Evaluation of Arctic and Antarctic Mixed-Phase Cloud Properties in E3SMv2 Using ARM Observations
Mixed-phase clouds are ubiquitous in the high-latitude regions. The accurate representation of mixed-phase cloud properties in global climate models is crucial because of their important roles in regulating the surface radiation budget. An earlier study, using the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) satellite observation and CALIPSO simulator, showed that the low bias in simulated ice clouds at high latitudes has been largely improved in the second version of the Energy Exascale Earth System Model (E3SMv2) compared to the E3SMv1. However, the lidar signals from the CALIPSO satellite may be attenuated by the optical thick clouds, which can lead to uncertainties in the evaluation of high-latitude mixed-phase clouds. Therefore, in this study, researchers utilized the ground-based remote sensing retrievals from the Atmospheric Radiation and Measurement (ARM) program to additionally evaluate the simulated high-latitude mixed-phase clouds in E3SMv2. By consistently sampling and collocating model-simulated and observed stratiform mixed-phase clouds at the North Slope of Alaska (NSA, Arctic) and McMurdo (AWR, Antarctic) sites, they found that the simulated cloud morphology is better simulated than the phase partitioning between liquid water and ice water compared to the observations at individual sites. The observed hemispheric differences in mixed-phase cloud properties (e.g., frequency of occurrence, cloud top temperature, cloud top height, cloud base height, and cloud thickness) between the NSA and AWR site are reasonably captured by E3SMv2, but not for the cloud phase.
By developing a novel sampling and collocation method to consistently compare the E3SMv2 simulated and ARM retrieved stratiform mixed-phase cloud properties, researchers identified that the simulated cloud morphology and their hemispheric difference are better simulated than the cloud phase partitioning between liquid and ice compared to observations. In particular, E3SMv2 is found to overestimate observed cloud liquid water and underestimate cloud ice water at the NSA site when compared to the ARM ground-based retrievals. Although the positive bias in liquid water is consistent with the earlier evaluation results using CALIPSO satellite data, the underestimation of cloud ice is in contrast to the earlier results in the Arctic. The difference is mainly caused by the different detection capabilities between space-borne and ground-based instruments. This study demonstrates the importance of considering complementary capabilities of different instruments when evaluating mixed-phase cloud properties in the model.
This study evaluated the simulated stratiform mixed-phase cloud properties in E3SMv2 utilizing one-year-long ground-based remote sensing retrievals at the ARM NSA site and McMurdo station (AWR). The nudging approach was applied to the E3SMv2 simulation for a constrained comparison with the ARM observations. Modeled stratiform mixed-phase clouds are consistently sampled using a similar definition as in the ARM data, and the simulated stratiform mixed-phase cloud samples are then collocated with the observations in order to perform a consistent evaluation between the model and observations, and between two ARM sites. This study showed that simulated cloud boundary properties such as cloud top temperature, cloud top height, and cloud base height are better simulated at the NSA site than the AWR site. The hemispheric differences in cloud boundary properties such as the warmer cloud top temperature, lower cloud top height, and lower cloud base height at the NSA than AWR are also reasonably captured by E3SMv2. However, large biases were found in the simulated cloud phase. The model largely overestimates liquid water path and underestimates ice water path at the NSA site. On the other hand, liquid water path is frequently underestimated at the AWR site. As a result, the observed hemispheric difference in cloud phase partitioning is misrepresented in E3SMv2. This study provided complementary insights in the cloud phase evaluation in E3SMv2 using ARM ground-based remote sensing retrievals, in addition to the earlier study that only utilized the CALIPSO satellite data.