Temporally and Spatially Adaptive Models for Estimating Transient Urban Populations: Integrating Interpolation Techniques With Spatial Criteria

Abstract

Accurate estimation of transient populations is crucial for understanding daily urban dynamics and informing effective planning. Traditional models often rely on static data and overlook temporal variations. This study introduces two new models, Temporally and Spatially Adaptive IDW (TSAIDW) and Temporal Ordinary Kriging (TOK), to improve transient population estimates by incorporating spatial characteristics and daily temporal patterns. TSAIDW uses adaptive spatial and temporal bandwidths based on residential, employment, and POI densities. Tested in three diverse Montreal neighborhoods, TSAIDW significantly reduced estimation errors compared to traditional IDW (minimum 27) and OK (minimum 15) and showed improvements over TOK. TSAIDW also allowed disaggregation of estimates to the finer Dissemination Block level. This approach offers a more flexible and accurate alternative for urban population modeling. A limitation is that the model depends on footfall camera data, which may not cover all intersections or reflect seasonal changes, suggesting the need for further data integration.