Know which glacial lakes above India's valleys are growing, and when to look closer.

On the night of 3 to 4 October 2023, about 14.7 million m³ of frozen moraine collapsed into South Lhonak lake in North Sikkim. Roughly 50 million m³ of water left the lake, and the flood did damage along 385 km of the Teesta (Eos; Sattar et al., Science, 2025). The front reached Chungthang, 62 km downstream, about 2 hours 17 minutes after the collapse.

In 2026, a suspected outburst from a lake in Tibet sent a flood down Nepal's Bhote Koshi and Trishuli on 26 August, and a flash flood without heavy rain hit the Budhi Gandaki on 11 October; both rivers reach Bihar as the Gandak. Sikkim now lists 40 high-hazard glacial lakes, 16 of them at the highest risk (PTI, 4 Sep 2026). Many of these lakes sit above 4,500 m, freeze for half the year, and are seen mainly from orbit.

• Measures open water at each lake in the latest clear Sentinel-2 scene, with a stated uncertainty, and shows the image with the detected shoreline.
• Judges growth from the cleanest evidence: the same season in 2019 to 2021 measured the same way, then the inventory trend, and flags any scene a person should look at instead of hiding it.
• Checks three triggers at every lake: earthquakes of magnitude 4 and above within 100 km in 30 days, heavy rain against IMD's categories, and warm spells that melt snow and ice.
• Combines them with each lake's documented hazard into a watch level (normal, watch or warning), labelled as a heuristic, and orders the board by a monitoring priority.
• For 20 lakes, lists towns downstream with river distance and an approximate flood-wave time, calibrated on the 2023 Teesta flood.

A Python pipeline reads two glacial lake inventories from Zenodo and keeps the watchlist lakes. For each lake it asks the Planetary Computer for recent Sentinel-2 scenes, reads a pixel window around the lake, and finds water with the normalised difference water index (green against near-infrared) at a threshold chosen for that scene. Scenes with cloud, missing data or ice over the lake are rejected; doubtful ones are flagged. USGS earthquakes and Open-Meteo forecasts at the lake's elevation give the triggers. Everything is written as small JSON files with the time each source was read, and the console reads only those files.

State disaster management authorities and district administrations decide which lakes need a field visit, a sensor or a drill, and when. CWC and IMD liaison officers get an independent second look at lakes between their own surveys. Programme teams for glacial lake risk reduction track which lakes are changing. Valley communities and their volunteers learn how much time the towns below a lake might have.

• Advisory only, on every page: Sarovara flags conditions; warnings come from CWC, NDMA, IMD, the state and district authorities, and in Nepal from DHM and NDRRMA.
• Every number carries its source and the time it was read. Doubtful measurements are shown with a flag, never hidden or silently corrected.
• Lakes in the Tibet Autonomous Region are shown only as natural hazards to the valleys below, in neutral language.
• Downstream lists name towns and villages only. Sarovara does not list dams, bridges or other infrastructure, or their vulnerabilities.
• No data about people: lakes, weather, earthquakes and places only.

• Sarovara does not detect or forecast outbursts. Collapses, ice avalanches and slow moraine failure leave little in earthquake or weather feeds, so a lake can be at normal on the day it fails.
• From October many lakes freeze, and monsoon cloud hides them in summer; the latest area can be up to 90 days old or fall back to the inventory.
• Thin grey ice and very turbid water can be missed as water; large cases are flagged for review.
• Model weather over steep terrain carries large errors; the rain thresholds are IMD station categories applied to model values.
• Flood-wave times come from two wave speeds and straight-line distances, not a hydraulic model. Use them for awareness and drills, never for evacuation timing.

Pilot: one state's glacial lakes through a melt season, weekly runs, with your field team's observations fed back to tune the flags and thresholds. contact@trinetra.in