अAmarnepalNepal Data
Glacial hazard · Reference

Nepal knows which lakes will kill people. It has been written down since 2020.

ICIMOD and UNDP identified and ranked 47 potentially dangerous glacial lakes six years before the Bhote Koshi flood. Twenty-five of them sit upstream, inside Tibet. Identification is the part Nepal did well. Instrumentation is the part that never followed.

ShareWhatsAppViberFacebookXReddit
The 2020 inventory

47 lakes, ranked, published, and largely unwatched

The joint ICIMOD–UNDP report was released on 7 September 2020. It did not merely count lakes: it ranked them by danger, which is exactly the input an instrumentation programme needs.

By country

25

Tibet Autonomous Region, China

Upstream of Nepal. Outside Nepali jurisdiction and outside its instrumentation.

21

Nepal

Inside the country. Two monitoring systems were reported operating in July 2025.

1

India

In the shared basins covered by the inventory.

By basin

42

Koshi basin

Nepal's largest river system, and the overwhelming concentration of risk.

3

Gandaki basin

Includes the Trishuli / Bhote Koshi system struck in 2025 and 2026.

2

Karnali basin

The far west.

Worth sitting with: the Bhote Koshi disasters of 2025 and 2026 both happened in the Gandaki system, which holds three of the 47 lakes. The Koshi basin holds forty-two.

The preparedness gap, drawn to scale

Nepal has identified 21 potentially dangerous glacial lakes. When the same river flooded in July 2025, 2 monitoring systems were working. The Bhotekoshi basin was not one of them.

2 of 21 identified dangerous glacial lakes were monitored.

Monitored (2) Not monitored (19)Grid shows 7 per row. Source: ICIMOD inventory; monitoring count reported July 2025.

Neither figure is the real indictment. Both floods came from outside Nepal, and no bilateral mechanism existed to tell Nepal either time.

The gap

A ranked list is a research output, not a warning system

Nepal did the hard scientific work, with international partners, and finished it six years early. What it did not do was put instruments on the lakes the report named, or build the protocol that would let a Chinese observation reach a Nepali duty officer in time to matter.

The distance between those two things is not scientific. It is budgetary and administrative. A telemetry station on a glacial lake is cheap relative to a single kilometre of the mountain road it protects, and it opens no ribbon.

Amarnepal analysis: the 21-versus-2 comparison is the headline, but it understates the problem. Even perfect Nepali instrumentation would not have caught either Bhote Koshi flood, because both began inside Tibet. More than half the ranked hazard sits where Nepal cannot install anything, which makes the cross-border protocol not one option among several but the only one that addresses the majority of the risk.

47

Potentially dangerous glacial lakes

Across the Koshi, Gandaki and Karnali basins: 25 in the Tibet Autonomous Region of China, 21 in Nepal, 1 in India. 42 of the 47 sit in the Koshi basin alone. Identified and ranked in a joint ICIMOD–UNDP inventory published on 7 September 2020, six years before this flood.

ICIMOD / UNDP, Inventory of glacial lakes (2020) ↗

2

Glacial-lake monitoring systems working in Nepal

Against the 21 dangerous lakes inside Nepal, as reported at the time of the July 2025 Bhotekoshi flood. The Bhotekoshi basin was not among the monitored two.

The Kathmandu Post, July 2025 ↗

560+ stations

DHM observation network

The Department of Hydrology and Meteorology runs roughly 280 rainfall stations, 100 climate stations and 180 hydrology stations, with 40-plus automatic stations feeding a web-based telemetry system. The network is real, and it is pointed at the wrong scale of watercourse for this hazard.

DHM / flood early warning literature ↗

25

Community flood early-warning sites

Across seven basins: Karnali, Babai, Narayani, Seti, Rapti, Bagmati and Koshi. These systems are built for rain-driven river flooding with hours of lead time, not for a debris-dam burst arriving in minutes down a mountain tributary.

Government of Nepal / DHM ↗
What would close the gap

The three that could exist within a year

Convert the 2025 agreement into a duty officer and a number

The August 2025 understanding with China needs to become a named 24/7 contact point on each side, an agreed trigger list (seismic signal above a threshold in the border ranges, sudden turbidity or level change on the Lhende and Bhote Koshi, satellite-detected mass movement), and a maximum notification time written down. Nothing about this is technically hard. It is a protocol, not a project.

Cost: Effectively nil. It is diplomatic and administrative work.

Instrument the 21 lakes that were listed in 2020

Nepal already knows which lakes matter, because ICIMOD and UNDP ranked them six years ago. Automatic level and temperature telemetry on those specific lakes, with satellite change-detection as the second layer, closes the identification-to-warning gap that this flood exploited.

Cost: A rounding error against the Rs 15bn of road and bridge damage from this one event.

Put sirens where the lead time is minutes

For a debris-dam burst, the warning cannot travel through a district administration. Settlements below a known cascade path need direct-trigger sirens and drilled evacuation routes to high ground, on the model used below dam spillways. Timure, Rasuwagadhi and Syabrubesi are the obvious first three.

Cost: Low, and the Melamchi and Chamoli precedents both argued for it.

All six failures, and all six fixes →Climate impacts on Nepal
Questions

Glacial lake outburst risk in Nepal

How many dangerous glacial lakes does Nepal have?+

A joint ICIMOD-UNDP inventory published on 7 September 2020 identified and ranked 47 potentially dangerous glacial lakes across the Koshi, Gandaki and Karnali river basins: 21 inside Nepal, 25 inside the Tibet Autonomous Region of China, and one in India. 42 of the 47 sit in the Koshi basin alone, making it by far the highest-risk system.

How many of those lakes are actually monitored?+

Two functioning glacial-lake monitoring systems were reported in Nepal at the time of the July 2025 Bhotekoshi flood, against the 21 dangerous lakes inside the country. The Bhotekoshi basin was not among the monitored ones, either before the 2025 flood or before the far deadlier 2026 flood on the same river.

Why does it matter that 25 of the lakes are inside Tibet?+

Because more than half the hazard is outside Nepali jurisdiction, which means no amount of domestic instrumentation can see it. Both the July 2025 and August 2026 Bhote Koshi disasters originated inside Tibet. Nepal and China agreed in late August 2025 to share cross-border glacial flood risk information, but twelve months later Nepal's Flood Forecasting Division still received no notification of the 2026 event, saying no formal mechanism was in place.

What is the difference between a GLOF and what happened in August 2026?+

A classic GLOF is the failure of a moraine or ice dam holding a standing glacial lake, which can exist for years and can therefore be monitored, as with the July 2025 supraglacial lake outburst. The August 2026 event was an ice-rock avalanche that created a temporary debris dam and burst it within minutes. Both send a destructive surge downstream, but only the first can be watched in advance.

Are glacial lake outburst floods becoming more common in Nepal?+

The physical drivers are strengthening. More than 70% of documented rock-ice avalanches in High Mountain Asia begin in probable permafrost zones, and 65% start where ground temperatures already exceed minus 1.5 degrees Celsius. As glaciers retreat they expose and stop buttressing rock faces, and permafrost at depth continues warming for centuries even under a 1.5 degree scenario. ICIMOD's disaster risk lead described the acceleration of such events in the Hindu Kush Himalaya as unprecedented.

Sources & data note

Lake counts are from the joint ICIMOD–UNDP inventory published 7 September 2020, which covers the Koshi, Gandaki and Karnali basins across Nepal, the Tibet Autonomous Region and India. The monitoring-system count is as reported at the time of the July 2025 Bhotekoshi flood; if it has since increased, we have not found a published source saying so.