Objectives Since the opening of Shanghai City, China as a treaty port in 1843, its urban space underwent a gradual process of segregation, leading to the formation of a number of differentiated functional zones. Architectural color became a visible marker of material culture and a key indicator of spatial differen‑ce. The objective of our study is to clarify how architectural colors vary across historical functional districts, how these colors evolved, and what cultural and historical processes shaped their differentiation.

Methods Following the research paradigm of spatially integrated humanities and social sciences, we select the Bund and the Hengfu Historic and Cultural Conservation Area as the study sites. Historical maps and archival records were used to reconstruct the functional districts of the Bund and the Hengfu Historical Landscape Area. Deep learning algorithms extracted color information from large-scale street-view imagery of historical buildings. Then spatial statistics were employed to assess the relationships between color patterns and functional zoning. Finally, random forest model was used to identify the main drivers influencing color differentiation.

Results The results indicate that: (1) Historical zoning and architectural color show strong spatial correspondence. The Bund, shaped by its long-term financial and commercial functions, is dominated by cool tones. Hengfu, characterized by residential and administrative uses, shows a concentration of warm tones. (2) Functional specialization in finance, elite housing, and administration forms the primary force behind color differentiation. (3) Two diachronic mechanisms are evident: The Bund displays a pattern of “historical fixation” due to durable façade materials, public ownership, and stable financial functions; Hengfu displays a process of “color stratification,” shaped by material renewal, modernist architectural trends, and policy interventions.

Conclusions By integrating spatialized and diachronic perspectives, we find architectural color in the Shanghai concessions reflects a spatially coupled and temporally layered formation process. The integration of spatial reconstruction, street-view-based color extraction, and machine-learning interpretation provides a clear framework for understanding color variation in historical urban areas. These findings offer a foundation for protecting the authenticity and integrity of cultural heritage landscapes.