Tutorials

Fourteen tutorial notebooks cover the full palmwtc workflow. Each is a thin wrapper around the library API: DataPaths.resolve() → run_step(...) → plot → narrative. All algorithmic logic stays in the palmwtc.* package.

Recommended starting point

000 — Integrated end-to-end runs the full pipeline (qc → flux → windows → validation → plot) in a single notebook against the bundled synthetic sample. Markdown cells explain the scientific meaning of each step. New users should open this first before diving into the per-stage tutorials below.

Pipeline spine (run in order)

#	Title	Runs on synthetic?
010	Core data integration (raw → unified monthly)	Stub (needs raw chamber files)
020	Rule-based QC	Stub (needs raw chamber files)
030	CO₂ / H₂O flux cycle calculation	Yes
032	Calibration window selection (production)	Yes (0 windows on toy data)
033	Science validation against literature	Yes (reports skip on synthetic)

Diagnostics & audits

#	Title	Purpose
011	Weather station vs chamber	Microclimate-artifact diagnostic
025	Cross-chamber bias	Sensor calibration consistency check
026	CO₂/H₂O segmented bias	Time-localised drift detection
034	QC + window audit	Pre-calibration sanity check

Optional / opt-in

#	Title	Purpose
022	ML-enhanced QC (optional)	IsolationForest contextual outliers (palmwtc[ml])
023	Field alert HTML report	Email-able operator report
031	Window selection (reference)	Methodology audit; superseded by 032
035	QC threshold sensitivity sweep	Operating-point tuning