From micro mobility to macro patterns: a vector field-based framework powered by a data-driven flow model for extracting coherent micro-macro traffic dynamics

The escalating complexity of urban traffic congestion makes single-scale analysis inadequate for supporting effective top-down management strategies. However, existing researches face challenges in representing continuous, interpretable macro-scale patterns without excessively compromising the fidelity of micro-scale movement details. To address this, we proposed a bottom-up vector field-based framework powered by a data-driven flow model to extract coherent macro-scale congestion patterns from micro-scale mobility. We first constructed a network-constrained demand vector field using a path flow model to capture micro-scale mobility. Building on this, we developed congestion exposure fields to characterize the direction and intensity of individual congestion exposure. To inform meaningful spatiotemporal aggregation and improve interpretability, we incorporated irrotationality theory as a scale-aware criterion, identifying scales where micro-scale mobility exhibited stable and coherent structures suitable for macro-scale representation. At these scales, a congestion potential field was derived to reveal system-wide risk dynamics. A case study in Wuhan demonstrates that the proposed framework effectively reduces micro-scale noise and highlights coherent macro-scale patterns, which offer actionable insights into local traffic complexity by serving as real-time warnings for areas prone to congestion or exhibiting increased vulnerability. This work provides an interpretable approach for selecting meaningful aggregation scales, supporting integrative assessments of micro-macro traffic dynamics.