GLACE is a dataset of annual Sentinel-1 summer composites of the European Alps.
It was primarily made for glacier mapping. Debris-covered glaciers are hard to separate from the
rock around them for humans and machine learning models alike. Interferometric
Synthetic Aperture Radar (InSAR) can separate them, because glacier flow lowers the
interferometric coherence between repeated observations.
The dataset
For each summer since 2015, GLACE provides two
measurements, in VV and VH polarization, at 10 m grid spacing:
- Coherence (COH12): how similar the radar phase stays between
two acquisitions 12 days apart. Flowing ice usually has lower coherence than
stable terrain.
- Backscatter (RTC): the strength of the radar echo,
radiometrically corrected for terrain (gamma-nought).
Each measurement comes with two quality bands, an observation count (QA: Count)
and a composite quality map (QA: Quality).
How it was made
Sentinel-1 SLC bursts are processed to 12-day coherence and terrain-corrected
backscatter. Each year's composite uses all ascending and descending acquisitions
in the 90-day window with the least snow cover, measured by the MODIS snow index
(NDSI). The acquisitions are combined by local resolution weighting (LRW;
Small et al., 2022): at each pixel, acquisitions that resolve
the slope more finely receive a higher weight. This reduces shadow, layover and
foreshortening artifacts in steep terrain. GLACE is, to our knowledge, the first
to apply LRW to coherence.
Getting the data
The composites are cloud-optimized GeoTIFFs, available as a STAC catalog on
Source
Cooperative, licensed CC BY 4.0. You can browse the catalog with the
Portolan/STAC browser. This map shows lossy
8-bit previews, so use the GeoTIFFs for analysis.
About this map
This map visualizes the GLACE dataset. The data is streamed from Source Cooperative. All streaming,
decoding and rendering happens in your browser, with no application server
required. Click the map to read a value. Share a view by copying the page URL:
the address keeps the extent and selections.
Built with:
Their licenses are in
third-party-licenses.txt.
The color maps are Fabio Crameri's
Scientific colour maps and matplotlib's viridis, magma and
cividis. The viewer is open source under the MIT license.
Credits
Contains modified Copernicus Sentinel data
2015–2026. GLACE is produced at the
Department of Geography, University of Zurich. Credits for the basemaps, terrain
and glacier inventories are shown on the map and beside each inventory. A dataset
citation will be added with the preprint.
Feedback
Problems or suggestions? Report them in the
dataset's issue tracker or, for the web map, the
viewer's issue tracker.