-
Notifications
You must be signed in to change notification settings - Fork 0
APIs and OSS
The National Oceanic and Atmospheric Administration (NOAA) provides various datasets related to climate, weather, and environmental data. In this project, we use the NOAA Global Historical Climatology Network Daily (GHCND) dataset to fetch temperature data from the nearest weather station.
To see a list of all NOAA stations: https://www.ncei.noaa.gov/pub/data/ghcn/daily/ghcnd-stations.txt
NCEI Data API Documentation: https://www.ncei.noaa.gov/support/access-data-service-api-user-documentation
NCEI Search (datasets) API Documentation: https://www.ncei.noaa.gov/support/access-search-service-api-user-documentation
NCEI Management (for metadata) API Documentation: https://www.ncei.noaa.gov/support/access-support-service
NCEI Order API Documentation: https://www.ncei.noaa.gov/support/access-order-service
- you'll need to buy a dataset w/ NOAA account that has the API key
- most datasets are free
NCEI GitHub: https://github.com/NCEI-NOAAGov
FCPG Tools: https://fcpgtools.readthedocs.io/en/latest/cookbook.html
- produce pre-computed girds of upstream basin characteristics
https://github.com/AIStream-Peelout/flow-forecast
- LSTM flow forecasting
- run pretrained model in inference mode: https://flow-forecast.readthedocs.io/en/latest/inference.html
- download the model with the model's ID using flow-forecast's pip package
- Find ID and model config file in the public wandb.io training run's
- download model as ONYX
- DA-RNN is easiest model to use (minimal features)
- model feature inputs: historical tempature, ....