The Himalayas needs a cross-border disaster-warning system

After the Lhende river tore through Nepal’s Rasuwa district on August 26, nobody was entirely sure where the disaster had actually started for the first few days. Scientists eventually traced it to an ice-rock avalanche on the Nepali side of the border, which dammed the river. When that dam gave way, the resulting flood carried debris nearly 100 kilometres downstream.

In its wake, 19 motorable bridges were damaged, 40 kilometres of road washed out along with a dozen hydropower projects. By the time authorities finished counting losses, over 1,000 were confirmed dead, and more than 4,500 missing.

Scientists studying climate change have long warned that a warming world would make the Hindu Kush Himalaya more vulnerable to disasters such as avalanches, floods and landslides. We are now seeing those warnings unfold in front of us. At the same time, a new UN report suggests that temporary rises above the 1.5°C warming threshold are becoming increasingly likely, and may soon be the new normal. Such temperature overshoot can intensify the risks facing the region, making extreme events more frequent and leaving countries, like Nepal in this case, with less time to prepare.

Covering blind spots

That detail about where the avalanche happened is worth sitting with. Most post-disaster commentary in the Himalayas assumes the failure runs one way, an upstream country sitting on data a downstream one needs. Here, neither government was watching that particular slope. As borders have nothing to do with where a glacier or a rockface decides to move, a monitoring network built around that border inherits its blind spots on whichever side happens to be watched less.

Nepal had, in fact, already flagged the risk. At a meeting in Kathmandu on 27 May, three months before the flood, Nepali officials pressed their Chinese counterparts for something beyond rainfall forecasts: readings on glacier movement and water levels, updated every ten minutes, fast enough to get a public alert out before a flood wave hit rather than after. China's team said they had no mandate to sign anything on the spot and asked for more time. 

Until then, what Nepal received from China amounted to little more than heavy-rainfall forecasts. Nepal's arrangement with India already runs on a routine basis by comparison, with precipitation and river-level readings moving back and forth as a matter of course. The northern border has nothing close to that yet.

China's caution here is real enough. But zoom out across the Hindu Kush Himalaya, and the deeper problem becomes structural: no shared data architecture exists anywhere in the region, only a patchwork of bilateral arrangements of wildly different strength. 

Collaboration is key

Take the India-China Expert Level Mechanism, running in some form since 2006 and anchored since 2013 by an umbrella MoU with no expiry date, probably the most developed river data-sharing arrangement anywhere in the Himalayas. 

Even this stretches to just two rivers — Brahmaputra and Sutlej — drawing on four monitoring stations in Tibet, and only during flooding season, from May 15 to October 15. 

Data on rainfall, water level and discharge are transmitted twice daily. Data on glacier movement and the risk of outburst floods are not included within the ambit of the agreement. For the exchange of hydrological data, India pays ₹1crore annually. 

But the tap has not always stayed on: during the 2017 Doklam standoff, China suspended data transmission, citing technical upgrades at its stations, even while Bangladesh, further downstream on the same river, continued to receive its share. Seasonal, narrow, metered, and vulnerable to whatever else is happening between Delhi and Beijing that year: this is roughly what Nepal is now negotiating toward.

Bhutan doesn't even have that option. India and Bhutan have made real progress. Thimphu agreed this year to share real-time rainfall and river-discharge readings with West Bengal. Governments of both regions also held a  fresh round of high-level talks on hydropower and river management last month. China and Bhutan, though, have no diplomatic relations at all. Their boundary talks have reached only their 25th round, and those are still settling where the line itself runs, well before discussing what should flow across it. Two of the four countries that would need to join a shared Himalayan data system don't yet have an embassy in each other's capital.

Moving beyond countries

The gap isn't technical. The Kathmandu-based International Centre for Integrated Mountain Development (ICIMOD),  which counts all eight Hindu Kush Himalaya states as members, already runs satellite-based flood guidance under its SERVIR-HKH programme and does regional glacier assessments. It has even written the manual, literally, for community-based flood early warning, which Nepal, Bhutan and Pakistan have each piloted in scattered valleys. What's missing is authority. Nothing gives that science a standing route across borders, so every valley that floods has to wait on a fresh round of bilateral diplomacy first.

Getting all countries to sign one treaty on real-time data pooling isn't realistic soon, especially with Bhutan and China still without formal ties and India-China relations prone to sudden freezes. A more workable path probably would be building on existing bilateral relationships rather than waiting on one grand multilateral deal. This would imply advocating for glacier movement or actual river stage telemetry to be considered hazard data, as opposed to strategic data. The model would have to be similar to meteorological figures that routinely cross borders under World Meteorological Organization conventions regardless of whatever is transpiring between two governments. 

It may also mean empowering institutions with a real operating budget and the mandate to run as a routine clearinghouse, simultaneously feeding all the national agencies so that strained relations with one agency are not the chokepoint for the entire valley’s warning time.

But none of that is of any use if the warning does not get past the ministries. A data-sharing agreement signed in Beijing or Delhi is of no use to a herder in Rasuwa if she is not given enough warning to move. The flood waves from a breached ice dam might give as little as five to 30 minutes of warning. Transmitting data from a glacier in Tibet to a cellphone in a Nepali village in that short a time frame is the other half of this problem. None of this has actually been done yet.

Anjal Prakash is Professor of Public Policy at FLAME University. He contributes to various IPCC reports and is presently a Lead Author in IPCC’s special report on Climate Change and Cities. Views expressed are personal.