Multi-Source Emissions Reconciliation: 2022 Snapshot Status

Why four sources

Single-source emissions dashboards present national totals as if they were measured. They are not. Every published greenhouse-gas total is the output of a measurement-modeling-reporting chain whose assumptions differ across the four global sources PlainEmissions reconciles. UNFCCC national inventories are politically negotiated self-reports under IPCC reporting guidelines; EDGAR is a globally-consistent bottom-up model that applies the same emission factors to every country; Climate TRACE uses satellite measurements and machine-learning attribution at facility resolution; the World Bank Climate Knowledge Portal aggregates multiple feeds for macro indicators. None is wrong. All disagree on specific countries. The disagreement is the information.

2022 EDGAR substrate

The current PlainEmissions database holds 14,310 MtCO2e for the world's top emitter (China) and 42,507 MtCO2e total across all 51 countries currently loaded — the EDGAR v8.0 snapshot for 2022. The next reconciliation pages will overlay the Climate TRACE 2023 release and UNFCCC most-recent submissions on this substrate, surfacing the per-country divergence directly via comparison tables and chart overlays.

Source: EU EDGAR v8.0, Joint Research Centre, European Commission (2024). CC BY 4.0.

Gas-level distribution

Pulling the 2022 totals by gas from emissions_records shows the global mix — CO2 dominates, methane follows at a fraction reflecting its shorter atmospheric lifetime, nitrous oxide and the F-gases round out. The figures below use AR6 GWP100 weighting; native-unit (megatonnes per gas) is also stored in emissions_records.value_native for users who prefer non-CO2e accounting.

Source: EU EDGAR v8.0 (2024). Query: SUM(value_mt_co2e) BY gas_code WHERE year=2022 AND source_code='EDGAR'.

Reconciliation roadmap

The reconciliation framework PlainEmissions applies follows the same logic across all four upstream sources. First, normalize country codes to ISO 3166-1 alpha-3. Second, fold each source's sector taxonomy into a single IPCC-aligned 8-sector hierarchy. Third, convert every value to CO2-equivalent megatonnes using consistent AR6 GWP100 multipliers (retaining native units alongside for transparency). Fourth, store provenance per row so multi-source comparisons remain queryable without joins. Fifth, never interpolate — when a country-year-sector-gas combination is missing from one source, the absence is preserved rather than back-filled. The user sees disagreement as it actually exists in the data.

The roadmap for filling out the substrate prioritizes the Climate TRACE 2023 release for methane validation (where the satellite-vs-inventory gap is largest), the UNFCCC 2024 submission round for Annex I parties, and the World Bank Climate Knowledge Portal for macro indicator alignment. As each source loads into emissions_records with its own source_code, the cross-source comparison pages light up automatically — sample queries are documented in /methodology.

Methodology

The PlainEmissions reconciliation framework holds four data sources side-by-side per country-year-sector-gas combination rather than picking one. The fact table emissions_records stores one row per (iso3, year, sector_code, gas_code, source_code, gas-unit) tuple. Each row preserves the upstream value verbatim (no rounding, no normalization beyond unit conversion) along with the native gas-quantity value and unit metadata. Cross-source comparisons are then a single SQL GROUP BY away rather than requiring multi-table joins. Source-specific quirks — UNFCCC's politically-negotiated reporting timelines, EDGAR's annual model release, Climate TRACE's quarterly satellite refreshes, World Bank's monthly indicator pulls — are all captured in the data_sources reference table with vintage metadata and license terms. For full pipeline detail see PlainEmissions methodology and the per-source documentation linked from the data-sources index.

Limitations

The current 2022 baseline relies on the EDGAR v8.0 snapshot loaded into emissions_records via the 05-seed-edgar ETL stage. Sectoral splits use country-specific share distributions for known outliers (Brazil's LULUCF dominance, Indonesia's land-use share, Saudi Arabia's fugitive emissions weight, New Zealand's agriculture share) and a default 8-sector distribution for the remaining countries. Full sectoral disaggregation per country arrives in 10-load-edgar-full.mjs once the bulk EDGAR extract is staged. Multi-year reconstructions (2017-2021) apply a country-specific growth trend to the 2022 baseline — when actual EDGAR time series load, these will replace the trend reconstruction. Climate TRACE and UNFCCC overlays are stubs until the corresponding stages ship. As with all emissions data, the methodology page tracks current vintage per source and flags known-stale dimensions.

Related

• Top 15 Emitting Countries (2022)
• Satellite vs Inventory Methane Gap
• Methodology & Data Sources