Identification of phytoplankton blooms from observational data.
The repo contains two parts:
- Data preprocessing: The modelled data is used to generate observations for training. Essentially these are nutrients point values and density profiles data. The modelled data comes from ROMS+NERSEM models.
- Training: JAX/FLAX based routines.
Install and activate a python environment, for example:
python3 -m venv venv/blooms-ml
source venv/blooms-ml/bin/activate
Then install blooms-ml:
git clone https://github.com/limash/blooms-ml
cd blooms-ml
pip install -e .[ml-cpu]