palmwtc 0.4.0

Automated whole-tree chamber workflow for oil-palm ecophysiology — QC, flux calculation, science validation.

palmwtc

Automated whole-tree chamber workflow for oil-palm ecophysiology.

palmwtc is the data-processing and analysis pipeline for the first automated whole-tree chamber (WTC) sized to enclose individual oil palm trees, deployed at the LIBZ field site (Riau, Indonesia) and instrumented with LI-COR LI-850 gas analyzers. It transforms raw sensor cycles into validated CO₂ and H₂O fluxes, applies multi-stage quality control, and produces inputs for the XPalm digital-twin calibration pipeline.

What it does

raw chamber cycles  ──►  QC (rules + ML + breakpoints)  ──►  flux calculation  ──►  science validation
        │                                                                                  │
        └── 30-min weather + soil + tree biophysics ──┬───────────────────────────────────►┘
                                                       └── high-confidence calibration windows

End-to-end run on the bundled synthetic sample:

pip install palmwtc
palmwtc run            # uses bundled sample if no PALMWTC_DATA_DIR set

For your own data:

export PALMWTC_DATA_DIR=/path/to/your/chamber/data
palmwtc run --skip 022 025          # mirrors the original notebook-runner CLI
palmwtc run --notebooks             # papermill mode, produces HTML reports

Install

pip install palmwtc                 # core only
pip install 'palmwtc[ml]'           # + scikit-learn IsolationForest QC
pip install 'palmwtc[interactive]'  # + ipywidgets / anywidget for Jupyter dashboards
pip install 'palmwtc[gpu]'          # + torch (Apple-Silicon MPS / CUDA)
pip install 'palmwtc[all]'          # everything

Requires Python 3.11–3.13.

Library use

from palmwtc.config import DataPaths
from palmwtc.qc import QCProcessor
from palmwtc.flux import calculate_flux_cycles

paths = DataPaths.resolve()                     # layered: CLI → env → yaml → sample
qc_result = QCProcessor(paths).run("CO2_C1")
flux = calculate_flux_cycles(qc_result.data)

Full API reference: adisapoetro.github.io/palmwtc/api/

Citation

If you use palmwtc in scientific work, please cite the Zenodo DOI. The concept DOI 10.5281/zenodo.19680893 always resolves to the latest version; to cite a specific release, use its own version DOI (visible on the Zenodo record → Versions panel).

@software{adisaputro_palmwtc_2026,
  author  = {Adisaputro, Didi},
  title   = {palmwtc: Automated whole-tree chamber workflow for oil-palm ecophysiology},
  year    = {2026},
  version = {0.2.0},
  doi     = {10.5281/zenodo.19680893},
  url     = {https://github.com/adisapoetro/palmwtc},
}

See CITATION.cff for machine-readable metadata.

Background

Most ecosystem-scale flux measurements over oil-palm plantations use eddy covariance, which integrates over hectares of canopy and cannot resolve single-tree behaviour. palmwtc is built around a different instrument: an automated whole-tree chamber, sized and ventilated to enclose an individual mature oil palm, with an LI-COR LI-850 gas analyzer cycling open and closed on a programmed schedule. This is, to our knowledge, the first WTC deployment for oil palm — the WTC method was previously applied to temperate broadleaf species (Medlyn et al. 2016).

What does an automated whole-tree chamber look like?

The illustration below shows one concrete example — the LIBZ chamber from which palmwtc was originally developed (Riau, Indonesia; whole-tree chamber around an individual mature oil palm; aboveground sensor pole carrying sonic anemometers, temperature/humidity sensors, and LI-850 air inlets at five heights; underground TEROS-21 + Watermark sensors at five depths).

This is one possible implementation, not a specification. palmwtc is a generic toolkit — it does not assume any one chamber design. Other deployments will differ in:

• Chamber dimensions — palmwtc takes the chamber volume as a configuration input; chambers from a few cubic metres to large enclosures all work.
• Tree species and growth form — the package is not oil-palm-specific; any plant that fits in a closed chamber can be processed.
• Sensor pole design — the number of heights and the sensor mix (sonic anemometer, T/RH, PAR, gas-analyser inlets) varies per deployment.
• Soil instrumentation — depth count and sensor brand (TEROS-21, Watermark, TDR, capacitance) varies per deployment.
• Datalogger choice — Campbell Scientific is one common option; others work as long as the data files can be loaded.

palmwtc consumes whatever data your chamber produces and computes flux. The hardware, materials, and dimensions are out of scope for the package itself — they belong to your specific field setup.

Contributing

See CONTRIBUTING.md. Bug reports and feature requests: github.com/adisapoetro/palmwtc/issues.

License

MIT — see LICENSE.