Distinct convective precipitation structures underlie similar surface rainfall and regional contrasts in rainfall extremes
Year: 2026
Type: preprint
Abstract: Conditioning radar profiles on near-surface rain rates and drop size distributions (DSDs) often obscures distinct convective structures. Instead, this study applies representation-learning-guided clustering to reflectivity-profile shapes from 1.65 million convective profiles observed by the Global Precipitation Measurement Dual-frequency Precipitation Radar, identifying six structural archetypes without near-surface inputs. They span shallow warm-rain to deep convection but overlap in near-surface precipitation properties. Despite this similarity, echo-top heights within the modal DSD bin differ threefold, reflecting contrasting ice-phase and warm-rain contributions. Across rainfall rates, these microphysical pathways diverge primarily at moderate intensities, where shallow archetypes show greater warm-layer rain-rate and drop-size increases, but converge at higher intensities as deeper archetypes show stronger drop-size growth. At the regional scale, this classification separates tropical land–ocean extreme-rainfall contrasts into opposing contributions from archetype prevalence and within-archetype extreme-rainfall probability. These findings provide structure-resolved benchmarks for evaluating convective precipitation in weather and climate models.
Source: Research Square
Institution: National University of Defense Technology
Language: English
Subfield: Atmospheric Science
Field: Earth and Planetary Sciences
Domain: Physical Sciences
Keywords: convective precipitation, radar reflectivity profiles, precipitation microphysics, droplet size distribution, warm rain processes, deep convection, extreme rainfall, tropical precipitation, representation learning, clustering, Global Precipitation Measurement, GPM Dual-frequency Precipitation Radar, echo top height, ice-phase precipitation
SDG: Climate action
Open Access status: green
Award: G5836031125
OpenAlex ID: W7225986170
