Eight notebooks ordered as a learning path. Each builds on concepts from the previous, taking you from loading and inspecting data through transforms, efficient storage, and into a PyTorch training loop.
| # | Notebook | Topic |
|---|---|---|
| 1 | Loading and Visualising ERA5 | from_source(), inspection, plotting |
| 2 | Writing to Zarr | Cloud-native chunked storage |
| 3 | Unit Conversion and Normalisation | Transforms, scaling for ML |
| 4 | Grid Formats and Regridding | Regular grids, HEALPix, sub-area extraction |
| 5 | Temporal Statistics and Derived Variables | Aggregation, anomalies, rolling means |
| 6 | Chunking Strategies for ML | Read performance, DataLoader alignment |
| 7 | Multi-source Ingestion and Polytope | Unified pipeline, DestinE access |
| 8 | From Zarr to PyTorch Training Loop | Dataset, DataLoader, training |