Zero - Browser-based IFS visualization using ECMWF open data

Datasets and Usage Open data
1

Hello,

I’d like to share a project that demonstrates what becomes possible with ECMWF’s open data commitment: Zero, a browser-based weather globe that visualizes IFS forecast data directly in the browser.

Live demo: https://zero.hypatia.earth

Source: GitHub - hypatia-earth/zero: Serverless weather globe rendering ECMWF forecasts in browser using WebGPU

What it does:

  • Renders IFS data at native O1280 resolution (6.6M grid points) — no downsampling or regridding

  • Fetches data via HTTP Range requests from Open-Meteo’s S3 mirror (~500KB slices from 44MB files)

  • Runs entirely client-side using WebGPU — no backend infrastructure required

  • Works offline after first load, using Service Worker caching

Currently available layers:

  • Temperature (2m)

  • Wind (10m, animated flow lines)

  • Pressure (MSL, with isobars)

  • Earth basemap with day/night terminator

Why this matters:

The project aligns with ECMWF’s vision of making forecast data accessible to all. A serverless architecture means communities with limited infrastructure can deploy hazard awareness tools without maintaining servers or paying for commercial services. Open source means anyone can adapt it for their specific needs.

Technical approach:

The irregular O1280 Gaussian grid is sampled directly in fragment shaders using binary search and precomputed lookup tables. This avoids the quality loss and memory overhead of regridding to regular textures — preserving ECMWF’s native resolution where it matters most.

Feedback welcome:

I’d appreciate any feedback from this community:

  • Are there specific parameters or layers that would be most useful to add?

  • Suggestions for improving data attribution or documentation?

Thank you to ECMWF for the open data program, and to Open-Meteo for providing the S3 mirror that makes this architecture possible.

Best regards,

–Torsten

1 Like