Datasets Quick Reference

GEOGLOWS V2 Datasets

All GEOGLOWS V2 datasets are sponsored by the AWS Open Data Sponsorship Program. This makes our datasets publicly available and available at no cost to the consumer. You do not need an AWS account, credit card, or a user name and password to use this AWS service to access GEOGLOWS data. You can view our listing in the registry of open data and in the AWS Data Exchange.

• Registry of Open Data: https://registry.opendata.aws/geoglows-v2/

• AWS Data Exchange: https://aws.amazon.com/marketplace/pp/prodview-aboaljwcz64zs 

Model Configs, GIS Data, Metadata

The stream network, model configuration files, and many supporting metadata files:

• http://geoglows-v2.s3-website-us-west-2.amazonaws.com/

Additional code used in GEOGLOWS:

• TDX-Hydro post processing preparation for hydrological modeling: https://github.com/geoglows/tdxhydro-postprocessing 

• Forecast computation scripts: https://github.com/geoglows/geoglows_ecflow 

• Retrospective weekly update scripts: https://github.com/geoglows/retrospective-update 

• Version of RAPID (muskingum channel routing) software used: https://github.com/c-h-david/rapid/releases/tag/20240224 

Forecast Data and Summaries

GEOGLOWS daily river forecasts are calculated at the ECMWF Supercomputing facility in Bologna, Italy. The forecasts are archived there as well as on AWS in S3. In addition, The forecast data are available through a data service.

• http://geoglows-v2-forecasts.s3-website-us-west-2.amazonaws.com/ 

Retrospective Simulation and Return Periods (Historical Simulation)

GEOGLOWS restrospective (or historical simulation) data and return period estimates. The retrospective simulation is updated weekly. The retrospective simulation is based on the ERA5 dataset. New ERA5 data becomes available daily with approximately 1 week lag (minimum 5 days) from real time which means the least possible amount of lag is 1 week. Once a week, we use all the newly produced ERA5 data since the last simulation to extend the retrospective simulation as far as possible. You should notice the historical simulation alternates between 1-2 weeks of lag from real time. 

• http://geoglows-v2-retrospective.s3-website-us-west-2.amazonaws.com/