A new study published in Atmospheric Chemistry and Physics exploits the recently launched EarthCARE satellite to dissect these transitions with a level of detail that previous missions could not achieve. Led by Johanna Mayer of the European Space Agency’s ESRIN centre, together with colleagues at ESA and the European Centre for Medium-Range Weather Forecasts, the team combined EarthCARE’s collocated active and passive instruments with geostationary imagery from the GOES satellites to reconstruct, hour by hour, what happens inside a cloud deck in the roughly one day before it breaks apart. Their central finding is that the seeds of the transition are sown long before the pattern visibly changes: liquid water and rain begin to build up in closed cells as much as 25 hours ahead of the breakup.

EarthCARE, launched in May 2024, carries three instruments that work in concert. The Multi-Spectral Imager provides wide-swath views of cloud structure, while the Atmospheric Lidar and the Cloud Profiling Radar probe the vertical column beneath. The radar is a significant advance over its predecessor on CloudSat: it is more sensitive, resolves finer vertical and horizontal scales, suffers far less from surface clutter, and, crucially, is the first spaceborne Doppler radar, meaning its velocity measurements constrain raindrop fall speeds and therefore raindrop sizes. Together with a synergistic retrieval product called ACM-CAP, which unifies clouds, aerosols and precipitation into a single physically consistent estimate, these capabilities allow scientists to measure light drizzle near the ocean surface, a regime that had long eluded satellites.

To classify cloud patterns automatically, the researchers applied a convolutional neural network, fine-tuned for both EarthCARE’s imager and GOES imagery, to 128-kilometre-square cloud scenes. The network assigns each scene to one of six mesoscale categories, including closed and open mesoscale cellular convection, with F1 scores reaching 0.86 to 0.94 for the two key classes. Applying the classifier to more than fifteen months of data across four subtropical stratocumulus regions, the Southeast Pacific, Northeast Pacific, Southeast Atlantic and Northeast Atlantic, the team assembled a dataset of roughly 5,000 closed-cell and 2,500 open-cell scenes, each tagged with detailed microphysical retrievals.

The contrasts between the two regimes are stark. Open cells contain far fewer cloud droplets, with number concentrations typically between 10 and 100 per cubic centimetre, and show much greater variability in both droplet size and liquid water path. They rain more often and more heavily, and their raindrops are larger, which matters because bigger drops are more likely to reach the surface, where they can scavenge aerosols and further deplete the cloud condensation nuclei that droplets need to form. Perhaps the most surprising result concerns closed cells: about 60 percent of them drizzle, a frequency that earlier instruments largely missed. The team attributes this detection to EarthCARE’s improved sensitivity, and notes that field campaigns may have reinforced the mistaken notion that closed cells do not rain simply because they tended to sample shallow, near-coastal clouds.

One subtlety emerged from the comparison with previous work. Earlier satellite studies using passive imagers generally found larger average droplet sizes in open cells, but the EarthCARE synergy retrieval found similar mean sizes with much greater variability. The explanation lies in what each type of instrument sees: passive imagers preferentially sample the optically thicker parts of broken cloud fields, while the active-sensor retrieval also detects the thin, small-droplet cloud layers that thread between open cells. When the team restricted their statistics to optically thicker profiles, their results converged with the passive picture, resolving the apparent discrepancy.

The heart of the study, however, is its treatment of transitions. Because EarthCARE is a polar-orbiting satellite that revisits any given region only about every 25 days, a single overpass cannot capture an evolving cloud scene. The team therefore computed wind trajectories driven by ERA5 reanalysis winds, forward and backward 30 hours from each EarthCARE overpass in the Southeast Pacific, and classified cloud structure along those trajectories using GOES imagery every two hours. In total they identified 803 closed-to-open transitions, 330 observed as closed cells at the overpass and transitioning later, and 473 traced back from open cells to an earlier closed state. For nighttime transitions, when optical-thickness retrievals are unreliable, they pinpointed the timing using the 11.2-micrometre brightness temperature, which rises sharply as cloud fraction collapses.

Plotting cloud properties against the time remaining before transition revealed a coherent and physically telling sequence. Liquid water path begins to climb about 25 hours before the transition, peaking near the breakup and falling afterwards as the thin interstitial layers of open cells appear. Rain water path rises in parallel, and a decomposition shows that the earliest phase of this increase, from roughly 25 to 15 hours before transition, is driven mainly by more of the cloud raining, while the later phase reflects intensifying rain rates. Droplet number concentrations then begin to fall and droplet sizes to grow, starting around 20 hours before transition, while raindrop sizes increase on a similar schedule. Notably, cloud-top and cloud-base heights barely move throughout, and the microphysical fields remain spatially homogeneous, with variability typical of closed cells, until only a few hours before the pattern actually flips.

From this timeline the authors build a physical narrative. Enhanced boundary-layer moisture, plausibly supplied by strong surface winds and warm seas, thickens the cloud and boosts its liquid water, which in turn feeds precipitation. Once rain reaches the surface it scavenges aerosols, cutting droplet numbers and enlarging droplets, which makes rain formation still more efficient, a self-amplifying rain-aerosol feedback. Eventually the rain and its evaporative cooling beneath cloud base stabilize the sub-cloud layer and generate cold pools whose collisions trigger the updrafts that reorganize the deck into open cells. Importantly, the observations suggest that low aerosol availability is not the initial trigger, since droplet concentrations stay near climatological values until after liquid water and rain have already risen; aerosol depletion appears to develop as part of the feedback rather than preceding it. The team also observed a pronounced diurnal cycle, with roughly 60 percent of transitions occurring at night and a peak around dawn, consistent with nighttime radiative cooling driving liquid water and precipitation to their daily maximum.

The implications reach into climate policy as much as cloud physics. Anything that raises boundary-layer moisture, stronger winds, warmer sea surfaces, could push stratocumulus decks toward open cells more often, reducing their albedo and weakening their cooling effect. Conversely, abundant aerosols can delay or suppress rain formation and thereby postpone breakup, which means that declining aerosol emissions worldwide may produce a warming influence by allowing closed decks to collapse into open cells earlier along their trajectories. The authors caution that their analysis is limited to daytime retrievals and subtropical regions, that the neural-network classification and the still-maturing EarthCARE products carry uncertainties, and that meteorological factors such as inversion strength likely modulate the timeline. Even so, the emergence of smooth, coherent trends across hundreds of independently detected transitions gives confidence that the signal is real. What the study delivers is something climate models have lacked: an observationally grounded, hour-by-hour chronology of how rain dismantles one of the planet’s most important cooling systems, over roughly a day of slow, self-reinforcing change.

Subject of Research: Precipitation-driven transitions from closed to open marine stratocumulus cells observed by the EarthCARE satellite

Article Title: EarthCARE reveals details on the role of rain in closed-to-open cell transitions

Article References: EarthCARE reveals details on the role of rain in closed-to-open cell transitions. (n.d.). https://doi.org/10.5194/acp-26-14085-2026

Image Credits: AI Generated

DOI: 10.5194/acp-26-14085-2026

Keywords: EarthCARE, stratocumulus, closed cells, open cells, precipitation, drizzle, cloud microphysics, aerosol scavenging, cloud albedo, GOES, convolutional neural network, marine boundary layer