One limitation for the grid-base global climate model (GCM) is that it cannot resolve aerosol sub-grid variability, adding extra uncertainties to the aerosol forcing estimation. To guide the development of an aerosol sub-grid variability parameterization, here we analyze the aerosol sub-grid variability over the southern Pacific Ocean simulated by the high-resolution (3km horizontal grid spacing) Weather Research and Forecasting model coupled to Chemistry (WRF-chem model). We find that within a typical GCM grid, the aerosol mass sub-grid standard deviation can account for 15% of the grid-box average near the surface. The fraction can increase to 50% at the top of planetary boundary layer (PBL). The relationship between the sea-salt mass concentration, meteorological variables, and sea-salt emission rate is investigated in the clear-sky and cloudy portion. The change of aerosol sub-grid standard deviation is highly correlated to the change of vertical velocity standard deviation in the free troposphere (between 2 km to 4 km), and to the change in the standard deviation of sea-salt emission rates near the surface. In the cloudy area, sea salt aerosol mass is smaller, but higher correlation is found between the sub-grid standard deviations of aerosol mass and vertical velocity. Additionally, we find that decreasing the model grid resolution can reduce the aerosol sub-grid variability but strengthen its correlation with the total water mixing ratio (sum of water vapor, cloud liquid, and cloud ice mixing ratios). This study will form the basis of an aerosol sub-grid variability parameterization for use in global climate models.