Traditional landscape design is mainly based on the personal experience and actual evaluation of designers, and cannot effectively and quickly complete design updates and iterations. To study and propose a new digital landscape design method that optimizes park path system design by combining spatial theory and landscape elements. The new method uses a convex space model to construct a model of the park path system, treating specific roads as convex spaces and the grasslands, fences, and shrubs on both sides of the roads as convex space boundaries. Using a two-step spontaneous geolocation data platform to visualize park path system data. Visualize the path system of the park landscape using digital technology. Introduce five characteristic elements of park landscape space and determine the relationship between the main landscape elements and spatial theory. Analyze the qualitative relationship between these elements and spontaneous geolocation data through linear regression analysis and correlation testing, and conduct simulation experiments. In the simulation experiment, the corresponding selectivity of the flower show area was 0.97 at a distance of 40 m. The path selectivity corresponding to the parent-child area was second only to the flower show area, with a specific value of 0.89. The path selectivity for these three areas of bamboo pavilion, badminton, and basketball was not significantly different, all located around 0.75. As the distance increased, the selectivity value gradually decreased, and the changing magnitude gradually decreased. This research method has significant effectiveness and feasibility in the important design parameters of parks. This can provide certain reference value for landscape architecture design.