Daily and 1/16 degree maps of CO2 fossil fuel emissions based on satellite retrievals of pollutant atmospheric data

In the context of the Paris Agreement on climate change and of a global effort to reduce greenhouse gas emissions, the monitoring of anthropogenic carbon dioxide (CO₂) emissions is needed to assist policy makers but represents a major challenge. While current inventories provide rather robust annual emission totals at country scale, they lag behind real time by many months and they lack spatial and sub-annual details. Here we map the daily surface fossil fuel CO₂ emissions at a 1/16 degree resolution over Europe, with the year 2021 as an example, based on spaceborne atmospheric composition observations.

As the high-resolution satellite monitoring of atmospheric CO₂ remains challenging, especially at a local spatial scale and a daily time scale, we take advantage of the co-emission of CO₂ and nitrogen oxides (NO_X) during fossil fuel combustion: we exploit images of nitrogen dioxide (NO₂) concentrations retrieved from the measurements of the Tropospheric Monitoring Instrument (TROPOMI) onboard the Sentinel-5P satellite.

From the TROPOMI NO₂ concentrations, we retrieve daily maps of NO_X emissions based on the divergence of the mass fluxes within the NO₂ images. We combine the changes of these maps from one year to the next with low latency national CO₂ emissions from Carbon Monitor (https://carbonmonitor.org/), and with a baseline of monthly spatially-distributed CO₂ emissions for a previous year (here 2020) from GridFED (https://mattwjones.co.uk/co2-emissions-gridded/) from which we removed aviation and shipping emissions beforehand.

The resulting maps of emission increments from 2020 to 2021 capture changes in highly emitting areas: major urban or industrial areas, and main transport corridors. The emissions for the year 2021 show good consistency with existing inventories. The dataset also produces realistic seasonal variability at a local scale and captures daily variability, although temporally smoothed due to a 5-day rolling average of Carbon Monitor data.

This method is both temporally and spatially scalable and can therefore be extended to the entire world and to additional years, which provides encouraging prospects for the continuation of this work.