Jump to Content

Resources

Connect with us

Accessing our forecasts

We aim to develop and release transparent models, forecasts, and data.

  1. We release real-time forecasts on the Flood Hub for locations where we are confident in the accuracy and reliability of our models. The Flood Hub Virtual Gauges provide forecasts in other locations as well, and is intended for experienced researchers.
  2. As part of our contributions to advance AI/ML in hydrology, we actively contribute to the GitHub NeuralHydrology codebase, which contains very similar (almost identical) functionality as our operational model. Additionally, the gitHub repository is fully documented and contains several tutorials.
  3. Google Runoff Reanalysis & Reforecast (GRRR) dataset consists of hydrologic predictions by Google's SOTA hydrologic model, with full global coverage (based on the HydroBasins dataset) in daily resolution. The dataset is available under a CC-BY-4.0 license.
  4. Google Inundation History dataset consist of inundation model, providing historical data on how often each 128-meter pixel in the world has been wet between 1999 and 2020, derived from satellite imagery. The dataset is available under a CC-BY-4.0 license.
  5. You can also join our API access waitlist, please note that the API pilot program is limited at the moment.

Add your historical data

Caravan is an open source global streamflow dataset, where global streamflow data providers (e.g., hydromet agencies) contribute daily. Several countries are currently represented in this dataset, including (but not limited to) Australia, Brazil, Canada, Chile, Denmark, Mexico, the UK, the US, and much of central Europe. Moreover, the United Nation World Meteorological Organization-supported Global Runoff Data Center recently contributed streamflow data for 25 additional countries.

Caravan includes a Python codebase that extracts catchment attributes from HydroSheds and meteorological forcing data (from ECMWF ERA5-Land) that are associated with each watershed in the dataset. This results in a globally-consistent rainfall-runoff dataset that can be used for large-sample hydrology research, and to train AI-based streamflow prediction models. Each data uploaded is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) license, which allows users to share the material in any medium or format for any purpose, and build upon the material for any purpose).

Useful links

Real time forecasts

  • Flood Hub - global flood forecasts, updated daily
  • Flood Hub Help Center
  • API access - join the waitlist to access our real-time forecasts programmatically

Google datasets

Caravan links