We used satellite data and machine learning to better understand how tiny particles in the atmosphere affect clouds and their brightness. At higher spatial resolution, we discovered a new “M”-shaped pattern in the relationship between cloud water and droplet concentration unlike the inverted-V shape observed at coarsely gridded scales. Cloud water increases more with droplet concentration when rain is present. These findings support the development of next-generation atmospheric models.