Artificial intelligence-based monitoring of occupancy and stay duration on urban beaches: Analyzing climate influence in Benidorm (Spain)
Abstract
Accurate data on beach occupancy and its relationship with climatic factors is essential for managing public services and mitigating overcrowding in high-demand tourist destinations. This study focuses on Poniente Beach in Benidorm (Spain), where nearly 5 million beach visits were recorded between July 2023 and June 2024. Using a computer vision system based on YOLOX and ByteTrack algorithms, combined with fixed video cameras, we developed an artificial intelligence-based methodology to detect beach entries and exits and calculate occupancy and stay duration in real time. The resulting data were analyzed using Random Forest models to evaluate the influence of key climatic variables. Our findings indicate that water temperature, Heat Index, and maximum air temperature are the primary drivers of beach use. Peak occupancy exceeded 7000 simultaneous users and occurred when water temperature was above 27.5 degrees C and the Heat Index ranged between 32 degrees C and 40 degrees C, with attendance declining under more extreme heat. Average stay durations reached 2 h in summer but dropped below 30 min in winter. In contrast, wind and precipitation showed limited influence: wind only reduced attendance above 30 km/h, and short rain events (<2 h) minimally affected daily occupancy but decreased average stay. These results demonstrate the feasibility of applying AI and big data analytics to monitor and predict beach usage patterns, enabling adaptive tourism management strategies under evolving climate conditions.