grib-map — desktop AROME weather map

For the slides you can find them here: ./slides/slides.pdf.

Overview

A reference implementation of the lab in arome-tauri-weather-lab-student.md: a Tauri + Rust + Leaflet desktop app that opens a Météo-France AROME GRIB2 file, decodes the 2 m air temperature in Rust, and draws it as a coloured grid over Montpellier.

JavaScript frontend            Rust backend
  - file dialog     invoke()     - read the file
  - Leaflet map   ───────────►   - decode GRIB2 (2 m temperature)
  - draw cells    ◄───────────   - return [{lat, lon, value}]

Course checkpoints (git tags)

Each part of the lab is a working, tagged checkpoint:

Tag	Part	What works
part1	1	Tauri + Vite vanilla app scaffolds, frontend builds, backend compiles
part2	2	A .grib2 file is available (data/, fetch script + synthetic sample)
part3	3	describe_grib + load_temperature decode 2 m temperature (unit-tested)
part4	4	Leaflet map fills with coloured temperature cells from a chosen file
part5	5	Polish: fit-to-data, min/max/avg, opacity slider, colour legend

git checkout part3 (etc.) to see the project at that stage.

Prerequisites

Rust (rustup) + Node.js (LTS), plus the system libraries for the webview and the GRIB decoders:

• webview / build: webkit2gtk-4.1, base-devel, openssl, librsvg, cmake, clang, pkgconf
• GRIB decoding: libaec

Météo-France open-data AROME files (what data/fetch_arome_sp1.sh downloads) use CCSDS/AEC packing (data-representation template 5.42), decoded via libaec — so the grib crate is built with the ccsds-unpack-with-libaec feature. (The handout mentions JPEG2000/openjpeg; that path isn't needed for the open-data files this lab uses.)

On Arch/Manjaro: pacman -S webkit2gtk-4.1 base-devel openssl librsvg libaec cmake clang pkgconf

Run

npm install
npm run tauri dev      # first build is slow; a window opens

Get a GRIB file first (see data/README.md):

./data/fetch_arome_sp1.sh        # latest real AROME +00H SP1 (~16 MB), or
python3 data/make_sample_grib.py # tiny offline synthetic sample

Then click Open a GRIB2 file… and pick it.

Test

cd src-tauri && cargo test       # decodes the synthetic sample + any real file in data/

Build a release binary

npm run tauri build              # installer/binary in src-tauri/target/release/