Abstract
The dual-carbon goals of China necessitate precise accounting of company carbon emissions, vital for green development across all industries. Not only the company itself but also financial investors require accurate and comprehensive company-level emissions data for climate risk management. This paper introduces the structure and methodology of the High-resolution Database for Carbon Emissions of China-listed companies, integrating three primary data sources: self-disclosed environmental data from listed companies, long-accumulated national power emission data, and regional high-precision emission data derived from multi-source satellites. The database's innovation lies in the employment of artificial intelligence (AI) algorithms to aggregate multi-source satellite data. This approach enables the precise identification of carbon emission sources and the prediction of company-level carbon emissions. Consequently, this methodology robustly cross-validates self-reported direct emissions, enhancing the accuracy and granularity of company-level emission records. Central to the database's utility includes the provision of high-resolution company carbon emission data, which is not only highly accurate but also instrumental in carbon management and emission market transactions. By offering a more nuanced and verifiable picture of company emissions, the database supports China's broader efforts to meet its ambitious dual-carbon targets and transition towards a more sustainable and environmentally responsible economy.
Introduction
As China pursues its "dual-carbon" goals—peaking emissions by 2030 and reaching neutrality by 2060—the demand for precise carbon accounting has reached a critical point. Traditionally, corporate emissions data has relied heavily on self-disclosure, a method frequently hampered by inconsistency, lack of transparency, and the risk of "greenwashing." For investors and regulators, this creates an "information gap" that makes assessing true climate risk nearly impossible. This paper presents a high-resolution database that addresses these challenges by fusing multi-source satellite data with artificial intelligence to provide a verifiable, top-down view of corporate emissions.
Database Architecture and Data Sources
The database integrates a massive array of data sources to build a holistic view of corporate environmental impact. It covers 5,226 Chinese A-share listed companies, utilizing over 30,000 documents from the Shanghai and Shenzhen Stock Exchanges.
The system architecture is built upon three primary pillars:
- Company Self-Disclosed Data: Processed via web scraping and OCR from ESG and annual reports.
- National/Regional Power Emission Data: Used to derive regional emission factors for indirect emission calculations.
- Satellite Remote Sensing Data: Including $CO_2$ concentrations from OCO-2, OCO-3, and TanSat, alongside high-resolution imagery from Landsat 8/9 and Sentinel-2.
As shown in the table above, the database captures a granular set of variables, ranging from direct energy usage (coal, gas) to indirect usage (purchased electricity, heat) and specific industry metrics like data center emissions for IT firms.
Methodology: AI-Driven Multi-Modal Estimation
The core innovation of this database lies in its multi-modal approach to estimating the three "Scopes" of emissions:
- Scope 1 (Direct Emissions): The researchers utilized two distinct AI models to validate and estimate direct emissions. CarbonNet, a BERT-based transformer architecture, processes satellite $XCO_2$ data. Simultaneously, a Vision Transformer (ViT) analyzes high-resolution imagery to detect and quantify carbon-emitting plumes from industrial sites. This "top-down" physical measurement serves as a "truth layer" to cross-validate self-reported figures.
- Scope 2 (Indirect Emissions): These are calculated by mapping a company’s geographic location to its specific power grid and applying regional emission factors derived from national power generation data.
- Scope 3 (Value Chain Emissions): These are estimated using "equity penetration data," which calculates a company’s share of emissions based on its ownership stakes in subsidiaries and assets.
A critical "Bias Adjustment" step ensures data integrity: if self-reported Scope 1 data deviates significantly from satellite-estimated values, the system flags the discrepancy and prioritizes the more reliable satellite-derived figure.
Key Results and Sectoral Insights
The resulting database provides a comprehensive look at the emissions landscape of China's listed companies from January 2019 to January 2023.
- Broad Coverage: The database tracks 5,226 companies across various sectors.
- Sectoral Dominance: The manufacturing industry represents the largest segment (3,490 companies), followed by Information Technology and Wholesale/Retail.
- Transparency and Access: The researchers have made the data publicly available via intelligentcarbon.ai, offering variables such as total emissions, emission intensity, and specific energy usage.
Conclusion
By bridging the gap between AI, remote sensing, and financial reporting, this high-resolution database provides a vital tool for carbon transparency. While challenges remain—such as the inherent latency in satellite data processing and the complexity of global Scope 3 supply chains—this methodology offers a robust framework for verifying corporate climate claims. It moves the industry closer to a world where carbon footprints are measured objectively from space, supporting China's transition toward a sustainable, low-carbon economy.
Summary updated