Carbon ACX - Trustworthy Carbon Accounting

Data Pipeline

CSV Inputs (data/)
    ↓
Pydantic Schemas (calc/schema.py)
    ↓
Derivation Engine (calc/derive.py)
    ↓
Figure Generation (calc/figures.py)
    ↓
Manifest Creation (calc/figures_manifest.py)
    ↓
Hashed Artifacts (dist/artifacts/<hash>/)

Manifest-First Architecture

Every data artifact is accompanied by a manifest containing:

Byte Hash (SHA256): Cryptographic hash of the figure for integrity verification
Generation Timestamp: ISO 8601 timestamp of when the artifact was created
Citation Keys: References to source data and methodology
Provenance Chain: Complete lineage from raw data to final artifact
Numeric Invariance: Test results ensuring calculations remain consistent

Data Sources

Emission Factors

Source: data/emission_factors.csv

Validated emission factors from authoritative sources including IPCC, EPA, and GHG Protocol.

Grid Intensity

Source: data/grid_intensity.csv

Regional electricity grid carbon intensity values updated quarterly.

Activities

Source: data/activities.csv

Standardized activity definitions mapped to emission factors.

Verification

All artifacts can be independently verified:

Retrieve manifest from /api/manifests/[id]
Download referenced figure file
Compute SHA256 hash of figure
Compare computed hash with manifest.figure_sha256
Verify numeric invariance tests passed

Reproducibility

The entire build process is reproducible:

# Clone repository
git clone https://github.com/chrislyons/carbon-acx.git

# Install dependencies
poetry install

# Run derivation pipeline
make build

# Verify artifacts match manifests
pytest tests/test_manifests.py

Methodology