Meltpools in the Himalaya

The flood on the Bhote Kosi on Tuesday morning was a timely reminder about the risks Himalayan communities face from melting permafrost and glaciers, as well as the need for transboundary early warning.

In the spring of 2024, I visited the Kanchenjunga Conservation Area (KCA) again. While in the village of Kampuchen (4,145) I visited Nupu Sherpa, friend and owner of the Kampuchen White House Lodge, to share a new paper we had just published about a massive ice-debris avalanche that had occurred in 2022 in the upper Nupchu valley.

Although Nupu was interested in the results and implications of the avalanche, he was even more concerned about the recent and rapid development of several large, pro-glacial lakes upon the terminus of the Kanchenjunga glacier, an hour walk north of Kampuchen and across the seasonal yak pasture of Ramdang Kharka (4,375m). 

Like many high altitude, low gradient glaciers in the Himalaya, those in the Kanchenjunga region had shrunk to mere shadows of their former selves during the previous 125 years, leaving behind debris-covered remnant ice and small meltwater ponds. 

Nupu’s fear was that the lakes might soon coalesce to form a large and potentially dangerous glacial lake. Given a large enough trigger, such as a massive landslide or avalanche, a glacial lake outburst flood (GLOF) could result that could damage Kampuchen, Ghunsa, Phale, the Kangchenjunga Base Camp trail, and even beyond to the lower Tamor into India. 

Since I was on my way to the village of Lhonak (4,792m) I stopped to take a look at the new lakes. Sure enough, from a lookout point on the lateral moraine of the Kanchenjunga Glacier I could see as many as ten newly formed and growing meltwater ponds upon the terminal moraine region of the glacier. 

The prospect was reminiscent of early satellite images taken of Imja Glacier in 1962 by the French, showing a number of meltwater ponds that within 15 years had grown and merged into a glacial lake half a square kilometer in size and containing 45 million cubic meters of water. Ten years after that Imja had grown to more than a square kilometer in size, was at least 120 m deep, and contained an estimated, and growing, 90 million cubic meters of water. Would the same scenario play out for the lower Kanchenjunga Glacier?

To try and understand the situation better, glaciologist Sonam Wangchuk of ICIMOD put together a time series composite of lake development over time, from 2016 to 2024, where it was clear that the majority of the new lakes had only formed since 2019.

Sonam Rinzin of Newcastle University estimated the potential depth and volume of the future glacier lake using standard procedures based upon the surface slope of the Kanchenjunga Glacier combined with volume-area curves for similar glacial lakes, estimating that the future lake would have a volume of approximately 33 million cubic meters of water and depth of more than 116 m. Other modeling methods enabled the mapping of probable ice, debris, and/or rock avalanches into the lake that could trigger a GLOF. 

A third model was then used to simulate four different avalanche scenarios that could trigger a GLOF: from small (debris avalanche of 3 million cubic meters) to the worst case (15 million cubic meters) of fast-moving debris. Even under the small magnitude scenario, avalanche-triggered outburst floods would travel distances of almost 120km downstream, and under the worst-case scenario the floods would reach the Indo-Gangetic Plain and cross the border into India, damaging or destroying up to 90 buildings and 44 bridges.

The prospective triggers of a future GLOF, however, are not limited to a nearby rock avalanche. A sudden release of water from water-filled caves further up the glacier could trigger a surge wave capable of breaching the terminal moraine complex holding in the lake water.

Weakening permafrost at the highest altitudes could result in massive ice-debris avalanches, such as those believed to have triggered recent and catastrophic floods in the Seti River, Barun Khola, Chamoli, South Lhonak lake, and Tuesday’s Bhote Kosi flood in Rasuwa.

Weakening and eventual collapse of the terminal moraine complex through sub-surface piping could also result in an outburst flood. And a range of cascading processes currently unknown to scientists could trigger a flood as well.

But will a large and potentially dangerous glacial actually form at the Kanchenjunga glacier? There’s really no way to tell at this point, although the rapid development of lakes since 2016 is a strong indicator that a large lake is in the process of forming. And rather than wait and see, our current uncertainty does not diminish the need for a more proactive approach to the development of a large glacial lake at some point in the near future.

For example, the lower Kanchenjunga Glacier clearly needs to be regularly monitored over the coming years to keep track of glacial lake growth and potential hazards. This could be done by the Department of Hydrology and Meteorology (DHM), ICIMOD, or Kathmandu University with results regularly shared with local people and local governments. 

In fact, such a service could be developed into a program that is offered to other communities throughout highland Nepal who experience similar situations of glacial lake development, growth, and prospective danger, especially since most communities presently feel that they have no one to turn to for similar information.

Inexpensive, locally appropriate, user-friendly early warning systems (EWS) need to be developed for the Kanchenjunga region now, instead of waiting for a tragedy to happen.

While the different options and costs are being assessed, the installation of new cell towers in the upper Ghunsa Khola would assure that local residents have access to one of the most effective and proven EWS systems available: cell phones, combined with a lake level monitoring device that broadcasts data and connects to cell phone alarm systems. 

For example, hundreds of lives were saved when upstream witnesses called their family and friends downstream to warn them of the Seti River flood near Pokhara in 2014.

New zoning policies that prohibit the construction of lodges and other infrastructure in high-risk floodplain regions need to be developed now. Where the infrastructure already exists, villages should be encouraged to install gabion rock-filled wire cages along susceptible river channel or river interfaces to divert the flow of water during a flood event, a method used successfully in both highland and riverine Nepal.

Local and national hazard training and response programs need to be developed now.

Several examples of excellent programs developed by different organisations in the upper Khumbu region exist that provide excellent models, such as those tested by the US Agency for International Development (USAID) and ICIMOD between 2011 and 2015.

Glacial lake reduction 

Nepal was recently awarded a $36.1 million award by the Green Climate Fund (GCF) to directly confront many of hazards from glacial lake outburst floods jointly implemented by the United Nations Development Programme (UNDP) and the DHM building on past successes at Tso Rolpa and Imja Lakes. There are some lessons learned over the past decades of field work based in Nepal and Peru:

1. The DHM wants to use the project to reduce GLOF risks in Kosi and Gandaki basins and glacial lake reduction programs are planned for Thulagi, Lower Barun, Lumding Tsho, and Hongu 2 glacial lakes. 

However, the planned lowering depth for Tso Rolpa in 2000 was 20m, with the project ending once 3m lowering was achieved due to lack of funds. Imja lake, which scientists also determined needed to be lowered by at least 20m, was in fact lowered by 3m in 2016 ‘because that’s what they did at Tso Rolpa’. 

If glacial lake reduction programs are to be implemented under the GCF program they should be based upon good science, and lowered to an effective depth, otherwise the activity will be a waste of funds. Peru offers experience for glacial lake lowering.

2. Although large glacial lakes (more than 1sq km) will always command the most media and even scientific attention, it has become abundantly clear in recent years that even small glacial lakes can be just as deadly, given the correct sequence of cascading events, rainfall, snowmelt, and/or other triggers. 

The 2017 Barun flood and September 2024 Thame flood are two excellent examples. Methods for assessing the flood risk of glacial lakes will need to be revised accordingly to include many of these smaller glacial lakes as well, in addition to the 21 large lakes already considered to be dangerous.

3. Existing and expensive EWS might benefit by some consideration of cheaper, more local, and possibly more effective alternatives. In 2015, for example, villagers living below Tso Rolpa complained that they had no idea how the recently installed EWS worked, what it sounded like, and what to do in the event that it went off.

With improved cell service and real-time lake level monitoring apps, local people could check on the lake whenever they wished, as opposed to waiting for some kind of communication from Kathmandu.

4. The GCF grant has the opportunity to become the ‘go to’ place for communities throughout highland Nepal to report any of their concerns related to glacial lakes and/or other observed cryospheric processes. This service could facilitate local people’s ability to adapt to, mitigate, or prevent a range of climate change-related hazards largely unknown to previous generations.

Alton C Byers is a faculty research scientist at the Institute of Arctic and Alpine Research (INSTAAR) University of Colorado at Boulder and has visited the KCA including a six-month residency in Ghunsa supported by the Fulbright Nepal Scholar Program. This article is largely based on Evolution of a Potentially Dangerous Glacial Lake on the Kanchenjunga Glacier, Nepal, Predictive Flood Models, and Prospective Community Response. Water, 2025. (Byers AC, Rinzin S, Byers E, Wangchuk S.)