Arunachal Pradesh and the 1% Flood
India's flood risk begins here — where the Brahmaputra enters the country and one lowland district holds the only depth you can trust
On 9 April 2000, a rock-and-debris avalanche crashed into the Yigong river in Tibet, building a natural dam roughly 90 metres deep across a 2.5 km² spread of valley floor. It held for two months. On 10–11 June it failed.
The resulting flood remains the largest outburst from a landslide-dammed lake documented in modern history. It reached Pasighat, where the Siang leaves the mountains for the Assam plains, at a peak discharge of about 44,200 cubic metres per second. It killed roughly 130 people in Arunachal Pradesh, left more than 50,000 homeless, and washed away most of the bridges in its path.
Nobody in Arunachal caused it, and nobody in Arunachal could have stopped it. That is the central fact about flood risk in this state: it is largely decided upstream, in another country.
The scale
- Of Arunachal Pradesh’s 731,487 building footprints, 168,578 — 23.0% — sit in the modeled 1% annual-chance floodplain. That is above the national figure of roughly 20%, and the third-highest in the Northeast.
- 3,250 km of road — 4.3% of the network — is exposed. Rail exposure is effectively nil: 3 km of 1,933 km (0.1%).
- The atlas returns a modeled mean depth of 14.01 m over exposed buildings. This is not a usable depth, and it is the second-most distorted figure in the entire atlas after Sikkim’s 29 m. It is an artifact, and the rest of this section explains why.
Arunachal is the largest Northeastern state by area and the most vertical. The Brahmaputra enters India here as the Siang, having already fallen thousands of metres through the Tsangpo gorge — the deepest canyon on Earth. Within a single 30 m elevation cell, the real ground can drop tens of metres from a valley wall to the river bed. When a flood model tries to fill that geometry, the arithmetic produces enormous “depths” that describe the terrain, not the water.
Where it concentrates
| District | % of buildings exposed | Exposed buildings | Modeled depth |
|---|---|---|---|
| Upper Subansiri | 53.3% | 16,463 | 34.47 m · artifact |
| Namsai | 37.2% | 21,928 | 0.82 m · reliable |
| East Kameng | 36.4% | 9,393 | 19.08 m · artifact |
| Shi Yomi | 34.9% | 2,997 | 11.26 m · artifact |
| West Kameng | 32.9% | 15,459 | 25.60 m · artifact |
| West Siang | 30.1% | 7,502 | 14.69 m · artifact |
| Changlang | 29.5% | 22,948 | 4.34 m · uncertain |
| Anjaw | 25.9% | 2,421 | 35.72 m · artifact |
Namsai is the district to read first. At 37.2% exposed with a modeled depth of 0.82 m, it is the only entry in this table that behaves like a normal floodplain — because it is one. Namsai sits in the lowland strip along the Assam border, on the plains of the Lohit and Noa-Dihing, and it shares Assam’s flat-valley hydrology. Its numbers are trustworthy, and they say that more than a third of its buildings are in shallow water.
Changlang (29.5%, 4.34 m) is a partial case — part lowland, part hill — and sits between the two regimes.
Every other district in the table reports depths between 11 m and 36 m, which no one should read as water. Anjaw’s 35.72 m and Upper Subansiri’s 34.47 m are the model describing gorges. What those districts’ percentages tell you is still useful: a third to a half of their buildings sit in the corridor a severe flood would follow, because in this terrain the only buildable flat ground is the valley floor the river claims.
Why Arunachal floods
- It is where the Brahmaputra arrives. The Yarlung Tsangpo enters India as the Siang, carrying the drainage of a vast Tibetan catchment. Arunachal receives that water before Assam does.
- Landslide-dam outburst floods. The 2000 Yigong event is the defining case: an upstream blockage forms, impounds an enormous volume, and releases it catastrophically. In actively eroding, seismically live mountains, this recurs.
- Extreme relief and rainfall. Arunachal falls from over 7,000 m to under 100 m. Steep, heavily rained-on catchments turn precipitation into fast, erosive torrents.
- Landslides, which frequently matter more than depth. Slope failure blocks rivers, severs the few roads, and kills. This atlas does not map it.
- Enormous sediment load. The Siang has run visibly black or muddy in the past — notably in 2017 — prompting concern about upstream construction and seismic disturbance in Tibet. Sediment raises beds downstream and worsens Assam’s flooding.
- Upstream infrastructure in Tibet. Chinese hydropower development on the Yarlung Tsangpo, including a very large project in the Medog/Motuo gorge region, has raised persistent Indian concern about both altered flows and the consequences of a failure or sudden release. There is no comprehensive water-sharing or real-time data-sharing treaty covering this river.
Reading the numbers honestly
This is the most important section for Arunachal, and it is worth being blunt.
What the atlas can tell you reliably:
- Where the flood corridor runs — along the Siang, Subansiri, Lohit, Kameng, Dibang and Noa-Dihing valleys. Extent does not depend on resolving exact depth.
- Namsai’s depths (0.82 m) and, with more caution, Changlang’s — the lowland districts where the terrain suits the model.
- The state-level extent figure of 23.0%, which is a real and significant number.
What it cannot:
- Any depth figure above about 5 m in this state. The 14.01 m state mean and the 11–36 m district values are terrain artifacts of a ~30 m DEM in near-vertical country. They should never be quoted as depths, and this atlas does not quote them as such.
- Landslides and outburst floods, which are Arunachal’s actual killers, and which a static 1%-annual-chance depth model does not represent.
On boundaries: Arunachal Pradesh is mapped here to the full official India extent, out to roughly 97.4°E, consistent with how India publishes its own maps. Boundaries shown are indicative; refer to the Survey of India for authoritative external boundaries.
District boundaries follow the 2011-census-derived set (25 districts), so the most recently created districts are folded into their parents.
A strategic approach
Pillar 1 — Get upstream data. The single highest-value intervention for Arunachal, and for Assam below it, is real-time hydrological and reservoir data from the Tibetan reach of the Yarlung Tsangpo. Hours of warning is the difference between the 2000 flood and a managed evacuation.
Pillar 2 — Monitor for landslide dams. Satellite monitoring of the upper catchments can detect a river blocked by a landslide while the lake is still filling. Yigong took two months to fail — an enormous warning window, if anyone is watching.
Pillar 3 — Treat Namsai as a conventional floodplain. It is the one district where standard tools apply: zoning, raised construction, drainage. Its 37.2% exposure in shallow water is a solvable problem.
Pillar 4 — Design the valley-floor infrastructure for scour and impact. In gorge terrain, bridges and roads fail to velocity and debris, not to depth. Arunachal’s road network is its lifeline and its most repeatedly destroyed asset.
Pillar 5 — Warn and evacuate rather than defend. Where a flood arrives as a wall of water down a gorge, embankments are irrelevant. Sirens, rehearsed routes to high ground, and resilient communications are what work.
The honest bottom line: Arunachal Pradesh carries 23.0% of its buildings in the modeled 1% floodplain, and India’s Brahmaputra flooding begins here. But this is the atlas’s hardest state to read: its 14.01 m mean depth is a DEM artifact, and only Namsai — 37.2% exposed at a credible 0.82 m — behaves like a floodplain the model can measure. Everywhere else, trust the extent and discard the depth. And note where the risk is actually decided: in June 2000, a landslide dam in Tibet released the largest such flood ever recorded, killed about 130 people here and left 50,000 homeless, and no gauge or embankment inside Arunachal had any say in it. For this state, flood resilience is mostly a question of what India knows about the water before it crosses the border.
Sources
- 2000 Yigong outburst flood — Academia.edu: Earthquake–landslide–flood nexus at the lower reaches of Yigong Tsangpo, Tibet · AGU Landslide Blog: The Yigong rock avalanche, Tibet · NCBI/PMC: Geomorphic effects of recurrent outburst superfloods in the Yigong River
- Siang floods and history — The Arunachal Times: Siang’s fury — unravelling the historic floods · The Better India: How a massive landslide turned Arunachal’s Siji river into a lake
- Sediment and upstream concerns — SANDRP: Muddy Siang is a sign of danger ahead · Scroll.in: China is not to blame for the black Brahmaputra — but people in Arunachal and Assam face imminent danger · The Wire Science: Earthquakes in Tibet are behind the darkening of Arunachal’s Siang river
- 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. Boundaries are indicative; refer to the Survey of India for authoritative external boundaries. Building footprints are structures, not people.