Decoding Built-Environment Impacts on Freight Truck Speeds Through Spatial Regression: A Case Study of Valparaíso, Chile

Freight mobility in Latin-American port cities is often constrained by steep topography; irregular street patterns and mixed land uses that jointly depress truck operating speeds. This research investigates how the built environment influences freight truck speed in port cities, using the case of Valparaíso, Chile. We integrate nine months of high-frequency GPS data from heavy-duty vehicles with multiscale spatial regression techniques, notably the Multiscale Geographically Weighted Regression (MGWR) model. Urban predictors include road hierarchy, land use, slope, population density, and distance to port terminals. The MGWR model outperforms both traditional Ordinary Least Squares (OLS) and single-scale Geographically Weighted Regression (GWR) in all performance metrics. It achieves adjusted R² values between 0.4816 (May) and 0.5854 (February), consistently surpassing GWR (0.3026 to 0.4032) and OLS (maximum of 0.1889). In terms of AICc, MGWR registers values as low as 637.2 (July) and never exceeds 732.3 (January), compared to OLS which ranges from 2321.4 to 2617.3. The spatial variation in coefficients reveals strong non-stationarity: service streets and distance to port consistently enhance speed, while residential roads and population density exert negative effects in key corridors. These insights support the development of locally targeted freight policies. The approach is transferable to other port cities seeking to balance freight efficiency with urban livability.