Determining the association of the built environment and socioeconomic attributes with urban shrinking in Yokohama City

Abstract Urban shrinking means a densely populated urban area or city where is experiencing a significant population loss. Currently studies have examined the relationship between built environment and socioeconomic attributes and urban shrinking, however ignore the local effects. Here, we show spatially heterogeneous associations of environment and socioeconomic attributes with urban shrinking reflected by the change of VIIRS night-time light radiance during the period 2014–2019 by mixed geographically weighted regression model, after variables screening by random forest. We found that during the period 2014–2019, there were 35 km2 in Yokohama was shrinking, with most of them being mixed-use land. In general, bus stop density (BSD), road intersection density (RID), aging population (PoAP), housing price (HP), distance to the nearest park (DNP), proportion of business areas (PoBA), and proportion of private houses and flats (PoH_ private) have mixed effects on urban shrinking. Furthermore, BSD, RID, PoAP, DNP and PoH_private have negative or positive association with urban shrinking across locations, suggesting spatial heterogeneous strategies should be considered to address urban shrinking. We anticipate our study to be a start point to use mixed GWR model in shrinking city and as additions in examining the relationship between built environment and socioeconomic attributes and population loss.