GribStream Blog
GDAS global analysis fields now available
GribStream now supports NOAA GDAS (Global Data Assimilation System): global analysis-first fields on a 0.25 deg grid with 0-9 hour lead times.
On February 8, 2026, we added NOAA GDAS (Global Data Assimilation System) to GribStream.
If GFS is your "what might happen" forecast stream, GDAS is your "what the atmosphere looked like" baseline near the analysis cycle.
What GDAS is
GDAS is NOAA/NCEP's global data-assimilation system. It blends observations (satellites, radiosondes, aircraft, surface stations, and more) with a model background to produce a physically consistent atmospheric state.
Those GDAS analyses are then used to initialize operational forecast systems, including GFS. That is why GDAS is so useful for verification and diagnostics: it sits close to the analysis pipeline used by the forecast models themselves.
Coverage on GribStream
- GribStream code:
gdas(model page) - Coverage: global
- Resolution: 0.25 deg (~28 km)
- Cycle times: 00/06/12/18 UTC
- Lead times: 0-9 hours
This is an analysis-first dataset with short horizons, not a long-range forecast feed.
What it is used for
GDAS is especially useful when you care about near-real-time atmospheric state and forecast quality:
- Forecast verification: compare GFS, GEFS, or AI model output against an assimilation baseline.
- Bias tracking: monitor where and when forecast systems drift from analysis.
- Event reconstruction: rebuild synoptic and mesoscale context around high-impact events.
- ML labels and QA: generate consistent "truth-like" targets for training and monitoring.
How to use it in practice
Use the same API patterns as other models:
/api/v2/gdas/runsfor single-cycle analysis/forecast slices/api/v2/gdas/timeseries(alias:/history) for best-available values over time
A common workflow is pairing GDAS with GFS or GEFS: use GDAS as the baseline state, then measure forecast error growth with lead time.