Land use regression (LUR) modeling has been applied to study the spatiotemporal patterns of air pollution, which when combined with human space-time activity, is important in understanding the health effects of air pollution. However, most of these studies focus either on the temporal or the spatial domain and do not consider the variability in both space and time. A temporally aggregated model does not reflect the temporal variability caused by traffic and atmospheric conditions and leads to inaccurate estimation of personal exposure. Besides, most studies focus on a single air pollutant (e.g., O3, NO2, or NO). These pollutants have a strong interaction due to photochemical processes. For studying relations between spatial and temporal patterns in these pollutants it is preferable to use a uniform data source and modelling approach which makes comparison of pollution surfaces between pollutants more reliable as they are produced with the same methodology. We developed temporal land use regression models of O3, NO2 and NO to study the co-variability of these pollutants and the relations with typical weather conditions over the year. We use hourly concentrations from the measurement network of the Dutch National Institute for Public Health and the Environment and aggregate them by hour, for weekday/weekend and month, and fit a regression model for each hour of the day. 70 candidate predictors that are known to have a strong relationship with combustion-related emissions are evaluated in the LUR modelling process. For all pollutants, the optimal LUR was identified with 4 predictors and the temporal variability was determined by the explained variance of each temporal model. Our temporal models for O3, NO2, and NO strongly reflect the photochemical processes in space and time. O3 shows a high background value throughout the day and only dips in the (close) vicinity of roads. The diminishing rate is affected by traffic intensity. The NO2 LUR is validated against NO2 measurements from the Traffic-Related Air ...