Cluster-Based Strategic Planning and Capacity Optimization of Assembly Areas in Disaster Management

Planning emergency assembly areas, which are vital in the first hours after a disaster, is a critical decision problem, especially in metropolitan areas with high population density and urbanization pressure. This study comprehensively evaluates the capacity adequacy, population pressure, and spatial imbalances of existing assembly areas in the Eyüpsultan district of Istanbul. A two-stage hybrid model is proposed: in the first stage, assembly areas are clustered based on geographical proximity using the K-means algorithm; in the second stage, an improvement is implemented for clusters falling below the per capita area target (3 m²/person), based on the reallocation of parks from neighboring clusters with excess capacity through administrative assignment. As a result of trials with different cluster numbers, the five-cluster configuration was selected as the preferred solution when the Silhouette score (0.568), minimum per capita area (3.21 m²), and spatial continuity criteria were evaluated together. Thanks to the proposed hybrid model, the per capita space in all clusters has been kept above the 3 m² threshold. These findings demonstrate that capacity gaps can be addressed solely through the rational optimization of service assignments, without increasing the urban open space stock. Consequently, the proposed approach offers a more equitable, accessible, capacity-balanced decision-support structure for disaster management for local governments.