From space to strategy: using satellite data to drive low-emission development

For further reading, review and download the authors' full research paper Satellite-Based Measures for Tracking Atmospheric CO₂ and CH₄ at National, Subnational, and Urban Scales here.

Closing the data gap for low-emission development

A low-emission growth path across key sectors — including agriculture, mining, forestry, manufacturing, infrastructure, and utilities — is critical for sustainable development, poverty reduction, and a livable planet. Integrating emissions mitigation into development planning is vital for long-term environmental and economic resilience. 

Yet emissions policy has long been constrained by the lack of reliable local and regional air pollution data needed for diagnosing problems, designing interventions, and tracking outcomes. Satellite-based atmospheric emissions monitoring is beginning to close this gap.  In this blog, we present new high-resolution emissions estimates derived from satellite readings by the World Bank’s Data and Research departments to support evidence-based policymaking. 

Source data: High-resolution satellite readings for CO₂ and CH₄ monitoring

We use NASA’s Orbiting Carbon Observatory-2 (OCO-2) to track CO₂, using daily, georeferenced XCO₂ measurements at a resolution of 1.29 × 2.25 kilometers. OCO-2 follows a sun-synchronous orbit — a satellite path that passes over the same part of the Earth at the same local time each day, ensuring consistent measurement of air pollution over time. It crosses the equator around 1:30 PM local time and repeats coverage every 16 days.

For CH₄, we use data from the European Satellite Agency (ESA)’s TROPOMI (Sentinel-5P), which follows a similar sun-synchronized orbit with daily revisits and a resolution of 5.5 × 3.5 kilometers. We use Level 2 Offline XCH₄ data, corrected for surface reflectance bias.

These datasets support consistent, high-resolution monitoring of emissions. 

Making satellite estimates comparable across space and time

To ensure that CO₂ and CH₄ estimates are comparable across different regions and over time, we process the raw satellite data to calculate local anomalies — differences between observed and background concentrations. Background values are estimated using the method of Hakkarainen et al. (2019), which accounts for both geographic location and time. Because daily background estimates at high spatial resolution are not reliable, we calculate daily median values of XCO₂ for each 10-degree latitude band and then interpolate these to a 1-degree resolution. Using medians avoids distortion from outliers. We then subtract the background from each observation to get the local anomaly. Finally, we average these anomalies by month for each 5-kilometer grid cell to create a consistent, high-resolution dataset for analysis.

Tracking CO₂ and CH₄ emissions over time

NASA and ESA assign quality scores to each satellite observation, and we include only those meeting or exceeding the acceptable thresholds to ensure reliable trend analysis. This leads to uneven spatial coverage, especially in cloudy regions. Rather than coarsening the grid uniformly — which would overlook small areas with sufficient data — we expand the observation area around each cell until it meets a minimum data threshold. This adaptive approach balances local detail with statistical robustness across the study area.

We track emission trends in each global 5-kilometer grid cell using two indicators:

• (1) a long-term monthly trend (2014–2024 for CO₂; 2018–2025 for CH₄); 
• (2) a short-term shift, defined as the difference between the past 12-month average and earlier values.

The second indicator highlights recent developments relevant for near-term policy.

We summarize trends for any area using two methods: 

• (1) Unweighted – All grid cells are treated equally. We compute the share of cells with increasing and decreasing trends (including insignificant ones) and define a score as: [Increasing % – Decreasing %]. 
• (2) Weighted – Same logic, but with emission-based weights from the Emissions Database for Global Atmospheric Research (EDGAR) database. Percentages reflect shares of total emissions rather than cell counts, giving more weight to high-emission areas.

In both methods, scores are normalized to account for the share of statistically significant results in an area. Final scores can range from –100 (all significant decreases) to +100 (all significant increases), allowing easy comparison across regions.

Scalable emissions performance scores for any geographic area

Our 5-kilometer grid level emissions trend tracking enables us to estimate emissions performance scores for geographic areas of arbitrary scale. Our datasets currently include long- and short-term CO₂ and CH₄ performance scores for 242 countries and disputed areas, 3,242 provinces (level-1 administrative units), 36,563 sub-provinces (level-2), 13,636 Functional Urban Areas, and 670 offshore oil and gas zones within national Exclusive Economic Zones (EEZs). Open-access coding supports updates as new satellite observations become available. 

Illustration for global emissions 

In addition, our subnational trend and change estimates displayed in the bar charts can inform strategic allocation of international climate resources, tailored to both persistent and emerging emissions challenges.

Over the long term, CO₂ levels have decreased in 13,179 sub-national administrative levels and increased in 9,616, while in the most recent period, they have decreased in 9,663 but increased to 10,790.

For methane (CH₄), long-term trends show declines in 13,734 subnational administrative areas and rises in 9,675, whereas recent data indicate decreases in 15,143 and increases in 8,689.

Extending the method to other air pollutants 

The same methods and code can be applied to satellite data from Japan’s GCOM-C and the ESA’s Sentinel-5P for high-resolution estimation of emissions and population exposure for other pollutants, including PM₂.₅, NO₂, SO₂, CO, and ozone—at national, subnational (level-1 and level-2), and urban scales worldwide.