Convective “butterflies” lead to tropical cyclone rapid intensification

Epitomized by the concept of the butterfly effect, atmospheric moist convection is among the most inherently unpredictable processes in the Earth’s atmosphere. Through upscale multi-scale interactions, moist convection limits the predictability of severe weather events such as tropical cyclones, thereby posing societal risks and hindering efforts to pinpoint which specific convective events trigger large-scale responses. Here we design a numerical sensitivity experiment framework that allows targeted suppression of individual convective events without altering the surrounding environment. Applying this approach to multiple tropical cyclones with different intensification histories, we find that specific small-scale convective events can shape the larger-scale intensification pathway through upscale interactions. Targeted suppression of inner-core convection inhibits convective organization, weakens warm-core development, and prevents or substantially delays rapid intensification. Our findings reveal how small-scale disturbances can have storm-scale consequences, providing a basis for more targeted observing, modeling, and data-assimilation strategies for severe weather prediction.