Back to the map

Credits, data and methods

Everything shown on the map is a measurement or a cited statement. This page lists every dataset, tool and reference used, and how the numbers were made.

NASA data

DatasetDetails
NISAR L2 GCOV (geocoded polarimetric covariance), provisional releaseNASA/JPL and ISRO NISAR mission. Collection C2854338529-ASF, processing version (CRID) P05023. Distributed by the Alaska Satellite Facility DAAC. Layers used: HHHH and HVHV backscatter (gamma-0) and the validity mask, L-band frequency A. Subsets requested through NASA Harmony on 14 Nov 2026. NISAR follows NASA’s standard data use and citation guidance.
Scenes, track 98 ascending, frame 1620 Jun, 2 Jul, 14 Jul, 26 Jul, 19 Aug, 31 Aug, 12 Sep, 24 Sep, 6 Oct and 18 Oct 2026 (same radar mode on every date).
Scenes, track 48 descending, frames 74 and 7516 Aug and 28 Aug 2026.
Scenes, track 149 descending, frame 7423 Aug and 4 Sep 2026.
NASA services and documentsNASA Harmony (subsetting), NASA Common Metadata Repository (granule search), ASF NISAR Data User Guide, NISAR L2 GCOV Product Specification JPL D-102274 Rev F.
NASADEM Merged DEM Global 1 arc second, version 1NASA LP DAAC, collection C2763264762-LPCLOUD, tiles n27e085 and n28e085, measured February 2000. Used for downhill routes, low ground near rivers and glacier height bands. Retrieved 14 Nov 2026.
High Mountain Asia Near-Global Multi-Decadal Glacial Lake Inventory, version 1 (HMA_GLI)NASA NSIDC DAAC, collection C3249539102-NSIDC_CPRD, period 2015 to 2018. 12 lake outlines inside the study box. Retrieved 14 Nov 2026.

Space agency partner data

NISAR is a joint mission of NASA and the Indian Space Research Organisation (ISRO).

Other data

DatasetDetails
OpenStreetMapWaterway lines, place and peak points. © OpenStreetMap contributors, Open Database License 1.0. Retrieved through the Overpass API on 14 Nov 2026.
BasemapOpenFreeMap “dark” style, © OpenMapTiles, data © OpenStreetMap contributors.
TerrainMapzen Terrain Tiles (Terrarium), hosted on AWS Open Data. Used for hill shading and the 3D view, drawn at true vertical scale.

References for the event

Method

  1. For each scene, NASA Harmony clips the three layers to the study area (85.10–85.65°E, 27.85–28.42°N, in two boxes) and returns GeoTIFFs in UTM zone 45N at 10 m (20 m for one radar mode).
  2. Pairs are consecutive passes on the same track, 12 days apart. The later scene is put on the earlier scene’s grid.
  3. Pixels flagged invalid in either scene’s mask are dropped. Backscatter power is averaged in a 5 × 5 pixel window, then change = 10 · log10(after / before), in decibels.
  4. River paths are traced downstream along OpenStreetMap waterway lines. Pixels within 100 m of a path are grouped into 500 m bins by distance downstream. Per bin we report the share of pixels whose change exceeds ±3 dB.
  5. The same statistic is computed for every other 12-day step on track 98 (June to October 2026) and for four neighbouring rivers, as a baseline for ordinary change.
  6. Map tiles are the change rasters resampled to Web Mercator and coloured: blue for a fall, red for a rise, transparent below 1.5 dB, fully opaque from 4 dB.

The step from 26 Jul to 19 Aug is not used: it spans a mission-wide data gap and is 24 days long.

What this is not

This is a hackathon prototype. Radar change shows where the surface changed between two passes. It cannot tell what was damaged. The data are provisional releases. Nothing here is an official assessment, a forecast or a warning.

Software

Python with rasterio, NumPy, Pillow and harmony-py; Node.js for web requests; MapLibre GL JS 4.7.1 (BSD-3-Clause); fonts Archivo and Source Serif 4 (SIL Open Font License) from Google Fonts; hosted on Cloudflare Workers static assets.

SI use

See the SI disclosure.

Flow directions and height above the nearest river were computed with the open-source library pysheds 0.5 (GPL-3.0). Full method: data and methods.