Top-Down Atmospheric Inversion for Urban Carbon Accounting

26 concepts20 papers2 proposals

Traditional carbon accounting relies on "bottom-up" inventories. As synthesized by R. J. Andres et al. in A synthesis of carbon dioxide emissions from fossil-fuel combustion, these inventories aggregate economic activity data—such as fuel sales, factory outputs, and traffic counts—and multiply them by estimated emission factors.

While this approach provides a reasonable baseline for global budgets (Pierre Friedlingstein et al. in Global Carbon Budget 2021), it degrades significantly at sub-national and city scales. Cities are the primary engines of anthropogenic greenhouse gas emissions, yet their self-reported inventories are frequently plagued by data gaps, delayed reporting, and structural uncertainties (Rong Wang et al. in High-resolution mapping of combustion processes and implications for CO 2 emissions).

To independently verify emissions for monitoring, reporting, and verification (MRV) frameworks, researchers have developed a "top-down" alternative: atmospheric inversion. Instead of counting fuel sales on the ground, top-down methods measure the actual concentration of $\text{CO}_2$ in the atmosphere and work backward using physics to infer the surface fluxes that must have caused them.

Isolating the Urban Signal via Gradients

The fundamental challenge of measuring a city's $\text{CO}_2$ emissions from the air is signal-to-noise ratio. The global atmosphere is flooded with a massive background concentration of $\text{CO}_2$ (currently exceeding 420 parts per million, or ppm), driven by global circulation and seasonal biospheric "breathing." A major metropolis like Paris or London might only add 2 to 10 ppm to the air passing over it.

To isolate this tiny urban signal from the massive background, top-down systems rely on the upwind-downwind gradient principle.

Imagine a parcel of air moving across a landscape. Before it enters a city, a sensor measures its baseline $\text{CO}_2$ concentration (the upwind station). As the air parcel travels over the city, urban combustion processes inject additional $\text{CO}_2$ into it. When the air exits the metropolitan boundary, another sensor measures the elevated concentration (the downwind station). By subtracting the upwind background from the downwind measurement, researchers calculate the net urban enhancement ($\Delta\text{CO}_2$).

  Wind Direction  --->  --->  --->
       |                        |
       v                        v
[Upwind Sensor]  --> [ CITY ] --> [Downwind Sensor]
 (Background)       (Emissions)     (Background + City Signal)

François‐Marie Bréon et al. demonstrated the power of this approach in An attempt at estimating Paris area CO 2 emissions from atmospheric concentration measurements, showing that focusing strictly on the gradient between upwind and downwind stations cancels out regional background fluctuations and dramatically improves the alignment between atmospheric models and real-world observations.

This gradient framework is highly adaptable. It can be deployed using near-ground in-situ sampling networks to capture localized urban-rural differences (Alex Boon et al. in Analysis of the potential of near-ground measurements of CO 2 and CH 4 in London, UK, for the monitoring of city-scale emissions using an atmospheric transport model), continuous high-altitude tower networks (Irène Xueref-Rémy et al. in Diurnal, synoptic and seasonal variability of atmospheric CO 2 in the Paris megacity area), or portable ground-based Fourier Transform Infrared (FTIR) spectrometers that measure the total column of air above the city (Felix Vogel et al. in XCO 2 in an emission hot-spot region: the COCCON Paris campaign 2015).

The Physics of Atmospheric Transport Inversion

Knowing the concentration enhancement ($\Delta\text{CO}_2$ in ppm) is only the first step. To verify an inventory, we must translate that concentration back into a mass flux (such as tons of carbon emitted per hour). This translation requires modeling the fluid dynamics of the atmosphere.

Atmospheric transport models simulate how wind fields, turbulence, and planetary boundary layer (PBL) dynamics dilute and disperse emissions. The intuition can be understood through two primary meteorological variables:

  1. Wind Speed: If the wind is blowing at 20 km/h, emissions are swept away and diluted into a large volume of air, resulting in a small downwind ppm enhancement. If the wind drops to 5 km/h, the same emission flux will accumulate over the city, creating a much larger ppm spike.
  2. Boundary Layer Height: The PBL is the layer of the atmosphere closest to the Earth's surface where air is actively mixed. Driven by solar heating, the PBL expands up to several kilometers deep during the day, providing a massive volume for $\text{CO}_2$ dilution. At night, the PBL collapses into a shallow "stable layer" just tens or hundreds of meters deep. As shown by Irène Xueref-Rémy et al. in Diurnal, synoptic and seasonal variability of atmospheric CO 2 in the Paris megacity area, this diurnal collapse traps urban emissions near the ground, causing massive spikes in measured concentration even if the actual rate of city traffic or heating remains constant.

An atmospheric inversion system combines these physical principles mathematically using Bayesian inference. The system takes a "prior" estimate (the bottom-up inventory) and runs it forward through a high-resolution transport model to predict what the atmospheric concentrations should look like at the sensor locations.

The model then calculates the discrepancies between these predictions and the actual observed gradients. Finally, an optimization algorithm adjusts the surface emission map to find the most statistically likely flux distribution that satisfies both the physical constraints of the weather and the atmospheric measurements. This requires high-resolution mesoscale modeling to accurately capture complex terrain and fine-scale coastal or urban circulations (G. Pérez-Landa et al. in Mesoscale circulations over complex terrain in the Valencia coastal region, Spain – Part 2: Modeling CO 2 transport using idealized surface fluxes).

Scaling Up: Satellites and Co-Emitted Proxies

While ground-based sensor networks offer high precision, they are expensive to maintain and cannot scale globally to cover every major emission source. To achieve global coverage, researchers are leveraging spaceborne instruments. However, satellite observations face severe signal-to-noise constraints.

As evaluated by Franck Lespinas et al. in The potential of a constellation of low earth orbit satellite imagers to monitor worldwide fossil fuel CO2 emissions from large cities and point sources, current and near-future Low Earth Orbit (LEO) satellite constellations can reliably quantify emissions primarily for large point sources and megacities emitting more than 0.5 million tonnes of carbon (MtC) per year.

To bypass the difficulty of detecting faint $\text{CO}_2$ plumes from space, scientists utilize proxy species. When fossil fuels are combusted, they do not emit pure $\text{CO}_2$; they co-emit trace gases like nitrogen dioxide ($\text{NO}_2$) and carbon monoxide ($\text{CO}$).

Unlike $\text{CO}_2$, which has a massive atmospheric background, the natural atmospheric background of $\text{NO}_2$ is nearly zero. Consequently, an urban combustion plume stands out against the background with an incredibly high signal-to-noise ratio, making it highly visible to satellite spectrometers.

Fossil Fuel Combustion ---> Emits CO2  (Low signal-to-background ratio)
                        ---> Emits NO2  (High signal-to-background ratio)

By establishing the characteristic emission ratios of $\text{CO}_2$ to $\text{NO}_2$ for different fuel types, researchers can use satellite observations of $\text{NO}_2$ columns as an indirect proxy to map $\text{CO}_2$ trends.

E. V. Berezin et al. successfully applied this methodology in Multiannual changes of CO 2 emissions in China: indirect estimates derived from satellite measurements of tropospheric NO 2 columns to derive multi-year emission trajectories over rapidly developing economies, revealing discrepancies where traditional bottom-up inventories over- or underestimated growth rates.

When combined with transport models, multi-proxy inversions ($\text{NO}_2$ and $\text{CO}$) provide an independent, top-down lens to identify systematic biases in national reporting (I. B. Konovalov et al. in Estimation of fossil-fuel CO 2 emissions using satellite measurements of "proxy" species).

Challenges in Sectoral Disaggregation and Uncertainty

Top-down atmospheric inversion has matured from a theoretical exercise into an operational verification tool. Multi-year inversion frameworks, such as the 6-year study over Paris by Jinghui Lian et al. in Can we use atmospheric CO 2 measurements to verify emission trends reported by cities? Lessons from a 6-year atmospheric inversion over Paris, have proven capable of tracking real-world emission declines and capturing sudden macroeconomic shocks, such as the COVID-19 lockdowns.

However, two major frontiers remain active areas of research:

To build reliable climate verification systems, ongoing research focuses on rigorously quantifying how these transport model errors, measurement uncertainties, and biospheric background fluxes propagate through the inversion equation (I. G. Enting et al. in Carbon Cycle Uncertainty in REgional Carbon Cycle Assessment and Processes (RECCAP); Ana Maria Roxana Petrescu et al. in The consolidated European synthesis of CO 2 emissions and removals for the European Union and United Kingdom: 1990–2018). By anchoring statistical models in physical atmospheric gradients, top-down inversion bridges the gap between reported climate commitments and the physical reality of the atmosphere.

History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_tracking_co2_emissions_696e710a,
  title = {tracking CO2 emissions},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/tracking-co2-emissions/rdr_696e710ac67e4aeeb3b99f1a7e937597},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-22},
}

This exact version (version 1, frozen text):

@misc{lacuna_tracking_co2_emissions_696e710a_v1,
  title = {tracking CO2 emissions},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/tracking-co2-emissions/rdr_696e710ac67e4aeeb3b99f1a7e937597/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-22},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake