Estimating global 0.1° scale gridded anthropogenic CO₂ emissions using TROPOMI NO₂ and a data-driven method.

Area of Science:

• Earth and Atmospheric Sciences
• Environmental Science
• Remote Sensing

Background:

• Existing satellite CO2 observations lack the resolution for fine-scale global mapping.
• Accurate, high-resolution CO2 emission data is crucial for climate change mitigation and policy development.

Purpose of the Study:

• To develop a novel data-driven method for estimating global anthropogenic CO2 emissions at a 0.1° scale.
• To leverage high-resolution nitrogen dioxide (NO2) satellite data to overcome limitations of direct CO2 measurements.

Main Methods:

• Utilized TROPOspheric Monitoring Instrument (TROPOMI) NO2 measurements as a proxy for CO2 emissions.
• Employed a long short-term memory (LSTM) neural network to predict CO2/NOx emission and NOx/NO2 conversion ratios.
• Implemented a random forest regression (RFR) model trained with the Emissions Database for Global Atmospheric Research (EDGAR) for CO2 emission estimation.

Main Results:

• Achieved a 0.1° scale global anthropogenic CO2 emission dataset with high accuracy.
• The dataset showed strong agreement with national inventories (R2=0.998 with GCB, R2=0.996 with EDGAR).
• Demonstrated consistency with city-level estimates (R2=0.824 with CMC).

Conclusions:

• The proposed data-driven method effectively estimates fine-resolution anthropogenic CO2 emissions.
• Satellite-observed NO2 provides a viable alternative for monitoring CO2 emissions at scales previously unattainable.
• This approach offers a new perspective for detailed global CO2 emission assessments.