We are committed to helping developers, researchers, and enterprise teams build successfully with WeatherNext models and datasets.
Support channels
1. Data access & allowlisting
For questions regarding access requests, project allowlisting status, or access permissions across BigQuery, Earth Engine, and Google Cloud Storage:
- Submit or check the status of your request using the WeatherNext Data Request Form.
- Requests are processed on a rolling basis within 5-7 business days.
- For urgent onboarding or account issues, contact: weathernext@google.com.
2. Operational service health & delivery delays
WeatherNext operational datasets target scheduled publication times on Google Cloud Storage, BigQuery, and Earth Engine (see the Dissemination Schedule).
- Target publication times have an expected variance of ±15 minutes.
- If data delivery exceeds +60 minutes past the estimated window, send
email to weathernext@google.com with the
subject line
[Delivery Delay] {Init Timestamp UTC}.
3. Technical questions & modeling issues
For dataset schema questions, data formatting issues, or unexpected model output:
- Email: weathernext@google.com
- Please include the following details in your email to help us troubleshoot
quickly:
- Surface: Google Cloud Storage (Zarr), BigQuery, Earth Engine, or Model Garden.
- Forecast run: Model initialization timestamp in UTC (for example,
2026-08-09 00:00:00 UTC). - Variables and coordinates: Variable name, pressure level, and latitude and longitude bounds.
- Reproduction: Minimal code snippet (Python or SQL) and full error trace.
4. Enterprise Google Cloud support
If your organization uses WeatherNext using managed inference on Google Cloud (Gemini Enterprise Agent Platform or Vertex AI) and has an active Google Cloud Support contract:
- Open a support case directly in the Google Cloud Console Support Center.
- Select your project and reference Vertex AI or Gemini Enterprise Agent Platform as the component.
5. Open source & research models
For issues, contributions, or questions regarding open-source research models (WeatherNext 2-mini, GraphCast, GenCast weights and Colab notebooks):
- File an issue or discussion on the google-deepmind/weathernext GitHub Repository.