Global simulations are often found to overpredict the super-cooled liquid with little ice production in the Arctic mixed phase clouds. Previous studies have highlighted the importance of continuous evaluation and improvement of cloud microphysics in climate models. To identify the meteorological and aerosol influences on the simulated high-latitude cloud properties, we conducted a long-term (2004-2014) regime-based analysis of the high-latitude clouds at the North Slope of Alaska (NSA) using the DOE ARM observations, ERA5 reanalysis and E3SM model simulations (with and without high-latitude dust sources). Using the ARM surface meteorological data at NSA, four categories with distinct combinations of the meteorological properties are identified. These categories were shown to correlate with synoptic meteorological regimes over the Arctic-wide region derived from ERA5 and different cloud phase and microphysical properties in the ARM observations at NSA. These synoptic regimes cannot be re-produced by E3SM based on the daily surface meteorology data, but four similar clusters are identified by a k-means classification of the model daily outputs, suggesting that E3SM captures the climatological occurrences of the synoptic regimes. Seasonal composite of the frequency of occurrence for the four regimes shows that the warm and moist biases in E3SM during the winter and spring time at NSA are associated with the less frequent model predictions of the Arctic anticyclones. This is consistent with the model overestimation of cloud liquid water and occurrences of the mixed-phase clouds, compared with the ERA5 and ARM cloud properties. In contrast, E3SM simulates the dominate mode of the local meteorology and synoptic regimes occurring at NSA in the summer and fall time, but still overpredicts the cloud liquid water content. Inclusion of the high-latitude dust sources in E3SM is shown to increase the ice nucleating particles in summer and fall, leading to higher ice water content. Overall, the regime-based analysis in the present study helps identify the model bias in the Arctic cloud properties that is linked to uncertainties in simulating both the large-scale meteorological systems, which can be largely improved through nudging to the reanalysis, and cloud microphysics related to the ice-nucleating particles.