Kilometer‐Scale Modeling of Atmospheric River Precipitation Using the Regionally Refined SCREAM Model

Abstract

This study evaluates kilometer‐scale simulations of atmospheric river (AR) precipitation using the Simple Cloud Resolving E3SM Atmosphere Model (SCREAM). Four AR events that cause flooding and impact the Puget Sound basin in the U.S. Pacific Northwest (PNW) are simulated using SCREAM with a regional refined mesh (RRM) at 3.125 km over the PNW within a global domain at 25 km resolution. Two SCREAM configurations are evaluated that differ only in their representation of ice‐cloud fraction: the original version (SCREAMv0) uses a relative humidity‐based formulation, whereas the revised version (SCREAMv0+) uses the cell‐averaged ice mass mixing ratio. For all AR cases, SCREAMv0+ simulates stronger lower‐level wind, more precipitable water, and increased moisture flux than SCREAMv0, aligning more closely with ERA5 reanalysis. With realistic large‐scale circulation, SCREAMv0+ increases precipitation over the Cascades and decreases it over the Coastal Range, leading to a closer resemblance of the spatial rainfall distribution according to multiple observational products. SCREAMv0+ shows a deeper layer of cloud ice and mixed‐phase hydrometeors than the SCREAMv0 simulations. It is likely that an altered latent heat release, caused by changes in the representation of ice‐ and mixed‐phase mass ratios, causes a large‐scale circulation response that eventually affects the precipitation distribution. Sensitivity experiments confirm these changes are mainly driven by how ice‐cloud fraction interacts with microphysics. Results show that altering cloud macrophysics can significantly change the connection between microphysics, diabatic heating, and large‐scale circulation in kilometer‐scale models. These findings emphasize the importance of cloud mixed‐phase processes in AR precipitation and demonstrate how regionally refined models like SCREAM can well simulate cold‐season extreme rainfall in complex terrain.