Multiscale oceanic precipitation extremes are determined by the morphology of rain events throughout the lifecycle of deep convective systems.

Extreme precipitation intensities in the tropics depend strongly on the spatiotemporal scale at which they are calculated, potentially introducing biases when assessing their physical drivers, impacts, and climate sensitivities. Furthermore, the contribution of Mesoscale Convective Systems (MCSs) to these extremes remains loosely constrained, especially on kilometer scales. Here, we use a new analysis framework for the co-occurrence of oceanic precipitation extremes at both convective (km) and mesoscale levels, and we compare their regional prevalence and rainfall morphology. We apply a storm tracking algorithm to ten global storm-resolving models (GSRMs) and one multi-year geostationary satellite product, focusing on various convective system types.

Our results reveal that the two scales of precipitation extremes are largely statistically independent, occurring in distinct regions with large model disagreement. Heavy km-scale events predominantly appear at the edges of convective zones, with 40% of such extremes in the satellite observations produced by MCSs. Their peak intensity is not correlated with the total area of precipitation features. In contrast, intense mesoscale events scale with the precipitating area, and are generated by MCSs in about a third of cases. We also observe a continuum of extreme precipitation features, spanning deep (DCS), very-deep (vDCS), and mesoscale convective systems.

We finally discuss the relative importance of cloud and rain morphology and life cycle parameters for understanding rain extremes on multiple scales, and we comment on relationships between environmental conditions and extreme-contributing DCS that emerge in our new multiscale analysis framework.When compared to observations, the models typically underestimate the precipitating surface and show substantial variability in the fraction of extreme rainfall attributable to different convective systems. These diagnostics highlight the need for further refining GSRMs to more accurately capture the relationship between convective organization and heavy rainfall.