Mapping nationwide 1 Km hourly human mobility (2022–2024) across South Korea using multi-source data

Population mobility is a fundamental driver of urban development and socioeconomic processes, providing a critical foundation for scientific analysis and evidence-based decision-making. Despite its importance, most existing studies and applications continue to rely on static resident (de jure) population data, which fail to represent the spatiotemporal dynamics of human activity. In this study, we develop a comprehensive framework to map nationwide hourly floating (de facto) population across South Korea at 1 km spatial resolution for the period 2022–2024 by integrating mobile phone–based records with a machine-learning approach. The proposed framework incorporates a diverse set of spatial and temporal predictors, including building characteristics, land cover, accessibility indicators, temporal variables, and meteorological conditions, enabling the model to capture spatial heterogeneity and short-term temporal variability that are often neglected in prior population mapping studies. Quantitative evaluation demonstrates strong model performance, with leave-one-year-out cross-validation yielding high temporal generalization (Pearson correlation coefficients r = 0.981–0.993) and spatial cross-validation across unseen metropolitan regions indicating robust transferability (r = 0.847–0.941). Qualitative analyses reveal interpretable and realistic mobility patterns, including elevated nighttime populations in residential areas and pronounced daytime peaks in commercial and industrial zones, particularly around midday (13:00). The model also successfully captures city-specific mobility responses, holiday-related population redistribution, and weather-dependent behavioral changes, such as increased residential presence on rainy days. These results represent a substantial improvement over conventional static population datasets and existing floating population products. By resolving hourly variations in human presence at high spatial resolution, the resulting dataset provides a valuable resource for a wide range of applications, including urban management (e.g. traffic modeling and infrastructure planning) and environmental risk assessment (e.g. disaster preparedness and wildfire prevention). Overall, this work offers new insights into human mobility dynamics and supports more informed analyses of environmental sustainability and urban resilience.