For decades, tracking global greenhouse gas emissions resembled managing a country's economy solely through annual tax returns. Traditional "bottom-up" emission inventories rely on aggregated statistical data—such as national energy consumption reports, economic censuses, and fuel sales records. Because compiling these datasets requires extensive auditing, official inventories typically suffer from a time lag of one to two years.
While annual reporting is sufficient for long-term climate modeling, it fails during rapid societal or environmental shocks. When global lockdowns occurred in 2020, or when extreme weather events abruptly altered natural carbon cycles, policymakers and researchers could not afford to wait two years to see the results. As argued by Bo Zheng et al., relying on annual proxies prevents us from understanding the immediate impacts of policy interventions or economic disruptions.
To bridge this gap, a new research paradigm has emerged: near-real-time (NRT) emissions monitoring. By combining high-frequency activity proxies with satellite observations and atmospheric transport models, researchers can now estimate gridded greenhouse gas and pollutant emissions on a daily or even hourly basis.
The Core Mechanism: Proxies and Activity Data
We cannot place physical meters on every tailpipe, smokestack, and building furnace in the world. To estimate daily emissions, NRT frameworks exploit a fundamental intuition: short-term fluctuations in emissions scale linearly with high-frequency indicators of human activity.
Instead of waiting for official fuel consumption statistics, bottom-up NRT models ingest daily, sector-specific data streams:
- Power Sector: Hourly electricity generation data from national grid operators.
- Industry: Daily steel, cement, and chemical production indices, alongside satellite-derived nitrogen dioxide ($\text{NO}_2$) plumes over industrial clusters.
- Ground Transportation: Real-time traffic congestion indices from GPS navigation services and city mobility reports.
- Aviation: Daily flight-tracking data from global transponder networks.
In a foundational study, Zhu Liu et al. introduced a framework to compute daily global $\text{CO}_2$ emissions by sector. By applying daily activity scaling factors to baseline annual inventories, the authors quantified an unprecedented 8.8% drop in global emissions during the first half of 2020.
[High-Frequency Proxies] (Mobility, Grid Load, Flight Logs)
│
▼
[Daily Scaling Factors] × [Baseline Annual Inventories]
│
▼
[Near-Real-Time Daily Emission Estimates]
This methodology has been scaled across different geographic and temporal resolutions. For instance, Piyu Ke et al. established the Carbon Monitor Europe dataset, providing daily sector-specific tracking for 27 EU countries and the UK. On a gridded global scale, the GRACED dataset developed by Xinyu Dou et al. and updated in subsequent work (Xinyu Dou et al.) distributes these daily estimates across a 0.1° × 0.1° spatial grid. This allows researchers to observe exactly where and when emissions fluctuate in response to localized economic shifts.
Top-Down Verification: Atmospheric Inversion
While proxy-based tracking is agile, it remains an indirect estimation method. To verify these estimates, researchers use a "top-down" approach called atmospheric inversion.
If bottom-up tracking asks, "Based on traffic data, how much gas did we emit?", top-down verification asks, "Based on the air composition downwind, what must have happened upwind?"
[Upwind Sources] ───(Wind & Diffusion)───► [Downwind Sensors]
▲ │
└─────────── [Atmospheric Inversion] ───────┘
(Physics-Based Reversal)
This method relies on two main components:
- Concentration Measurements: Ground-based sensor networks or satellite instruments (such as TROPOMI) measure the actual concentration of gases in the atmosphere.
- Transport Modeling: Physics-based meteorological models simulate how wind currents, temperature, and topography move and dilute air parcels.
By running these transport models in reverse, an inversion system calculates the exact spatial configuration of surface sources required to produce the observed atmospheric concentrations.
For example, Hui Lĭ et al. used satellite $\text{NO}_2$ observations combined with an inversion model to capture a steep 15% decline in China's $\text{CO}_2$ emissions during the 2022 Omicron wave. Because $\text{NO}_2$ is co-emitted with $\text{CO}_2$ during fossil fuel combustion but has a much shorter atmospheric lifetime, it serves as an excellent tracer for recent industrial and transport activity. Similarly, Anthony Rey-Pommier et al. applied a mass-balance inversion to TROPOMI data to generate high-resolution global gridded $\text{NO}_x$ emission datasets, bypassing the delays of traditional reporting.
Scaling Down to Cities
The utility of NRT monitoring is most apparent at the urban scale, where local governments require timely data to assess climate action plans. Cities are highly dynamic environments; a single holiday or a transit strike can radically alter their carbon profile.
Traditional bottom-up inventories struggle at this scale due to boundary definitions and data scarcity. To resolve this, Da Huo et al. built Carbon Monitor Cities, providing daily $\text{CO}_2$ estimates for 1,500 cities worldwide.
When these proxy models are paired with local atmospheric monitoring networks, the results become highly verifiable. In Paris, Jinghui Lian et al. utilized the 2020 lockdown as a natural experiment to test an urban sensor network. Their atmospheric inversion system successfully detected a sub-annual emission drop of over 40%, proving that ground-based networks can verify sharp, short-term shifts in urban carbon footprints. In a follow-up 6-year study over Paris, Jinghui Lian et al. demonstrated that combining hourly, 1-km resolution bottom-up priors with atmospheric measurements allows for the reliable verification of long-term urban emission trends.
Why Temporal Granularity Matters
Moving from annual to daily resolution transforms emissions data from a historical record into an actionable management tool.
Capturing Short-Term Anomalies
Annual averages smooth over critical spikes caused by human behavior or environmental stress. For example, Peng Wang et al. demonstrated that a sudden surge in vehicle emissions during China's National Day holiday triggered unexpected regional ozone exposure and public health risks. An annual inventory completely misses these brief, high-impact events.
Monitoring Natural Sinks
NRT tracking is not limited to human emissions; it is equally vital for monitoring the biosphere. Philippe Ciais et al. deployed a low-latency global carbon budget framework to discover a severe weakening of the terrestrial land carbon sink in 2024, driven by widespread droughts and wildfires. Detecting these shifts in real-time allows scientists to alert climate agencies to rapid changes in the Earth's natural capacity to absorb $\text{CO}_2$.
Validating Policy and Rebounds
When global events disrupt society, NRT data reveals how quickly systems recover. While the initial lockdowns caused historic drops in emissions, NRT tracking by Robert B. Jackson et al. showed that global fossil $\text{CO}_2$ emissions rebounded almost immediately to pre-pandemic levels as economic activity resumed.
As noted by Marc Guevara et al., integrating these diverse, high-frequency data streams requires careful validation to reconcile sector-level discrepancies. However, the transition toward near-real-time monitoring fundamentally changes climate governance, providing the empirical agility needed to manage emissions dynamically in an uncertain world.
Key Foundational Work
Near-real-time global gridded daily CO2 emissions (2021) — contributed 2 concept elements to this direction.
Satellite-based estimates of decline and rebound in China’s CO 2 emissions during COVID-19 pandemic (2020) — contributed 2 concept elements to this direction.
Near-real-time monitoring of global CO2 emissions reveals the effects of the COVID-19 pandemic (2020) — contributed 1 concept elements to this direction.