Explaining the urban mechanisms behind OpenStreetMap updates: Evidence from 88 German cities

OpenStreetMap plays an increasingly important role as a global source of urban spatial data. It exhibits content incompleteness and uneven updating activity, but existing research offers only limited insight into why cities with similar urban characteristics display different updating dynamics. Understanding these processes is essential for interpreting how OpenStreetMap is maintained and for designing strategies to reduce persistent mapping inequalities. The black-box nature of machine learning-based spatial modeling acts as a barrier and can obscure the mechanisms underlying observed spatial patterns. Explainable artificial intelligence (XAI) can partly mitigate this limitation; however, without explicit geovisualization, its conclusions could overlook spatial heterogeneity. We propose an analytical framework that integrates Geovisual XAI (GeoVisX) with clustering analysis to explain how urban characteristics are translated into distinct OSM updating mechanisms across cities. Drawing on data from 88 German cities, we examine how urban characteristics shape updating activity. Results show that updating activity is not driven by any single attribute. Instead, it emerges from nonlinear, contextual interactions among functional intensity, spatial structure, and population composition. Clustering cities using XAI-derived features identifies three distinct urban updating mechanisms that are (1) location-driven, (2) facility-decoupled residential maintenance driven, and (3) facility- and activity-reinforced. Overall, OSM updating inequality reflects heterogeneous urban mechanisms and the proposed framework supports mechanism-based studies of volunteered geographic information. • GeoVisX fingerprints explain OSM update mechanisms across 88 German cities. • Updates emerge from nonlinear, context-dependent interactions of urban features. • City-level clustering reveals distinct and heterogeneous OSM updating mechanisms.