Understanding intercity travel distribution: A quantitative explanation approach using dual-layered zone delineation

Urban agglomerations are national growth engines, yet surging intercity travel intensifies congestion and CO2 emissions, making sustainable land-use-mobility alignment imperative. Recent national and regional policies in China emphasize agglomeration-scale transport integration, yet the mechanisms generating intercity travel within urban agglomerations remain insufficiently specified and weakly quantified. We propose a dual-layer delineation that fuses multi-source features to construct traffic analysis zones (TAZs) that are contiguous, functionally coherent, and demand-balanced. Combined with an interpretable deep-learning framework—Kolmogorov-Arnold Networks (KAN) with SHapley Additive exPlanations (SHAP)—the approach quantifies nonlinear built environment effects on intercity travel. In the Beijing-Tianjin-Hebei urban agglomeration (BTH-UA), the delineation reduces the dependent variable's coefficient of variation by 24.01% and increases between-zone heterogeneity among explanatory variables eightfold, enabling fine-grained management. KAN also yields equation-level representations that accord with SHAP attributions. Distance and origin/destination trip volumes emerge as dominant drivers, while socioeconomic, transport, and land-use attributes exert significant nonlinear effects. The framework identifies priority travel areas and their travel distribution mechanisms, and supports demand-concentrating, high-capacity public transport services, advancing sustainable, coordinated land-use and intercity transport planning.