Visualization dashboard for the BSEC Weather Station Data; built using svelte as the app framework, maplibre-gl-js for the map, and observable plot for the graphs.

The data is served via hyparquet which allows efficient compression and partial file access. To prepare the data, first download the record from MSD-LIVE, then use the transform.ipynb notebook to transform into parquet files.

Ideally, the data could be served directly from GitHub Pages, but I've found that Pages gzips the files and thus prevents the incremental file access that hyparquet expects. So for now you'll see that I'm serving the raw data from a Google Cloud bucket. May be best in the future to provide access through a public or anonymously authenticatable MSD-LIVE AWS bucket instead.

Install dependencies with npm i. Run a local dev server with npm run dev. The dev server will serve the data from the local static/data/ directory; the production code will serve from the GCP bucket.

Commits to the main branch automatically trigger a build and deployment to GitHub Pages. The MSD-LIVE BSEC project landing page embeds the GitHub Page as an iframe, so deployed changes will be picked up immediately (or once the viewer's cache clears).