How Industry Types Shape Commuting and Non-Commuting Network Centrality: A Geographical Detector Analysis of Shanghai and Its Adjacent Areas

The polycentric spatial structure of metropolitan areas has become a frontier topic in urban studies. Yet existing research has largely remained descriptive regarding commuting networks, with limited investigation into how different industry types are associated with commuting and non-commuting centers. Taking Shanghai and its adjacent areas as a case study, this research constructs grid-scale (425 m × 425 m) origin–destination networks for commuting and non-commuting trips using mobile phone GPS data, and uses weighted in-degree centrality to identify the two types of centers. Above-scale enterprise data are classified into six industry types, and the factor and interaction detectors of the geographical detector are employed to quantify their explanatory power and pairwise interactions. The results show that (1) commuting centers are highly concentrated in Shanghai’s central city and form an employment belt extending from Jiading through Kunshan to Suzhou, whereas non-commuting centers are more dispersed in suburban areas; (2) for commuting centers, producer services and consumer services both exhibit strong explanatory power (q = 0.392 and 0.380, respectively), while for non-commuting centers on the surveyed Saturday, consumer services show the strongest explanatory power (q = 0.249); (3) manufacturing shows weak global explanatory power (q = 0.118) but forms localized high-intensity centers in advanced manufacturing clusters; and (4) all interactions are of the two-factor enhancement type, with producer–consumer services having the strongest effect on commuting centers (q = 0.477) and consumer services–real estate dominating non-commuting centers (q = 0.284). These findings suggest that industry co-location is prevalent and closely associated with urban mobility networks. The study offers a division-of-labor perspective, extends the geographical detector to travel network research, and provides implications for regional integration planning.