Himalayan study puts peak water at mid-century and 21 gauges on 40,000 glaciers

A study released this month and reported on Monday says the Hindu Kush-Himalaya is approaching peak water, the point at which river flows fed by melting ice stop rising and begin to fall, while only 21 of an estimated 40,000 glaciers in the range are monitored on the ground.

The paper was produced by Systemiq with the Integrated Mountain Initiative, the International Centre for Integrated Mountain Development in Kathmandu, and India's G.B. Pant National Institute of Himalayan Environment. Reuters summarised the findings from Hong Kong on 14 September. The authors say glacier mass loss has accelerated compared with a decade ago and that the mid-century turn in runoff will hit farms, cities and hydropower that treat today's swollen rivers as a permanent supply.

The timing of the report is not academic. In late August a glacier collapse on the Nepal-Tibet border sent ice and rock down 4,000 vertical feet into a tributary of the Bhote Koshi. Cascading landslides and flash floods followed. At least 1,300 people have been confirmed dead and more than 5,300 remain missing across Nepal and China's Tibet region. Reconstruction costs on the Nepal side have been discussed in a range that reaches $5 billion. India has already sent a first consignment of Bailey bridge parts to the flooded districts.

As ice retreats it leaves lakes held by loose moraine. The study says close to 200 glacial lakes in India are now classed as high risk, 56 of them very high risk, above valleys where millions of people live. A lake that replaces a glacier can drain in hours if a slope fails or if ice falls into the water. The August disaster did not need a lake on the Nepal side of the initial collapse. The next one might.

The monitoring gap is the number that should stay on a planner's desk. Forty thousand glaciers, 21 with instruments on the ice. Satellite data can track area and some mass change. It cannot replace stakes, weather stations and lake-level gauges when a district officer has to decide whether to evacuate a town. ICIMOD has said for years that the region is data-poor. This paper puts a hard fraction on that complaint.

Peak water is a date, not a slogan. Separate modelling of High Mountain Asia runoff, published in the open literature, has put the Himalayan turn as early as the 2020s in some sub-ranges under lower-emissions paths and in the 2050s under higher ones. Hengduan Shan has already been projected to peak first. West Kunlun, by contrast, may not peak until late century. The Systemiq study's mid-century framing is a regional average. Local rivers will not wait for the average.

Two billion people drink or grow food from rivers that start in this range, a figure the authors and subsequent coverage use as the outer bound of exposure. The inner bound is smaller and more useful: the towns immediately below the 56 Indian lakes marked very high risk, the hydropower intakes on the Koshi and the Sutlej, and the irrigation canals that were built for a rising hydrograph and will have to run on a falling one.

What the paper does not do is assign a single year when a given Indian district should expect less water. That work still sits with the 21 instrumented glaciers and with the state disaster authorities who have the lake lists. The practical next step is to put gauges on more ice and to treat the 56 lakes as engineering sites, not as scenery.

Nepal's August flood will be studied as a slope failure and as a climate event. The Systemiq report is an attempt to say the next failure is already visible in the lake inventory and in the runoff math. Peak water, once it arrives on a given river, does not reverse. The only choice left is how much notice the people below the ice receive.