Issue 045 - Natural hazards - Warning-time throughput
How much warning time did Nepalese communities have?
Reuters reported that on August 26, 2026, part of a glacier near Langtang Lirung collapsed, sending ice, rock, mud, and water down valleys near the Nepal-China border. The initial debris flow reportedly reached about 50 m/s and the collapse occurred roughly 20 km upstream of the Rasuwagadhi border area.
The problem
Estimate how much usable warning time a community about 20 km downstream could have had between the glacier collapse and the arrival of the destructive flood.
Then estimate whether an average person receiving that warning could realistically have reached safe high ground before the flood arrived.
Was the physical travel time of the flood long enough for an effective warning system to have made a major difference, or was the event moving so quickly that even immediate detection would have left little time to evacuate?
Because Fermi problems target an order of magnitude, I normally use no more than two significant digits and write most calculations in scientific notation; the Fermi reference explains both conventions.
Before checking sources
Matt's first pass
I assumed that the peak debris flow might have been 50 m/s, but I would use an average of 30 m/s.
In a perfect world, if the emergency had been understood immediately and communicated downstream instantaneously:
distance ~= 20 km ~= 2.0 x 10^4 m
average speed ~= 30 m/s
travel time ~= distance / speed
~= 2.0 x 10^4 m / 3.0 x 10^1 m/s
~= 6.7 x 10^2 s
~= 11 minutesLet's say the average person was about 500 m from a safe location, and they can move on foot at about 5 m/s:
evacuation time ~= 500 m / 5 m/s
~= 100 s
~= 1.5 to 2 minutesIn that idealized case, they would have enough warning.
In reality, recognition of the flood event would have taken some period of time, maybe 1 to 2 minutes. Receipt of that warning downstream and communication of the emergency might take another 2 to 5 minutes. Then people would need to gather essential belongings, children, or elderly family members and get moving toward safety, which might take another 3 to 6 minutes even under optimistic conditions.
If roads and walkways are crowded, people probably move slower than the ideal, maybe 2 to 3 m/s. That puts them in a safe location anywhere from 8.5 to 17 minutes after the flooding starts, which is not nearly a safe cushion of time.
Calibration Score
Matt's Calibration Score: 80 / 100
Higher is better: earn points for accurate pegs, sound models, correct math, and a result close to the sourced answer. The image shows percent full of it: 100 minus the Calibration Score.
Pegs: 20/30. The 30 m/s average flood-speed peg and 5 m/s evacuation-speed peg were optimistic, but still in the right rough neighborhood for bounding the problem.
Model: 30/30. Distance divided by speed, followed by subtracting warning-chain and evacuation delays, is the right structure.
Math: 10/10. The unit conversions and travel-time arithmetic were clean.
Result: 20/30. The ideal physical travel time and realistic no-cushion conclusion were within the right order of magnitude, even though the usable-warning assumptions were optimistic.
Grounding facts
USGS describes the event as a rapid slope failure involving a glacier that turned into a far-traveled flood and debris flow laden with boulders and other rubble. The mapped event ultimately traveled nearly 100 km, far beyond the 20 km prompt distance.
Nepali Times reported that the origin was close enough to the border area that the debris flow would have reached the Rasuwa checkpoint in only a matter of minutes, and that there was no warning this year.
The key distinction is physical warning time versus usable warning time. A 10-minute travel time can become almost no usable evacuation time once detection, communication, belief, gathering family members, and movement to high ground are included.
After checking sources
Check and recalibrate
The physical travel-time calculation is brutally short. If the flood front averaged the reported initial speed for the first 20 km:
fast case
~= 2.0 x 10^4 m / 5.0 x 10^1 m/s
~= 4.0 x 10^2 s
~= 7 minutesIf it slowed to Matt's 30 m/s average:
slower case
~= 2.0 x 10^4 m / 3.0 x 10^1 m/s
~= 6.7 x 10^2 s
~= 11 minutesAnd even if the average speed over that reach were only 20 m/s:
very slowed case
~= 2.0 x 10^4 m / 2.0 x 10^1 m/s
~= 1.0 x 10^3 s
~= 17 minutesSo a reasonable physical-arrival range for a community 20 km downstream is about 7 to 17 minutes, with a central Fermi answer near 10 minutes.
The evacuation part is much less forgiving than Matt's ideal 5 m/s pace. A healthy adult can run that fast briefly, but an average warning response needs to include children, older people, confusion, gathering family, rough terrain, mud, and bottlenecks. Using 1 to 2 m/s for a mixed group moving 500 m:
evacuation time
~= 500 m / 1 to 2 m/s
~= 250 to 500 s
~= 4 to 8 minutesNow subtract the warning chain:
usable time
~= 7 to 17 minutes physical travel
- 1 to 3 minutes detection/recognition
- 1 to 5 minutes communication
- 2 to 6 minutes human reaction
usable movement window
~= near zero to perhaps several minutesA good calibrated answer is that the physical travel time was probably minutes, not hours. A very fast automated warning system could have mattered, especially for people already near high ground and trained to move immediately. But a warning system with ordinary detection, communication, and human delay would leave little or no cushion for many residents.
Post-check reflection
Matt's reflection
Looks like some of my assumptions were pretty optimistic. A 5 to 6 m/s evacuation speed is an adult runner's speed rather than a reliable average pace for evacuees, and the flood front probably moved faster than my 30 m/s average assumption.
Ultimately my realistic scenario was probably quite close to the real-world situation.
This really helps frame the disaster and understand why so many people seemed to be caught either unaware or directly in the path of the flood.
Recommended memory peg
Remember warning time ~= distance / hazard speed, but usable warning is smaller: usable warning ~= travel time - detection delay - communication delay - reaction delay. For mixed-group evacuation on foot, use 1 to 2 m/s unless you are deliberately modeling healthy adults running.
Reader results
Bars show how submitted estimates sort into the answer choices from the gut-check prompt.