Effects of landscape constraints on street patterns in cities: Examples from Khorramabad, Iran

The development of street patterns within cities is partly related to internal, social factors and partly to external, primarily landscape factors. Here we explore the effects of landscape constraints on the street pattern in the city of Khorramabad (population 334 thousand), West Iran. The city is located in a valley whose comparatively wide north and south parts are connected by a narrow (1-km-wide) pass. The surrounding mountains rise 400–900 m above the valley floor. These landscape constraints result in a crescent-shaped city. A total of 8481 street trends and lengths were measured, as well as the street spacing (the distance between adjacent streets), in five subareas/subsets of the city. The streets have two main trends, one perpendicular and one parallel with the city boundaries. The street-length distributions are generally power laws. On log–log plots, the distributions show abrupt changes in scaling exponent at lengths of 120 ± 20 m, suggesting different street populations. In the narrow pass between the mountains the average street spacing is much less (and the street density thus much higher) than in the other parts of the city. Using the Gibbs formula, the entropies of street populations were calculated and compared with their scaling exponents and length ranges (differences between the longest and the shortest streets). The entropies show a strong linear correlation with the scaling exponents (R2 = 0.9916) and with the length ranges (R2 = 0.8886). ► A quantitative GIS study on the effects of landscape constraints, mainly steep and irregular mountain edges, on city shape. ► Street patterns in a city reflect its shape, and thus the external landscape constraints, with implications for urban growth. ► Demonstrates that street spacing changes as a function of landscape constraints. ► Shows that different street populations (with different function) have different scaling exponents and length ranges. ► Demonstrates strong linear correlations between (Gibbs) entropies, scaling exponents, and length ranges of street populations.