Evaluating Mesoscale Convective Systems Over the US in Conventional and Multiscale Modeling Framework Configurations of E3SMv1

Organized mesoscale convective systems (MCSs) contribute a significant amount of precipitation in the Central and Eastern US during spring and summer, which impacts the availability of freshwater and flooding events. However, current global Earth system models cannot capture MCSs well and misrepresent the statistics of precipitation in the region. In this study, we investigate the representation of MCSs in three configurations of the Energy Exascale Earth System Model (E3SMv1) by tracking individual storms based on outgoing longwave radiation using a new application of TempestExtremes. Our results indicate that conventional parameterizations of convection, implemented in both low (LR; similar to 150 km) and high (HR; similar to 25 km) resolution configurations, fail to capture almost all MCS-like events, in-part because they underestimate high-level cloud ice associated with deep convection. On the other hand, the multiscale modeling framework (MMF; cloud-resolving models embedded in each grid-column of similar to 150 km resolution E3SMv1) configuration represents MCSs and their annual cycle better. Nevertheless, relative to observations, the E3SMv1-MMF spatial distribution of MCSs and associated precipitation is shifted eastward, and the diurnal timing is lagged. A comparison between the large-scale environment in E3SMv1-MMF and ERA5 reanalysis suggests that the biases during the summer in E3SMv1-MMF are associated with biases in low-level humidity and meridional moisture transport within the low-level jet. The fact that conventional parameterizations of convection, even with high-resolution, cannot capture MCSs over the US suggests that methods with explicit representation of kilometer-scale convective organization, such as the MMF, may be necessary for improving the simulation of these convective systems. Organized thunderstorms, known as mesoscale convection systems (MCSs), can contribute to large amounts of rainfall and flooding in the United States, especially during the summer and spring. However, current Earth system models have difficulties representing the complex physical processes that are important to these storm systems, which occur at smaller scales than the models resolve. In this study, we assess the ability of the Energy Exascale Earth System Model v1 to simulate MCSs using two advanced configurations (high-resolution and multiscale modeling framework), and compare to the conventional low-resolution configuration and observations. To identify MCS-like events, we implemented a tracking method based on only the outgoing longwave radiation, which we apply to assess MCS characteristics and associated precipitation. Our results suggest conventionally parameterized configurations, with low- (similar to 150 km) and high- (similar to 25 km) resolution grids, fail to capture MCSs, due in part to an under-simulation of high-level cloud ice. On the other hand, the multiscale modeling framework configuration (i.e., kilometer-scale cloud-resolving models embedded in a standard similar to 150 km resolution global grid) better simulates MCSs and related characteristics. Nevertheless, we identify biases in the location and timing of simulated MCSs and associated rainfall, which can be linked to biases in simulating the large-scale environment during summer. Energy Exascale Earth System Model (E3SMv1) with conventional parameterizations of convection underestimates high-level cloud ice and fails to capture spring and summer mesoscale convective systems (MCSs)E3SMv1 with the MMF configuration captures MCSs, but has some biases, including an eastward shift in location and lag in diurnal timingE3SMv1-MMF MCS-precipitation biases are linked to biases in low-level humidity and moisture transport associated with the low-level jet