The quality of urban green space has an impact on the health and well-being of populations. Previous studies have shown that consideration of crowd activity characteristics is the key premise of landscape space design and planning. However, there is limited research on the correlation between features of the spatial layout of Mini Urban Green Spaces (MUGS) and the behavior of people, and it is difficult to take into account the possible distribution of people and their activity characteristics during the design phase of MUGS. This study aims to construct a technical workflow utilizing the AnyLogic platform and agent-based simulation methods for analyzing the characteristics of landscape spatial layouts considering dynamic human behavior. One MUGS, named 511 Park in Nanjing, China, was selected as the case for the application of the method and exploration of the impacts of spatial elements and layout on crowd activity types and density. We investigated the impact of four types of spatial elements—paths, facilities, nodes, and entrances—on human activities in MUGS. The results showed that path layout emerged as the most significant influencing factor. Changes in nodes and the number of facilities have a relatively minor impact on people’s activities. There was an apparent impact of changes in path orientation around nodes on the dynamics of the flow of people. This study could provide valuable insights for landscape designers, aiding informed decision-making during the construction, renovation, and management of MUGS.

The visual attributes of urban green spaces influence people’s perceptions, preferences, and behavioural activities. While many studies have established correlations between landscape perception and visual attributes, they often focus on specific timeframes and overlook dynamic changes in the spatial form of urban green spaces. This study aims to explore the long-term changes in the visual attributes of urban green spaces. We propose a method to quantitatively analyse changes in visual attributes using point clouds to simulate visual interfaces. Using an unmanned aerial vehicle, we conducted a five-axis tilt photography survey of Qinglvyuan Park in Nanjing, China, in August 2018 and September 2023. Point cloud models were generated for the two periods, and five visual attribute indicators, openness (OP), depth variance (DV), green view ratio (GVR), sky view ratio (SVR), and skyline complexity (SC), were analysed for long-term changes. The results indicate that OP, DV, and SVR decreased after five years, while GVR increased. The maximum increase in GVR was 26.6%, and the maximum decrease in OP was 12.8%. There is a positive correlation between GVR and its change (d_GVR). Conversely, there are negative correlations between SC and its change (d_SC), as well as between SVR and d_GVR. Tree growth emerged as a primary factor influencing changes in the visual attributes of urban green spaces. This study highlights the importance of adopting a long-term and dynamic perspective in visual landscape studies, as well as in landscape design and maintenance practices. Future research on predicting long-term changes in the visual attributes of urban green spaces should focus on understanding the relationships between tree properties and environmental conditions.