In spring 2020, Earth's atmosphere seemed to hold its breath. Transport ground to a halt, factories fell silent, and carbon dioxide emissions from burning fuel dropped by nearly eight percent. But spotting this "carbon echo" on a global scale proved tricky: the atmosphere is constantly mixing, forests and oceans breathe, and weather adds its own noise. Scientists set out to feel the pulse of the carbon cycle through that noise, to see whether such a brief anthropogenic disturbance could be detected.

To do this, they built a chain of models. First, they assembled a map of daily fuel-burning emissions, combining country-level data with the spatial distribution of factories and roads. Then they calculated how much carbon plants absorbed and released, and accounted for ocean exchange. All of this was fed into the GEOS-Chem atmospheric transport model, which pushed air masses around a virtual globe from January 2008 to May 2020. To isolate the anomaly, they subtracted the multi-year trend and the average seasonal cycle from each value, leaving only the deviation from the familiar pattern.

The result was clear. From February to April 2020, the CO₂ column over the band from 50°S to 50°N dropped by an average of 0.13 parts per million compared with the 2010–2019 climatology. That may seem tiny, but it was the largest sub-annual anomaly in a decade. The GOSAT satellite confirmed it: in March and April, the decline was the sharpest on record. When the authors separated the contributing factors, they found that COVID-19 accounted for 54% of the drop, the biosphere for 23%, and weather for another 23%.

At the level of individual cities, the picture was mixed. In Beijing, the expected drop of 5–10 ppm was hidden by weather fluctuations, but in Chengdu the graph showed a clear step: CO₂ concentrations fell by 10–20 ppm right on the day of the lockdown. Roadside measurements in Beijing recorded a 30 ppm reduction in near-road CO₂ during the quarantine, comparable to six out of ten cars vanishing from the streets. Thus, the global experiment imposed by the pandemic itself allowed us to test the sensitivity of our tools.

This work showed that modern monitoring systems can capture even short anthropogenic signals against natural variability. Previously, it was unclear whether such a brief CO₂ reduction caused by lockdown could be detected amid the breathing of the biosphere and the whims of weather. Now we know it can, provided we combine models and observations. The method will be useful for quickly verifying how countries meet their climate commitments, because emission reductions need to be not only promised but also seen in the atmosphere. The pulse of the carbon cycle, felt through the noise, has become more audible.