A Multi-Scale Workflow for Analysing the Urban Morphological Spectrum: A Comparative Analysis of Three Mid-Sized Cities

Abstract

Traditional urban morphological analyses are structurally limited in terms of both systemic diversity and empirical scale due to reliance on manual methods. In contrast, decoding the complex fabric of rapidly growing cities necessitates data-driven, scalable approaches. To address this gap, this study proposes a multi-scale pipeline that classifies urban form systematically and reproducibly from open spatial data, and applies it comparatively to three Anatolian cities of contrasting typo-morphological character: Elaz & imath;g, Erzincan and Mardin. From street networks and building geometries, an integrated morphometric matrix was assembled by computing network topology and orientation metrics, space syntax configurational accessibility, morphological tessellation, coverage area ratio, building form-volume indicators, neighbourhood and adjacency measures, and Local indicators of spatial association (LISA) spatial autocorrelation. After transformation and standardisation, urban typologies were derived through Principal Component Analysis and grouped with spatially weighted k-means (Geo-KMeans). Three findings stand out. First, the cities trace a distinct morphological spectrum that runs from Mardin's organic historical fabric to Erzincan's relatively planned grid structure, with the radial polarised Elaz & imath;g occupying an intermediate, transitional position between the two. Second, accessibility and built density prove only weakly related (r = 0.05-0.25). Third, six, five and four morphological typologies emerged, triangulated against LISA hot spot clusters and space syntax maps. Overall, this reproducible framework offers planners a systematic, data-driven basis for exploratory morphological assessment rather than a definitive, universal typology.