<p>Residential areas are the primary location for people’s daily activities in cities, and green spaces are essential for creating a healthy environment. This study examines clustered green spaces in high-density residential areas of Xi’an, investigating how their landscape patterns influence PM2.5 concentration distribution. First, we conducted field surveys to characterize the landscape patterns of green spaces in 18 representative residential areas. Second, using landscape ecology, we selected six groups of typical and comparative spatial pattern indicators of cluster green spaces. Finally, we employed ENVI-met software for numerical simulations to analyze how different landscape patterns influence PM2.5 distribution within green spaces. The results show that the concentration and size of green patches are negatively correlated with the overall PM2.5 concentration in the green space. When the area of cluster green patches is limited, the smaller the shape index of green patches, the lower the PM2.5 concentration in the green space is. The fragmentation and porosity among green spaces increase the landscape heterogeneity and instability within the green space. The larger the dispersion and porosity, the higher the PM2.5 concentration in the green space is. If the cluster green space in the residential area is a multi-patch configuration with regularly distributed layout, the overall PM2.5 concentration in the green space patch is low. Therefore, this study provides reasonable and optimized design strategies and scientific and rational index suggestions for the landscape pattern of residential cluster green spaces, aiming to provide a pleasant outdoor activity space for urban residents.</p>