Spatiotemporal Interpolation Method for Population Flow Data in Urban Areas

Abstract. Capturing the intricate dynamics of population movements in urban areas holds substantial implications for urban planning and management, particularly in the context of disaster mitigation. There is an attempt to introduce methods for estimating the spatiotemporal distribution of population flows, leveraging demographic data from various kind of sources. In earlier spatiotemporal interpolation methods, some key assumptions were made to obtain data at shorter intervals. In this study, we present an alternative spatiotemporal interpolation method by loosening assumptions and increase its versatility and facilitates flexible application across various contexts and objects. This is achieved by estimating the square root of the movement probability matrix for longer time intervals. The efficacy of our approach is demonstrated through its application to actual data from the Tokyo 23 wards, allowing for the estimation of the spatiotemporal distribution of population flows across various time intervals. Our results not only affirm the accuracy of the estimates but also provide insights into the intricate population flows within the densely populated regions of the Tokyo 23 wards. Moreover, by estimating population data at shorter time intervals, we explore the characteristics of these flows, offering an understanding of the dynamics that shape urban demography.