The urban traffic state is a self-organized behavior constrained by the road network structure. This study proposed an empirical approach to explain the relationship between the urban traffic state and road network structure from a spatial perspective, which was verified by real link travel speeds derived from floating car data in a typical urban road network. The characteristics of the road network structure were investigated with five quantitative indicators: hierarchy, adjacency, accessibility, efficiency, and density. A regression-kriging model was introduced to explain changes in the traffic state by incorporating the road network structure and the spatial variance caused by the significant spatial characteristics of the state of urban traffic. The model was applied to two designed typical scenarios. The results demonstrated that the model sensitively captured the change of traffic state caused by local changes in the road network structure at historical periods of city development. The process can provide robust initial values for model-driven transportation planning and is useful in practical decision making.