Runnable notebook vignettes¶
These are real Jupyter .ipynb files tracked in the SpectralBridge repository.
Each link opens the notebook in GitHub's repository viewer so you can read its
Markdown and code cells on the web. GitHub does not execute the cells; clone or
download the notebook when you are ready to run it locally.
The notebooks mirror the learning modules and call existing SpectralBridge functions. They contain configuration cells, explanations of inputs and outputs, and validation checkpoints. Copy a notebook into your own analysis directory before changing scientific assumptions.
The NEON and drone examples follow the two active root research notebooks. The local bulk example uses the public API on an already curated file tree. A separate production notebook configures the package-owned CyVerse drone campaign workflow. It contains no remote parsing, transfer loops, identity regexes, checkpoint implementation, bulk validation, or upload plumbing.
| Order | Notebook | Use it when |
|---|---|---|
| 00 | Full NEON pipeline | You want download through QA in one restart-safe call |
| 01 | Acquire NEON HDF5 | You only want to obtain or reuse the source HDF5 |
| 02 | Correct NEON reflectance | You want raw ENVI plus standard topo/BRDF outputs |
| 03 | Harmonize to Landsat | You already have corrected hyperspectral ENVI and want target-sensor products |
| 04 | Build and inspect analysis tables | You want Parquet/CSV inspection and output checks |
| 05 | Review QA and validation | You want to render QA and inspect machine-readable diagnostics |
| 06 | Process drone imagery | You have local drone HDF5 inputs |
| 07 | Extract polygon spectra | You want polygon-indexed spectra from a completed flightline |
| 08 | Insert a custom correction | You are developing a reviewed correction after topo/BRDF and before convolution |
| 09 | Build a bulk cross-run analysis | You want canonical catalogs, virtual queries, balanced regressions, and held-out-site validation |
| 10 | Run CyVerse drone-to-bulk production | You have remote drone ExportPackages and need tested staging, producer checkpoints, strict bulk analysis, reports, and verified closeout upload |
Opening them¶
From a clone:
python -m pip install -e ".[notebooks]"
jupyter lab docs/vignettes/notebooks/
The notebooks use repository-relative paths and begin with an editable
configuration cell. Large runs can take substantial memory and storage. Start
with one flightline, engine="thread", and max_workers=1.
What “runnable” means¶
The notebooks have valid kernels and no saved outputs. The numbered learning
examples use RUN = False; set it to True after editing their configuration.
The CyVerse production notebook calls one restart-safe public function. Remote
upload remains disabled unless upload_results=True; the package never
overwrites an existing non-matching remote result.
The supplied PDF was used for review but is not published as runnable guidance: its print layout clips wide code cells. Use the tracked notebook to copy or run code.