IceNet Dashboard deployment repo

This meta-repository coordinates a modular, containerised icenet dashboard application composed of:

• icenet-dashboard-preprocessor – An installable Python package that converts netCDF files to Cloud-Optimised GeoTIFFs (COGs) for use by the tiler.
• icenet-tiler-api – A TiTiler + FastAPI-based backend that serves map tiles and static data.
• icenet-dashboard – A Plotly Dash web app for visualising IceNet forecasts.

---

Repo Structure

meta-repo/
├── icenet-dashboard/ # Plotly Dash web-app served with Gunicorn
├── icenet-tiler-api/ # Tile and data server using FastAPI + Uvicorn
├── icenet-dashboard-preprocessor/ # Data transformer: NetCDF → COGs
├── data/ # Output COGs and STAC catalog generated by running `icenet-dashboard-preprocessor` on IceNet prediction netCDF files
├── compose.yaml # Orchestrates all containers
└── Makefile # Top-level automation commands

This is the basic workflow:

---

Getting Started

1. Install Docker

Ensure Docker is installed on your system.

---

2. Install and Run IceNet netCDF -> CoG Preprocessor

pip install -e icenet-dashboard-preprocessor/

icenet_dashboard_preprocess <path_to_icenet_netcdf_predictions>

Example usage:

To point to a directory with netCDF files:

icenet_dashboard_preprocess -i results/predict/

Using wildcards:

icenet_dashboard_preprocess -i raw_data/*.nc

This will create a data/ directory with the JSON catalog and CoG outputs.

data/
├──  cogs/
│   ├──  north/
│   └──  south/
└──  stac/
    ├──  north/
    ├──  south/
    └──  catalog.json

3. Set up configuration ENV file (WIP)

Each service can be configured via environment variables. They are orchestrated via compose.yaml for consistency across environments.

Run the following script which will output a minimal .env file which includes the ports the components should be deployed one, and the IP address that the COGs should be accessible from.

./generate_env.sh

There are quite a few duplications (e.g. hostname, ports) to be taken care of since originally each component of the stack was run independently.

4. Build & Launch the Full Stack

make up

This will:

• Build all service images (icenet-dashboard, icenet-tiler-api)
• Launch the dashboard and tile server stack via docker-compose

5. Access the UI Interfaces and Docs

• Dashboard UI: http://localhost
• Tiler API: http://localhost:8000
• Data Server: http://localhost:8002