GribStream Blog
Query forecasts like a database
GribStream expanded into a searchable weather archive with expressions, time-travel queries, and richer demos.
On June 8, 2025, we shared a major product update: GribStream now behaves like a queryable weather database, not just a forecast API.
What that means
Instead of only fetching raw fields, you can compute, filter, and search for conditions directly in the request:
- As-of (time-travel) queries for proper backtesting.
- Server-side expressions to compute derived fields (heat index, wind speed, custom formulas).
- Condition-based filtering so you can return events (e.g., wind > 15 m/s, freezing rain risk) rather than every row.
- Consistent query shape across models like NBM, GFS, HRRR, and RAP.
Why it matters
This approach makes it practical to treat the forecast archive like a searchable dataset: you can scan for thresholds, build alert criteria, and generate clean training data for ML without heavy post-processing.