TL;DR
This paper introduces GRACED, a new near-real-time global gridded daily CO2 emissions dataset for 2021. It offers unprecedented temporal and spatial resolution, crucial for policymakers to monitor climate mitigation efforts and adapt strategies quickly.
What Problem Does This Paper Address?
Accurate and timely monitoring of CO2 emissions is vital for effective climate change mitigation. Existing global gridded CO2 datasets often have a significant data lag, making it difficult to track emission changes in near real-time. This lag hinders the ability of policymakers and researchers to quickly assess the impact of policies, economic events, or seasonal variations on emissions. The paper addresses this by providing a dataset that bridges the gap between national-level estimates and the need for fine-grained, daily global emissions data.
Methods
The GRACED dataset is constructed using a top-down spatial gridding approach. The core methodology involves:
- Downscaling National Emissions: Daily national CO2 emissions estimates from "Carbon Monitor" (a near-real-time dataset) are used as the primary source. These national figures are then disaggregated to a 0.1° x 0.1° spatial resolution.
- Spatial Allocation using EDGAR: The spatial patterns from the Emissions Database for Global Atmospheric Research (EDGAR) are employed to distribute the national emissions across finer grids. This ensures that the spatial distribution of emissions aligns with established, albeit monthly, patterns.
- Temporal Refinement with Satellite Data: To capture daily variations and sub-national changes, particularly in response to events like COVID-19 lockdowns, satellite-derived Nitrogen Dioxide (NO2) column concentrations from TROPOMI are used. NO2 is a short-lived pollutant co-emitted with CO2, making its atmospheric distribution a good proxy for recent anthropogenic activity changes. This allows for a more dynamic adjustment of emissions beyond static EDGAR patterns.
- Sectoral Breakdown: Emissions are categorized into seven sectors: industry (including cement production), power, residential consumption, ground transport, international aviation, domestic aviation, and international shipping.
Datasets
The GRACED dataset is built upon the following key input datasets:
- Carbon Monitor: Provides near-real-time daily national CO2 emissions estimates for fossil fuel combustion and cement production.
- EDGAR (v5.0_FT2019): Offers gridded spatial activity data for CO2 emissions at a 0.1° x 0.1° resolution, used for spatial allocation.
- TROPOMI NO2 Data: Satellite-derived tropospheric vertical column densities of NO2 are used to capture daily and sub-national variations in emissions.
The paper also compares GRACED with other existing datasets, including:
- ODIAC (Open Source Data Inventory for Anthropogenic CO2)
- CEDS (Community Emissions Data System)
- EDGAR (as a comparison dataset, distinct from its use as an input)
- GCP-GridFED
- IEA (International Energy Agency)
Key Insights and Ideas
- Timeliness is Paramount: The most significant contribution is the near-real-time, daily resolution of the dataset. This is critical for agile climate policy monitoring, moving beyond the typical monthly or annual lags of other datasets.
- Satellite Data as a Dynamic Proxy: The innovative use of TROPOMI NO2 data to dynamically adjust sub-national emissions is a key methodological advancement, allowing the dataset to reflect real-world events like localized lockdowns or economic shifts.
- Uncertainty Quantified: The paper provides a detailed uncertainty analysis, giving users a clear understanding of the dataset's reliability. The grid-level two-sigma uncertainty of ±19.9% for 2021 is presented as a testament to the dataset's accuracy despite its high resolution.
- Sectoral Differences in Uncertainty: The analysis reveals significant variations in uncertainty across different emission sectors, with residential consumption and international aviation showing higher uncertainties, often linked to the underlying Carbon Monitor data.
Key Results
- GRACED 2021 Dataset: The primary output is the GRACED dataset for 2021, providing daily CO2 emissions at a 0.1° x 0.1° spatial resolution across seven sectors.
- Uncertainty Estimates: The dataset exhibits a grid-level two-sigma uncertainty of ±19.9% for 2021, ±19.9% for 2020, and ±23.1% for 2019.
- Spatial Distribution: The dataset reveals distinct spatial emission patterns, with high emissions concentrated in high-income and rapidly developing regions, and lower emissions in less populated or lower-income areas.
- Weekend vs. Weekday Emissions: A general trend of lower weekend emissions compared to weekday emissions was observed globally, with this difference being more pronounced in high-income regions.
- Sectoral Uncertainty: The power sector shows the lowest uncertainty (±13.5% in 2021), while residential consumption exhibits the highest (±42.0% in 2020-2021).
Limitations and Future Work
- Time Coverage: The GRACED dataset's time coverage (2019-2021) is shorter than some other datasets, though the authors state this will be addressed as data accumulates.
- Uncertainty Sources: The primary source of uncertainty is identified as the Carbon Monitor data. While satellite data helps refine temporal and spatial patterns, the fundamental national emission estimates still carry inherent uncertainties.
- Spatial Proxy Limitations: The assumption that EDGAR's spatial proxies remain unchanged from 2019 for downscaling might not fully capture rapid sub-national changes in all regions.
- Future Work: The paper emphasizes the importance of continuing to update GRACED in a timely manner to aid policymakers in monitoring and adjusting climate policies. Further research could focus on extending the time series and refining the uncertainty analysis for specific regions or sectors.
Summary updated