South Lhonak lake had pipes in it. For years, engineers had run siphons over its moraine wall to drain the water down and buy the valley a little safety. On the night of 3 October 2023, none of it mattered. A slab of frozen mountainside, about 14.7 million cubic metres of rock and ice, broke off and crashed into the lake. The water heaved back a wave some 20 metres high, tore through the loose wall of debris that had held it for decades, and sent roughly 50 million cubic metres down the Teesta.
By the time the surge reached Chungthang, 60 km below, it was deep enough to swallow the 1,200 MW Teesta-III dam, the largest in Sikkim. It did not stop there. The flood was tracked for hundreds of kilometres down the river, through Singtam and Rangpo and on towards Bangladesh. At least 55 people were killed, scores more were never found, and around 80,000 were hit across four districts.
We have since reconstructed that night almost minute by minute, in a study led by Ashim Sattar of IIT Bhubaneswar and published in Science in January 2025. The harder thing to explain is that we did not need to wait for it. South Lhonak had sat on every glaciologist’s watch-list for years. It had been studied, modelled, even fitted with siphons to lower it. Its swelling was visible from space. The disaster was not unforeseeable. It was unforewarned.
That gap, between what we can see and what we actually act on, is the real Himalayan hazard of this decade.
A glacial lake outburst flood, or GLOF, is one of the few catastrophes that announces itself in slow motion. As a glacier retreats, meltwater gathers behind the rubble the ice leaves, a natural dam often no stronger than a heap of gravel. The lake behind it grows, season upon season, until something, a rockfall, a heatwave, an avalanche into the water, tips it over. The trigger comes without warning. The loaded gun, however, sits in plain sight, filling visibly, often for a generation.
And we can watch it fill. The European Union’s Copernicus satellites photograph the whole Himalaya every few days, at 10-metre resolution, and give the pictures away. The archive runs back a decade. Anyone with a browser can replay a lake, year by year, and watch it spread. I know, because I have done it, and the first lake I looked at was Tsho Rolpa, in Nepal’s Rolwaling valley. In the 1950s, it was a pond. Today, it is about 1.6 square kilometres of meltwater, still creeping outward, with around 6,000 households, by ICIMOD’s count, living in its path downstream.
Here are the numbers that should trouble us. A 2020 assessment by ICIMOD and the UNDP identified 47 potentially dangerous glacial lakes across the Koshi, Gandaki and Karnali basins: 25 in Tibet, 21 in Nepal, one in India. Forty-two of those 47 sit in a single basin, the Koshi, whose waters cross the border and pour into Bihar. In July 2025, the Green Climate Fund approved a 36-million-dollar grant, 50 million with co-financing, for Nepal and UNDP to lower just four lakes over seven years: Thulagi, Lower Barun, Lumding and Hongu-2. Four, out of 21 in Nepal alone. The rest remain as they are, full, growing, and perched above valleys where people sleep.
This is not a distant, foreign problem for India to watch from across a fence. South Lhonak was ours, in Sikkim. And because 42 of the 47 dangerous lakes drain through the Koshi, a large outburst in eastern Nepal is, in good part, an Indian flood, arriving on the Bihar plains whether or not anyone there was told to expect it.
What failed in Sikkim was not knowledge. It was the last mile. The river gauges meant to warn Chungthang were among the first things the flood destroyed. The first alert that night came from an Indo-Tibetan Border Police post at Zanak, 8 km below the lake, not from any instrument. People further downstream learned what was coming when they heard it. For all our satellites and supercomputers, the one signal that mattered never reached the person who needed it, in time, and in a language they could read.
India has moved since on paper. In 2024 the National Disaster Management Authority launched a Rs 150-crore National GLOF Risk Mitigation Programme covering 195 high-risk lakes across six Himalayan states, and expeditions have reached about 40 of them. That is a real start. But the money buys assessment far more than it buys warning, and the warning chain still runs through too many hands: the Central Water Commission’s gauges, the state disaster authority’s sirens, the dam operator’s emergency plan, and the district officer who must decide, at two in the morning, whether to wake a town. In Sikkim, every link in that chain was either missing or washed away.
Closing that gap is not a moonshot. It needs three unglamorous things. First, treat the open satellite record as public-safety infrastructure: a dated, plain-language lake-watch, issued jointly by NDMA and the Central Water Commission that a district officer in Mangan or a clerk in Rasuwa can open on a phone, not a dataset only a specialist can read. Second, fund mitigation past the easy four, and tell the villages below the rest, honestly, why they are waiting. Third, and most of all for India, build a real transboundary warning compact with Nepal and Bhutan, with gauge data flowing between the Central Water Commission and Nepal’s Department of Hydrology and Meteorology in real time, because a lake that bursts in Solukhumbu is India’s emergency within hours.
None of this is beyond us, and much of it is nearly free. The imagery costs nothing. The rain and river models are public. The science is done and published. What is missing is the decision to treat seeing as the start of a duty, not the end of one.
The old alibi for disaster, that we never saw it coming, is expiring. For the first time, the growth of a killer lake above a sleeping town is there for anyone to see, for the price of nothing. South Lhonak has already shown us how the story ends.
So, when the next lake goes, and one of them will, we will not be able to say we could not have known. We could watch it fill for years, while we drained four of twenty-one. What, then, will we say to the village that was waiting?
Views expressed are the author’s own and don’t necessarily reflect those of Down To Earth