Influence of Horizontal Model Resolution on the Horizontal Scale of Extreme Precipitation Events

A fundamental characteristic of extreme precipitation events (EPEs) is their horizontal scale. This horizontal scale can influence the intensity of an EPE through its effect on the timescale of an EPE as well as its effect on the strength of convective feedbacks. Thus, to have confidence in future projections of extreme precipitation, the horizontal scales of EPEs in global climate models (GCMs) should be evaluated. Analyzing daily output from 27 models participating in the Coupled Model Intercomparison Project phase 6 (CMIP6), including 13 models participating in the High Resolution Model Intercomparison Project (HighResMIP), we computed the horizontal scales of EPEs and extreme ascent for annual maximum EPEs during 1981–2000. We found that the horizontal scales of both EPEs and the associated ascending motion are resolution‐dependent: for a factor of seven increase in horizontal resolution, the horizontal scale decreases by a factor of approximately two to five, with higher sensitivity in the tropics than in the midlatitudes. Further analysis in the southern hemisphere midlatitudes reveals that this resolution dependence results from precipitation during the simulated EPEs that is almost entirely resolved rather than parameterized. However, the EPEs are not simply grid box storms, and analysis of the horizontal scales of geopotential anomalies suggests that the planetary‐scale dynamics in GCMs is not resolution‐dependent. Thus, the dominance of resolved precipitation during EPEs is more likely due to convection on the model grid or formation of strong, poorly resolved fronts, and additional work is needed to explore these possibilities and find a remedy for this resolution dependence. Plain Language Summary Extreme precipitation is of great human interest because of the potentially severe flooding that it can cause. We are reliant on global climate models (GCMs) to project future changes in extreme precipitation, so there is a need to assess the realism of extreme precipitation in GCMs. One fundamental characteristic of an extreme precipitation event (EPE) is its horizontal size or “scale.” In this study, we analyze a large archive of output from simulations using state‐of‐the‐art GCMs and compute the horizontal scale of EPEs during 1981–2000. We found that the horizontal scale depends on the spacing between grid points in the model. These grid points are like pixels in a digital photo, and the smaller the distance between grid points, the higher the mod