Earth's climate sensitivity depends on how shallow clouds in the trades respond to changes in the large-scale tropical circulation with warming. In canonical theory for this cloud-circulation coupling, it is assumed that the clouds are controlled by the field of vertical motion on horizontal scales larger than the convection's depth (~1 km). This assumption has been challenged both by recent in situ observations, and idealized large-eddy simulations (LESs). Here, we therefore bring together the recent observations, new analysis from satellite data, and a 40-day, large-domain (1600 x 900 km2) LES of the North Atlantic from the 2020 EUREC4A field campaign, to study the interaction between shallow convection and vertical motions on scales between 10 and 1,000 km (mesoscales), in settings that are as realistic as possible. Across all data sets, the shallow mesoscale vertical motions are consistently represented, ubiquitous, frequently organized into circulations, and formed without imprinting themselves on the mesoscale buoyancy field. Therefore, we use the weak-temperature gradient approximation to show that between at least 12.5–400 km scales, the vertical motion balances heating fluctuations in groups of precipitating shallow cumuli. That is, across the mesoscales, shallow convection controls the vertical motion in the trades, and does not simply adjust to it. In turn, the mesoscale convective heating patterns appear to consistently grow through moisture-convection feedback. Therefore, to represent and understand the cloud-circulation coupling of trade cumuli, the full range of scales between the synoptics and the hectometer must be included in our conceptual and numerical models.

To advance the understanding of the interplay among clouds, convection, and circulation, and its role in climate change, the Elucidating the role of clouds-circulation coupling in climate campaign (EUREC4A) and Atlantic Tradewind Ocean-Atmosphere Mesoscale Interaction Campaign (ATOMIC) collected measurements in the western tropical Atlantic during January and February 2020. Upper-air radiosondes were launched regularly (usually 4-hourly) from a network consisting of the Barbados Cloud Observatory (BCO) and four ships within 6-16°N, 51-60°W. From 8 January to 19 February, a total of 811 radiosondes measured wind, temperature, and relative humidity. In addition to the ascent, the descent was recorded for 82 % of the soundings. The soundings sampled changes in atmospheric pressure, winds, lifting condensation level, boundary layer depth, and vertical distribution of moisture associated with different ocean surface conditions, synoptic variability, and mesoscale convective organization. Raw (Level 0), quality-controlled 1 s (Level 1), and vertically gridded (Level 2) data in NetCDF (Stephan et al., 2020) are available to the public at AERIS (https://doi.org/10.25326/137 https://doi.org/10.25326/137). The methods of data collection and post-processing for the radiosonde data set are described here.