Fluvius
Important
This documentation was generated on Mar 24, 2022, and is rebuilt on push events to the main branch.
Fluvius is a collaboration between the Analytics Lab at Conservation Science Partners, the Instituto Tecnologico Vale, and Microsoft Brazil. Fluvius uses satellite imagery and AI to monitor the health of rivers in the Amazon and the U.S. by enabling prediction of suspended sediment concentration. This website serves as documetation for the code in the project’s GitHub repository. The web app can be viewed here.
In the following sections, we provide an introduction to the project, and setup information.
Getting Started
The following command line scripts represent the building blocks of the fluvius
workflow. These scripts are meant to be run in the order specified by the
numeric prefix in the filenames. All of these files can be found in ‘bin/’, and
should be executed from the Fluvius repo’s root directory (e.g.
python3 bin/01-usgs-station-acquire.py)
Command Line Scripts
- 01-usgs-station-acquire.py
- 02-preprocess-data.py
- 03-image-join.py
- 04-data-merge.py
- 05-prep-qa-chip-dataset.py
- 06a-download-chips-for-qa.py
- 06b-upload-good-chips-list.py
- 07-remove-bad-obs.py
- 08-partition-data.py
- 09-MLP-grid-search.py
- 10-compile-grid-search-results.py
- 11-fit-top-model.py
- 12-prediction-inputs.py
- 13-predict-tabular.py
- 14-make-prediction-chips.py
- 15-prep-data-for-app.py
Internals