Uttarakhand and the 1% Flood
Two states in one — a broad, shallow Terai plain in the south and a violent, uncertain Himalaya in the north
In the early hours of 17 June 2013, after days of extreme monsoon rain, the Chorabari Lake (Gandhi Sarovar) above Kedarnath overtopped and breached, sending a wall of water, boulders and debris down the Mandakini valley. The temple survived; the town around it did not. Across Rudraprayag and Chamoli districts, the Government of Uttarakhand later listed more than 5,700 people as “presumed dead” — an administrative estimate, not a body count, reached when thousands of missing pilgrims and tourists could not be found after weeks of searching. Roughly 300,000 people were trapped in the valleys, and over 110,000 were evacuated by the armed forces. It remains India’s worst natural disaster since the 2004 tsunami. Less than eight years later, on 7 February 2021, a rock-and-ice avalanche of some 27 million cubic metres peeled off Ronti Peak in Chamoli, turned into a debris flood down the Rishiganga and Dhauliganga, and swept away the Rishiganga hydro project and the under-construction Tapovan dam — killing or leaving missing about 200 people, many of them workers trapped in a tunnel.
Neither of those was an ordinary flood. And that is the essential fact about Uttarakhand: it is really two states stacked on top of each other, and they flood in ways that share almost nothing. This piece uses the India Flood Atlas (a modeled 1% annual-chance flood layer) to show the scale, explain why the two regimes behave so differently, and lay out a realistic, phased path to resilience — while being honest that in the mountains, the exact numbers are the least certain the atlas produces.
The scale
The 2013 and 2021 disasters are observed history. The atlas adds the modeled 1% annual-chance (100-year) flood — the severe event that has a 1% chance of arriving in any given year, mapped consistently across the state.
Across Uttarakhand, that event reaches:
- ~499,000 building footprints — 14.6% of the state’s 3.41 million buildings — sit where the modeled 100-year flood would arrive.
- The mean flood depth over those exposed buildings is 6.55 m — the highest of any atlas state by a wide margin. But that single number is deeply misleading, and it is worth pausing on why. It is pulled up almost entirely by the steep mountain valleys, where the terrain model is least reliable and where relatively few buildings sit. The flooding in the plains — where most exposed buildings actually are — is shallow, on the order of half a metre. Of the exposed buildings statewide, 52% face water under 1 m, 15% between 1 and 3 m, and 33% deeper than 3 m (about 37% over 2 m) — a bimodal split that mirrors the two-regime geography exactly.
- ~10,822 km of road (about 7.6% of the state network) and ~206 km of railway (about 8.1% of the track) lie in the modeled floodplain — a lower share than the big plains states, because most of Uttarakhand’s mountain terrain sits high above any river.
The defining fact here is not a share or a depth — it is that the two halves of the state answer to different physics. In the south, the flat Terai plains of Udham Singh Nagar and Haridwar sit on the broad Ganga–Sharda floodplain, where a big flood spreads wide and shallow over a densely built landscape. In the north, the steep Himalaya — Uttarkashi, Chamoli, Bageshwar, Pithoragarh, Tehri — floods not by spreading but by charging: cloudbursts, flash floods and glacial-lake outbursts that stack water into incised valleys and race downstream at destructive speed. For the mountains, read the atlas for extent, not depth.
Where it concentrates
Risk splits cleanly between the plains, which hold most of the exposed buildings, and the mountains, which show the extreme depths. The most-exposed districts by share of their buildings in the floodplain:
| District | % of buildings exposed | Exposed buildings | Mean depth | Regime |
|---|---|---|---|---|
| Udham Singh Nagar | 30.1% | 142,018 | 0.46 m | Terai plains |
| Haridwar | 26.8% | 119,693 | 0.83 m | Terai plains |
| Uttarkashi | 15.6% | 23,266 | 17.36 m | Himalaya |
| Bageshwar | 13.9% | 15,371 | 16.65 m | Himalaya |
| Chamoli | 11.2% | 20,530 | 22.14 m | Himalaya (Kedarnath/Rishiganga zone) |
| Dehradun | 10.0% | 51,978 | 4.77 m | Foothills |
| Tehri Garhwal | 9.4% | 21,720 | 17.42 m | Himalaya |
| Pithoragarh | 9.1% | 16,812 | 22.92 m | Himalaya |
Read that table as two lists interleaved. The top two rows are a different world from the rest. Udham Singh Nagar (30.1%, 142,018 buildings) and Haridwar (26.8%, 119,693) are flat Terai districts on the Ganga and Sharda floodplains, and together they hold roughly 262,000 exposed buildings — about half the state’s entire exposed total — at mean depths of just 0.46 m and 0.83 m. This is the real centre of gravity of Uttarakhand’s building exposure: broad, shallow, plains flooding over a densely populated landscape. It is dangerous because it is widespread and populous, not because it is deep.
The mountain districts — Chamoli (22.14 m), Pithoragarh (22.92 m), Tehri (17.42 m), Uttarkashi (17.36 m), Bageshwar (16.65 m) — carry far fewer buildings but show the staggering modeled depths that lift the state mean to 6.55 m. Chamoli is the Kedarnath and Rishiganga zone, the epicentre of both marquee disasters. Dehradun sits between the worlds: foothill terrain, a mid-range 4.77 m, and the largest exposed count in the hills at nearly 52,000 buildings, threaded by the Rispana and Bindal and their cloudburst-prone tributaries.
Why the deepest numbers deserve a caveat
Those mountain depths — Pithoragarh’s 22.92 m, Chamoli’s 22.14 m, Tehri’s 17.42 m — deserve to be read very carefully, because they are the least certain figures in the entire atlas.
They point in a direction that is unquestionably real. The Himalayan valleys do flood violently: in 2013 the Mandakini surge scoured Kedarnath, and in 2021 the Rishiganga debris flood ran up the valley walls to over 200 metres above the floor. Water genuinely stacks tens of metres deep when a cloudburst or an avalanche-dammed pulse funnels into a narrow, incised gorge. The mountains are not modeled as deep by mistake.
But the exact metres cannot be trusted. The atlas is built on a roughly 30 m digital elevation model, and DEM accuracy is at its very weakest in exactly this terrain — steep, deeply incised valleys where a single 30 m grid cell can span a large vertical drop, and where a modeled water surface translates into wildly uncertain depths. So treat Chamoli’s 22 m and Pithoragarh’s 23 m as “catastrophically deep, direction reliable, exact figure not.” They are best read as a flag that says these valleys flood violently and need extent-based hazard mapping and evacuation planning — not as a claim that a specific building will see 22 metres of water. This is why, for the mountains, the honest reading of the atlas is extent first, depth a distant second — and why the shallow, well-constrained plains figures (Udham Singh Nagar’s 0.46 m, Haridwar’s 0.83 m), sitting on flat ground the DEM handles well, are the ones to lean on hardest.
Why Uttarakhand floods
Two regimes mean two sets of mechanisms, stacked on radically different ground:
- The Terai plains: broad, shallow river flooding. In the south, the Ganga and the Sharda (Kali) and their tributaries spread across the flat, densely settled Terai. Here a big flood behaves like the plains states — it sheets out wide and shallow over Udham Singh Nagar and Haridwar, submerging huge numbers of buildings to a modest depth. This is where most of the state’s exposed buildings live, and it is a manageable, well-understood hazard.
- Cloudbursts in the steep hills. Above the foothills, intense, highly localised cloudbursts dump extraordinary rain onto near-vertical, landslide-prone slopes. Rivers fill and flash within minutes, and saturated hillsides fail. The August 2022 Sarkhet cloudburst near Dehradun, and the deadly 2010 monsoon across Almora, Bageshwar and Nainital (around 200 monsoon-season deaths, compiled from news reports), are the recurring pattern.
- Glacial-lake outburst floods (GLOFs). As the climate warms, glaciers retreat and leave moraine-dammed lakes that can breach catastrophically. A Sentinel-2 inventory catalogued over 400 glacial lakes larger than 1,000 m² in the state (counts vary with the size cutoff). The 2013 Chorabari Lake breach above Kedarnath is the marquee GLOF-type event — and the risk grows every year the glaciers shrink.
- Rock-and-ice avalanches — not always a “lake burst.” The 2021 Chamoli disaster was not a GLOF. Early “glacier burst” reports were corrected: a ~27-million-cubic-metre slab of rock and ice detached from Ronti Peak and transformed into a debris flood. It is a distinct, hard-to-predict mountain hazard, and worth naming precisely so it is not confused with lake-outburst risk.
- A young, fragile, seismically active range. The Himalaya is geologically young, steep and earthquake-prone, which makes its slopes inherently unstable and its rivers prone to sudden blockage and release. This is the backdrop against which every mountain flood plays out — and the reason human interventions matter so much.
- Hydropower and construction in the wrong places. After 2013, the government-appointed expert committee chaired by Dr Ravi Chopra (2014) concluded — as the majority finding, formally disputed by some members and developers — that existing and under-construction hydropower projects aggravated the disaster: muck dumping, blasting and tunnelling destabilised slopes and raised riverbeds, amplifying downstream damage. It recommended dropping at least 23 proposed projects and barring new hydro in the highest, most fragile zones. In 2021, the Tapovan and Rishiganga projects sat directly in the debris flood’s path. Road-cutting, riverbed encroachment and unregulated hill-town construction add to the same problem.
- Climate change is intensifying the extreme-rainfall bursts and accelerating the glacier retreat that drive both the cloudbursts and the GLOFs — pushing today’s rare event toward the more-frequent event of 2050 and 2100.
A strategic approach
Because Uttarakhand floods two ways, it needs two strategies sharing a common backbone of hazard mapping and early warning.
Pillar 1 — Know the risk, and warn early. Statewide hazard mapping (like this atlas) plus real-time rainfall, river-gauge and cloudburst nowcasting. In the plains, that means conventional river-gauge warning on the Ganga and Sharda. In the mountains, the highest-leverage life-saver is cloudburst and GLOF early warning — automated weather stations, lake-level monitoring on the most dangerous glacial lakes, and sirens with rehearsed evacuation, because in a flash flood the margin is measured in minutes, not hours. The USDMA (Uttarakhand State Disaster Management Authority) and IMD are the backbone; pre-positioned NDRF/SDRF teams are the last-mile complement.
Pillar 2 — Map and monitor the glacial lakes. Identify and instrument the potentially dangerous glacial lakes, monitor their levels, and where feasible lower the most hazardous ones — the single most direct way to reduce GLOF risk to downstream valleys.
Pillar 3 — Govern construction in the fragile zone. The cheapest flood control in the Himalaya is not building in the path of the next debris flow. Enforce riverbed and valley-floor setbacks, regulate hill-town and roadside construction, and apply the Chopra committee’s caution about hydropower and heavy blasting above the fragile paraglacial zone. In the Terai, enforce floodplain buffers along the Ganga and Sharda.
Pillar 4 — Protect the plains, where the buildings are. Udham Singh Nagar and Haridwar hold half the state’s exposed buildings at shallow depth — a classic, tractable plains problem. Embankment maintenance, drainage, raised plinths and flood-safe design across the exposed Terai protect the largest number of people for the least money.
Pillar 5 — Defend the mountain slopes and valleys. For the hills, the decisive levers are landslide-susceptibility mapping, slope stabilisation, and evacuation — because where a cloudburst meets a saturated Himalayan slope, no wall saves the valley below. Deep-water valley towns need raised critical assets and rehearsed evacuation far more than they need embankments.
Pillar 6 — Make buildings and communities resilient. For those already exposed — broadest in the Terai, most violent in the hills — flood-resilient materials, protected utilities and rehearsed evacuation, paired with flood-inclusive home insurance (covered under “STFI” perils in standard Indian home cover such as Bharat Griha Raksha, yet rarely held).
A realistic timeline
Uttarakhand has institutional footholds — a USDMA, an active SDRF, and the hard lessons of 2013 and 2021 — but resilience across two regimes is a 15–25 year programme, not a finished project.
Phase 0 — Foundation & quick wins (0–2 years). Low cost, high impact. Statewide hazard mapping and exposure prioritisation; cloudburst nowcasting and GLOF early warning with sirens and rehearsed evacuation for the highest-risk mountain valleys (the fastest life-saver); river-gauge warning and pre-monsoon drainage clearance in the Terai. These save lives in the next monsoon.
Phase 1 — Monitor & regulate (2–5 years). Instrument the most dangerous glacial lakes and lower the worst; enact and enforce valley-floor and floodplain setbacks; regulate hill-town and hydropower construction in the fragile zone; map and stabilise priority slopes; fund a permanent annual maintenance budget.
Phase 2 — Basin- and valley-scale resilience (5–10 years). Manage the Ganga–Sharda plains and each mountain valley on their own terms — retrofit exposed public assets (hospitals, schools, substations, the char dham pilgrim routes) in the deep-water valleys, and extend the model to every high-exposure district.
Phase 3 — Climate-proof & sustain (10–20+ years). Design to future climate — size drainage, dam rules, slope protection and evacuation systems for 2050/2100 rainfall, cloudburst intensity and accelerating glacier retreat (the SSP scenarios) — and lock in monitoring, maintenance and updating as permanent functions.
The honest bottom line: Uttarakhand is two states in one, and its single most important flood fact is that its worst deaths and its most exposed buildings are in different places. The Terai plains — Udham Singh Nagar and Haridwar — hold about half the state’s exposed buildings at barely half a metre of water: a broad, populous, tractable problem. The Himalaya holds the catastrophe — Kedarnath, Chamoli — where cloudbursts, GLOFs and rock-ice avalanches flood violently and where the atlas’s 17–23 m depths are its least certain numbers, reliable in direction but not in exact metres. Read the plains for depth and the mountains for extent; protect the plains at scale and warn-and-evacuate the mountains; and be honest that the deepest figures are also the ones the terrain hides best. That is what lets a limited budget protect the most people across a state that floods two utterly different ways.
Sources
- 2013 Kedarnath / North India floods — 2013 North India floods (Wikipedia) · Dialogue Earth: “A decade after the floods, is Kedarnath safer?” (the ~5,700 figure is “presumed dead,” an administrative estimate — the true toll is unknowable, as many bodies were never recovered)
- 2021 Chamoli / Rishiganga flood — 2021 Uttarakhand flood (Wikipedia) · Shugar et al., Science (2021): a rock-and-ice avalanche, not a GLOF (~200 dead/missing; ~83 bodies recovered — cite both)
- 2022 Uttarakhand cloudbursts — SANDRP: Uttarakhand cloudburst 2022
- Hydropower controversy & the Ravi Chopra expert committee (2014) — SANDRP: Report of the Expert Committee on the Uttarakhand flood disaster · The Print: 24 hydel projects stuck after the 2013 floods (majority expert finding, formally disputed)
- Tehri Dam — Tehri Dam (Wikipedia) (~260 m, India’s tallest; ~100,000 displaced — figure approximate)
- GLOF risk — Glacial-lake inventory of Uttarakhand, Natural Hazards (2025) (lake counts vary with size cutoff)
- Exposure statistics — India Flood Atlas (modeled 100-year / 1% annual-chance event), NEER.
Exposure figures are modeled estimates (~30 m) for a 1% annual-chance flood, intended for awareness and prioritisation — not parcel-level certainty, and least reliable in the steepest terrain.