Modelling Wind Direction for Commuting Cycling Behaviour in Rotterdam & Gothenburg

This study assesses the effect of weather on commuting cycling behaviour. Like previous work, the relationship between various meteorological variables (air temperature, precipitation, wind speed) and commuting cycling behaviour variation (during rush hour) is tested using multiple linear regression analyses. Unlike previous work, it includes wind direction in its analysis. Additionally, the analysis uses ‘reference’ wind direction data from meteorological stations as well as modelled ‘locational’ wind direction data. Using URock (2023a), an open-source diagnostic wind model, this study models locational wind direction in a radius around specific segments of bicycle lanes since reference wind direction data frequently does not accurately represent specific wind conditions in the urban environment. This is done for eleven locations in Rotterdam (the Netherlands) and Gothenburg (Sweden) to account for geographical heterogeneities between and within these cities. The results indicate that air temperature, precipitation, and wind speed are significant variables explaining commuting cycling behaviour variation in both cities, and in both morning and afternoon rush hours. Commuting cyclists in Gothenburg, however, are found to be disproportionally more weather sensitive than their counterparts in Rotterdam. Although descriptive statistics show that locational wind direction values predictably provide a more intuitive image of local wind conditions than reference wind direction data, testing the effect of wind direction on commuting cycling behaviour variation yielded no consistent results, and no consistent difference between locational and reference wind direction values could be discerned.