How It Works

AtlasCore is an Australian climate data infrastructure platform. It normalises data from 10+ government sources into a single REST API, producing audit-ready outputs for AASB S2 climate disclosure and company climate intelligence.

Three-layer architecture

Layer 1 — Connectors

Stateless source adapters that extract data from government APIs, spreadsheets, and databases. Each connector handles the quirks of its source (merged Excel cells, SOAP APIs, NetCDF grids) and produces normalised records.

Current climate connectors include NGA Factors (XLSX), AEMO CDEII (ZIP/CSV), CER NGER (XLSX), BOM SILO (NetCDF), ABR Lookup (SOAP), ABS (SDMX), GreenPower (HTML), GNAF (bulk download), BOM Water (NetCDF), CER ACCU/LGC registries (CSV), and ABS I-O tables (CSV).

Layer 2 — Canonical datasets

Normalised, deduplicated, versioned storage. Every dataset gets a globally unique slug following the convention {jurisdiction}_{provider}_{content} — for example, au_dcceew_nga_emission_factors or au_aemo_cdeii_grid_intensity.

Key properties:

• One slug = one upstream source — no cross-provider aggregation
• Snapshot versioning — every extraction creates a versioned snapshot
• Capability tagging — datasets declare capabilities and tags for dynamic pack binding

Layer 3 — Packs

Vertical product layers that consume datasets. Packs define which datasets they require, business rules, report templates, and acceptance test suites. Packs never own ingestion, normalisation, or entity resolution — they declare intent and the engine does the work.

AtlasCore currently ships two climate packs:

Pack	What it does
AU Climate Disclosure	Emission factor lookups, GreenPower reference data, statistical surfaces, AASB S2 framework registry
AU Company Climate Intelligence	Company climate profiles, risk assessment, sector benchmarking, emissions trends, filing readiness, disclosure bundles

Request flow

When you make an API request like GET /v1/companies/{abn}/climate-profile:

1. Authentication — bearer token validated, tenant and pack entitlement checked
2. Entity resolution — ABN resolved against persisted CER NGER and ABR matches
3. Data composition — emissions, grid intensity, BOM SILO climate data, sector classification, benchmarks, and trends composed from canonical stores
4. Deterministic output — profile computed from persisted inputs only (no live API calls at query time)
5. Evidence — response includes evidence_hash and structured provenance for audit

What you get in one call

A single company climate profile includes:

• Scope 1 and 2 emissions from CER NGER
• Sector classification from ANZSIC division and industry codes
• Sector benchmarking — percentile rank against peers, median/quartile statistics
• Emissions trend — multi-year CAGR and linear trajectory to 2030
• Transition risk — sector intensity, Safeguard coverage, regulatory trajectory
• Physical risk — BOM SILO temperature, rainfall, extreme heat days
• Scope 3 screening — indicative category-level estimates
• Provenance — five government authorities with named datasets and evidence hash

What AtlasCore is not

• Not a calculator — AtlasCore retrieves and normalises authoritative government data. It does not estimate or model emissions beyond screening-grade Scope 3 indicators.
• Not a filing tool — Evidence bundles support AASB S2 reporting but AtlasCore does not submit filings.
• Not a data warehouse — Canonical storage is purpose-built for government data normalisation, not general analytics.

Design principles

• Deterministic outputs — same inputs always produce the same outputs. No stochastic models.
• Provenance by default — seven-file evidence bundles on every disclosure report (JSON, PDF, XLSX, Markdown, provenance, checksums, ZIP).
• Licence-first — every government source is audited for commercial repackaging rights before inclusion.
• Pack isolation — pack code never contains engine logic or connector references.