sentier-methods

What it is

LCIA methods for the Sentier platform: EF 3.x, IPCC, ReCiPe. Characterization-factor tables keyed by elementary flow, plus optional normalization and weighting sets.

A Data Layer repo. It ships loadable parquet and the schemas that describe it. No fetch, parse, or calculate code.

• Fed by sentier-importers, which delivers parquet by PR.
• Read by sentier-brightway, which fetches files pinned by commit. sentier-platform (LCIA matrices) is planned.
• Flow to CF links resolve via sentier-mappings.

Install

There is no package. Clone sentier-dev/sentier-methods and read the parquet directly.

Use

Read a table with pyarrow:

uv run --with pyarrow python -c "import pyarrow.parquet as pq; print(pq.read_table('data/01-ef-3.1/methods.parquet').num_rows)"

Validate all datasources against the schemas (this is what CI runs on every PR):

uv run --with pyarrow --with pyyaml --with 'jsonschema[format]' python scripts/validate.py

Layout

schema/                  # YAML column definitions + JSON Schema for metadata.json
data/                    # one subfolder per datasource, NN rank prefix
  01-ef-3.1/             #   EF 3.1: 25 methods, 319,393 CFs
  02-ipcc-2021/          #   scaffold, metadata.json only
  03-recipe-2016/        #   scaffold, metadata.json only
scripts/validate.py      # CI validator
.github/workflows/ci.yml # runs validate.py on every PR and non-main push

Data

Each data/<NN>-<datasource>/ folder holds:

file	rows
methods.parquet	one per (method, impact category); PK method_id
characterization-factors.parquet	one per CF; FK method_id
normalization-weighting.parquet	optional; FK method_id
metadata.json	datasource provenance

NN orders datasources and sets resolution precedence. Lower wins. method_id is <datasource>:<impact_category>, for example ef-3.1:climate-change.

Schema

Plain-YAML table definitions in schema/. LinkML is used only in sentier-vocab.

file	describes
method.yaml	methods.parquet
characterization-factor.yaml	characterization-factors.parquet
normalization-weighting.yaml	normalization-weighting.parquet
metadata.schema.json	metadata.json

The schemas are the contract. sentier-importers reads them to validate and cast data before delivering parquet here.

Contributing

• New datasource: add data/<NN>-<name>/metadata.json first. It validates without parquet.
• Parquet arrives by PR from sentier-importers. Do not add loaders or importers here.
• CI runs scripts/validate.py on every PR: metadata, required columns, dtypes, PK uniqueness, FK integrity.
• No git-LFS and no release artifacts. Parquet is committed directly.

License

MIT — open by default, client-loadable.