 Data sharing on Beijing’s terms

If the physical transformation of the Plateau is one half of the problem, the withholding of information is the other. China has no water-sharing treaty with any of its South Asian neighbours. It is not a party to the UN Watercourses Convention, and neither are India or Bangladesh; China prefers bilateral arrangements in which its upstream position gives it the stronger hand. What exist instead are limited memoranda of understanding under which China supplies flood-season hydrological data from a handful of stations, historically for a fee, on the Brahmaputra and the Sutlej. These arrangements have repeatedly lapsed or been suspended when relations sour. During the Doklam standoff in 2017, the data flow to India broke down at the height of the monsoon, precisely when Assam needed it most for flood forecasting.

The situation remains unresolved. In August 2026, India was still pressing China to resume sharing water flow data, as concern mounted over Beijing’s construction of the world’s largest dam on the Yarlung Tsangpo, and New Delhi again called for an early meeting of the Expert Level Mechanism on trans-border rivers. Seasonal, partial, and politically conditional data sharing is not a serious answer to the hazards now emerging from the Plateau.

9.2 Nepal’s experience in 2025 and 2026

Nepal’s experience is even more telling. After the 2025 Purepu flood, it was clear that the Lhende valley posed a transboundary risk and that Nepal had no eyes on the Tibetan side. Nepalese officials sought environmental monitoring data from their Chinese counterparts, including on glacier related hazards, three months before the 2026 disaster, but never received it. When the flood came, it took the Chinese government 11 days to permit foreign media to report from the Gyirong border area. Beijing responded to criticism by insisting it was already sharing whatever data it had, and its state broadcaster argued that glaciers are hard to predict before they collapse. That is true as far as it goes, but it misses the point. Prediction of a single event may be impossible; systematic, open monitoring of known hazard zones and real time exchange of data are not.

Nepal and China signed an agreement in 2019 that referred to cooperation on disaster risk, and held ministerial level talks on an information sharing mechanism as recently as May 2026. None of this produced a working warning system on the Lhende. As one widely cited analysis of the disaster put it, timely information on the upstream environment should not be regarded as diplomatic generosity from China. For communities living beneath the Plateau, it is a matter of survival.

9.3 Opacity inside Tibet

The same opacity applies within Tibet. Independent scientists, journalists, and environmental organisations have almost no access. Local Tibetans who document pollution, protest against dams or mines, or simply share images online risk arrest. Chinese official environmental data is itself affected by the well-known pressure on officials to meet targets, which leads to falsification. The result is that the region most critical to Asia’s water security is also one of the least transparent places on earth.

10. The Downstream Stake

10.1 Nepal

No country illustrates the downstream dilemma more sharply than Nepal. Every major Nepali river system that crosses the Himalaya, from the Karnali to the Koshi, has headwaters in or near Tibet. Kathmandu sits below climatic and geological processes it cannot observe, let alone control. The 2026 flood killed more people than any natural disaster in Nepal since the 2015 earthquake, displaced whole communities, destroyed schools and health posts, and hit the country’s economic backbone. The hydropower sector, which Nepal sees as central to its development and its export ambitions, suffered damage to hundreds of megawatts of capacity. The Rasuwagadhi trade route, rebuilt after 2025, was wiped out again.

Nepal’s growing economic reliance on the Gyirong corridor also means that its trade, customs revenue, and supply chains are now tied to a border crossing that sits in one of the most hazardous valleys in the Himalaya. Beijing has promoted this corridor, and the proposed Kerung-to-Kathmandu railway, as a way of pulling Nepal closer. The events of 2025 and 2026 show what that dependence can cost. Nepal needs near-real-time access to upstream satellite observations, river gauges, weather stations, seismic records and glacier change data for every river that enters it from Tibet. It also needs to insist that Chinese infrastructure on the border be subject to joint hazard assessment before, not after, disaster strikes.

10.2 India

India’s exposure runs the full length of the Himalaya. In the west, the Sutlej and Indus carry water and floods from Tibet into Himachal Pradesh and Ladakh, and the 2000 Sutlej flood and the Parechu scare showed how little warning India gets. In the centre, the rivers of Nepal feed the Gandak, Ghaghara and Koshi, which cross into Uttar Pradesh and Bihar. The 2026 flood itself travelled into India’s Gandak two days after the collapse, much reduced by then but a reminder that a disaster born on the Plateau can reach the Gangetic plain. The large number of Indian pilgrims caught in the Rasuwa corridor adds a human and diplomatic dimension that New Delhi cannot ignore.

In the east, the Brahmaputra is the lifeline of Assam and Arunachal Pradesh. Chinese officials and some commentators point out that most of the Brahmaputra’s annual discharge is generated within India from monsoon rainfall and tributaries, and argue that upstream dams can therefore have only a limited effect. This misreads the risk. What matters for the Northeast is not only average annual volume but the timing of flows, the sediment that builds and renews the floodplain, the lean-season supply in winter and spring, and above all the possibility of sudden surges from landslide dams, outburst floods or the operation or failure of structures at Medog. The Adi communities along the Siang have already voiced fears that the Medog project could reduce flows and threaten their livelihoods. The concentration of Chinese dams and military installations just north of the McMahon Line gives the whole question an unmistakable security dimension. India’s own response, including the proposed Siang Upper Multipurpose Project intended partly as a counterweight, carries its own heavy ecological and social costs, and has met strong local resistance. That is the logic of an upstream arms race in concrete, and it is a race Beijing started.

10.3 Bangladesh

Bangladesh is the country most exposed to the cumulative effect of upstream decisions and the one with the least leverage. The Brahmaputra becomes the Jamuna as it enters Bangladesh, and together with the Ganga and Meghna it shapes the life of a deltaic nation of more than 170 million people. The river is central to agriculture, drinking water, and fisheries across the northern and central districts. Crucially, the sediment eroded from the Plateau and the Himalaya forms the world’s largest source-to-sink sediment system and has built much of Bangladesh’s land mass, along with a submarine fan extending deep into the Bay of Bengal. Any large-scale trapping of sediment behind dams upstream threatens the delta’s ability to keep pace with sea-level rise and erosion.

Bangladesh already contends with severe monsoon floods, dry-season shortages and saline intrusion in the south. A reduction in lean-season flows or a change in the flood pulse caused by upstream storage would compound all of these. At the same time, sudden releases or outburst floods from the Tibetan reaches could add to flood peaks in a country where millions live on low-lying char islands and riverbanks. China has shared some flood-season data with Bangladesh since 2008, but on the same limited, discretionary basis as with India. Dhaka has no seat at the table when Beijing decides what to build at Medog, and no treaty right to be consulted. Its concerns are real, and they deserve far more attention from the international community than they currently receive.

10.4 Bhutan and the wider region

Bhutan, though not the focus of this report, faces the same combination of glacial-lake hazards and dependence on hydropower. It also offers a useful lesson: its sensor and siren system in one valley shows that effective early warning is possible even in difficult terrain, but at a cost of millions of dollars per valley. Scaling such systems across thousands of catchments requires cooperation and data from all states sharing the mountains, which brings the question back to Beijing’s willingness to participate as an equal partner rather than as a gatekeeper.

11. Assessment

Taken together, the evidence points to a consistent model of governance on the Plateau. Beijing treats Tibet’s land, water and air as state property to be mobilised for national energy, security and political goals. It pursues these goals through large, centrally directed projects with little public scrutiny and little regard for the knowledge and wishes of Tibetan communities. It presents many of these projects in the language of green development, clean energy, and ecological civilisation, even when the practical effect is forced resettlement, drained wetlands, fragmented rivers and heavier construction in hazardous valleys. And it treats information about all of this as a strategic resource to be shared sparingly, on its own terms, and withdrawn when politically convenient.

This model does not create climate change, and it did not trigger the collapse on Langtang Lirung. What it does is make a warming mountain system more dangerous and less knowable. It increases exposure by pushing infrastructure and people into hazard zones. It weakens the natural buffers—such as wetlands, grasslands, and free-flowing rivers—that moderate floods and droughts. It adds new, poorly understood risks, from mega dams in seismic gorges to weather modification. And by withholding data, it strips downstream populations of the time and knowledge they need to protect themselves. The 2026 flood showed what happens when a fast moving mountain hazard meets a border region full of assets and people and empty of shared warning systems.

12. Recommendations

For Nepal, the immediate priority should be a binding Nepal-China hazard data protocol covering every river that enters Nepal from Tibet, with near-real-time exchange of satellite imagery, gauge readings, weather and seismic data and glacier and lake monitoring, and with automated alerts rather than phone calls after a disaster has begun. Nepal should also insist on a joint, independent multi-hazard assessment of the Rasuwagadhi and Gyirong corridor before any rebuilding, and of any future Chinese infrastructure along the border, including the proposed railway. Hydropower projects in glaciated catchments should not be approved without a full cascade hazard assessment, and every underground project should have independent emergency exits, protected communications and rehearsed evacuation drills.

For India, the lapsed hydrological data arrangements on the Brahmaputra and Sutlej should be pursued not as favours to be purchased but as obligations owed by an upstream state. New Delhi should press for year-round, real-time data rather than monsoon season snapshots, and for full disclosure of the geological, seismic, dam break and operating plans for Medog. India should also invest heavily in its own high-altitude monitoring along the Himalaya, including satellite based detection of unstable slopes and lakes in Tibetan catchments, and in siren based warning systems in vulnerable valleys in Arunachal Pradesh, Sikkim, Himachal Pradesh and Uttarakhand. Its own hydropower expansion in the Himalaya should be held to the same standards it demands of China.

For Bangladesh, the priority is to build a coalition with India and Nepal so that downstream concerns are raised jointly rather than one country at a time. Dhaka should seek formal assurances on sediment flows and lean-season releases from any Chinese storage on the Yarlung Tsangpo and should push for its own access to upstream data independent of India.

At the regional level, ICIMOD and similar bodies should be supported to build an open, shared inventory of dangerous glacial lakes and unstable rock and ice faces above populated areas and infrastructure, ranked by risk, across all Himalayan states including the Tibetan side. Downstream governments should jointly insist that the Tibetan Plateau and the Himalayan cryosphere be placed at the centre of adaptation and loss-and-damage discussions under the UNFCCC. A region that supplies water to a quarter of humanity has a strong claim on global attention and finance.

Internationally, governments and scientific bodies should press for independent access to Tibet for environmental research and monitoring, and should treat restrictions on such access as a matter of regional security, not merely of human rights. Weather modification over the Plateau should be subject to international notification and assessment, given its potential to affect monsoon rainfall in neighbouring countries.

Conclusion

The flood of 26 August 2026 will be remembered in Nepal as one of the worst disasters in its history. It should also be remembered across the region as a warning. The trigger lay on a mountain face on the Nepali side of the border, and no one should pretend otherwise. But the disaster unfolded in a corridor that had been hit only a year earlier by a flood from Tibet, beside a Chinese port built on the most exposed ground in the valley, in a borderland where Nepal had asked for upstream data and received none, and where the world was kept away from the Chinese side for days afterwards. These are not accidents of nature. They are the results of political choices.

The broader picture is clear. On the Tibetan Plateau, Beijing is damming rivers, draining wetlands, removing nomads from grasslands, expanding mines, cutting roads and railways into unstable valleys, fortifying its border, and even experimenting with the weather—all in a region that is warming faster than almost anywhere else and that feeds the great rivers of Asia. Tibetans have been denied a voice in these decisions, and the countries downstream have been denied both a voice and the information they need. As the climate crisis deepens, this combination of aggressive engineering and deliberate opacity will turn more mountain hazards into regional catastrophes.

Nepal, India, and Bangladesh share a common interest in changing this. Their populations bear the consequences of what happens on the Plateau, and their security increasingly depends on what they can see and know about it. The central lesson of Rasuwa is that environmental vulnerability in the Himalaya is a shared vulnerability. It can only be managed through openness, shared monitoring and genuine consultation. Until Beijing accepts that the rivers it controls at their source are not its property alone, the people living along them will continue to pay the price.