Nepal landslide

A landslide is not the rain. The rain is the trigger; the slope is the disaster. Every hillside in the Himalaya and the Western Ghats holds loose material at an angle it can just about keep, and it keeps it until something changes — water in the pores, the toe cut away for a road, the trees that held the top thirty centimetres taken off, a quarry face a kilometre uphill. Then it goes in seconds, at night, on the people living below it, and it goes on going: the initial failure at Wayanad in 2024 became channelised debris flows that travelled eight kilometres. India has mapped where this can happen. What it has not done is count it.

12.6% 0.42 million km², excluding snow-covered area — Of India's land area, prone to landslide hazard. The Geological Survey of India's own figure, and the basis of its National Landslide Susceptibility Mapping programme. It is an area of EXPOSURE, not a forecast: it says where the ground can fail, not where it will. Geological Survey of India

0.18 million km² North East Himalaya, including the Darjeeling and Sikkim Himalaya — The largest exposed region. Then 0.14 million km² in the North West Himalaya — Uttarakhand, Himachal Pradesh, Jammu and Kashmir; 0.09 million km² in the Western Ghats and Konkan hills across Tamil Nadu, Kerala, Karnataka, Goa and Maharashtra; and 0.01 million km² in the Eastern Ghats around Araku in Andhra Pradesh. Geological Survey of India

~80,000 1998–2022 — Landslides mapped across India. NRSC/ISRO's inventory, in three kinds — seasonal, event-based and route-wise. IT IS A MAPPED TOTAL AND NOT AN ANNUAL RATE, and the difference matters: the pan-India seasonal layer covers the 2014 and 2017 rainy seasons only, so the record is dense where somebody looked and thin where nobody did. National Remote Sensing Centre, ISRO

147 in 17 states and 2 union territories — Districts assessed for landslide risk. National Remote Sensing Centre, ISRO

Rudraprayag Uttarakhand — The district with the highest landslide density in India. It also carries the highest exposure of population and of houses of the 147 districts assessed. Tehri Garhwal is second. The atlas's socioeconomic risk ranking is, in the words of the landslide scientist who reviewed it, "quite experimental" — the ordering is a reading, the exposure is not. National Remote Sensing Centre, ISRO, Dave Petley, The Landslide Blog (AGU)

351 2024 — Recorded landslide deaths in India. 239 in 2023, a rise of 46.9 per cent — the second largest increase of any named cause in that table. THIS IS A FLOOR, NOT AN ESTIMATE: a death enters it only if a reporting authority attributed it to landslide and filed it under that cause. National Crime Records Bureau, Ministry of Home Affairs

4.4% 2024 — Landslide's share of all recorded deaths from forces of nature. For comparison, in the same table "causes other than above" is 1,626 deaths — 20.6 per cent of the total, with no named cause at all. Any reading of a cause-by-cause table has to carry that unattributed fifth alongside it. National Crime Records Bureau, Ministry of Home Affairs

392 with a further 150 people missing, as of 6 August 2024 — Killed by one landslide, at Mundakkai and Chooralmala in Wayanad. ★ READ THIS AGAINST THE NATIONAL ROW ABOVE. NCRB recorded 351 landslide deaths in the whole of India for 2024. One landslide in that year killed more. The two numbers are not both wrong; they are answers to different questions, and the gap is what the register does to a disaster — see `not_counted`. Dave Petley, The Landslide Blog (Eos, AGU)

570 mm — Rain in the two days before the Wayanad slope failed. Dave Petley, The Landslide Blog (Eos, AGU)

8 km affecting 86,000 m², per ISRO's analysis — How far the Wayanad debris flows travelled. A LANDSLIDE'S REACH IS NOT THE SIZE OF ITS SCAR. The initial failure became channelised debris flows that ran down existing stream courses and through the villages — which is why an evacuation drawn around the visible slope is drawn in the wrong place. Dave Petley, The Landslide Blog (Eos, AGU)

48 at least; 15 of them inside Ecologically Sensitive Areas — Stone quarries in Wayanad district. Counted from satellite imagery by Down To Earth. Nine are inside forests in two ESA villages, Kidanganad and Noolpuzha. Down To Earth

9,993.7 km² reduced from 13,108 km² — Kerala's Ecologically Sensitive Area, after the state asked for it to be cut. The Union environment ministry accepted the request. More than a decade after the Kasturirangan report was accepted, none of the states sharing the Western Ghats has finalised its Ecologically Sensitive Areas at all. Down To Earth

Figures in circulation that do not hold up

• 'Landslides in India are 66.5% North West Himalaya, 18.8% North East Himalaya, 14.7% Western Ghats' — widely repeated of the NRSC atlas, but the figures do not appear on ISRO's own page for the atlas and the landslide scientist who reviewed it reports no such split. It would in any case be a share of what was RECORDED, from an inventory whose pan-India layer is two rainy seasons.
• '151 dead at Malin' — the final official toll is reported this way almost everywhere and is very probably right; it has not been traced to a primary return here, so the contemporaneous count of 56 confirmed and about 120 missing is what this pack states, with the 151 named beside it.
• 'Meppadi was one of 18 Ecologically Sensitive Localities named by the Gadgil panel in 2011' — repeated in several outlets after the 2024 landslide, and not traced to the panel report itself. The ESA facts this pack does carry are Down To Earth's, which are counted rather than recalled.
• A death toll for the 2023 Himachal Pradesh monsoon. Credible figures run from about 320 to about 428 depending on who is counting and over what window, and the state's own Memorandum of Damages is the official series. Nothing here picks one, and no figure for that season is stated as this page's own.
• An annual number of landslides in India. There is no national series that counts events per year: GSI maps susceptibility, NRSC maps a 25-year inventory, NCRB counts filed deaths. None of the three is an event rate, and adding them does not make one.

Rainfall-triggered slope failure

The rain gauge nearest the slope, and the antecedent fortnight. Usually available within a day; usually the first thing stated and the least contested.

Channelised debris flow

The runout, on satellite imagery before and after. Distinguishing a flow from a slide decides where the danger was, and it is often settled only once a clear pass is available.

The toe cut away

The cut face itself, and the date it was made. Visible on imagery and in project records, and rarely named in the first reporting.

Quarrying above the failure

Quarry locations against the failure's catchment. Countable from imagery; contested in permits.

What was taken off the slope, and what was put on it

Land-cover change over years, not days. This is the slowest cause to establish and the one most likely to be missing from an event page written in the first week.

The mechanism in general

Trigger

Water. Rain intense enough or prolonged enough to fill the pore space in the slope material, which removes the friction holding it together. There is no single national threshold: it depends on the material, the angle, the vegetation and how wet the slope already was, which is why forecasting a landslide is a different and much harder problem than forecasting the rain.

Conditions

Loose slope-forming material at a steep angle — weathered rock, old debris, laterite, tea and cardamom soils in the Ghats, glacial and fluvial deposits in the Himalaya. Susceptibility is a function of slope, geology, the material and the vegetation, and GSI maps it nationally at 1:50,000.

What science says

The exposure is mapped and the annual count is not. GSI has a susceptibility map of 0.42 million km²; NRSC has ~80,000 mapped landslides over 25 years; NCRB has a deaths row that a single event can exceed. India knows where its slopes can fail far better than it knows how often they do, or how many people they kill.

Human factors

What was done to the slope, and what was built below it. Road and rail widening cuts the toe out from under a hillside; quarrying loosens it from above; terracing and levelling change where water goes; clearing the top of a slope removes the root mat that holds its first half-metre. None of these is weather. Down To Earth counted at least 48 stone quarries in Wayanad district from satellite imagery, 15 of them inside Ecologically Sensitive Areas and nine inside forests in ESA villages. The Western Ghats Ecology Expert Panel reported in 2011; the High Level Working Group that replaced it reduced the protected extent; the environment ministry then accepted Kerala's request to cut its own ESA from 13,108 to 9,993.7 km². A decade on, not one of the states sharing the range has finalised its ESAs. That sequence is a record of decisions, not of rainfall.

The mechanism for this kind of event, not a finding about this one. Himalayan disasters are often reclassified once fieldwork is done.