Indicators for station classification.

Mountainous urban rail transit stations exhibit distinct characteristics. To investigate how these features affect passenger flow variations at rail stations, we analyze geographic-environmental data surrounding the stations and integrate road network topology, automatic fare collection data, and point-of-interest (POI) data. We propose a method to classify rail transit stations by considering the mountainous features and establish a multiscale geographically weighted regression (MGWR) model to assess the classification results. This study focuses on 189 rail stations in Chongqing, identifying six station categories: comprehensive mountainous, comprehensive non-mountainous, employment mountainous, employment non-mountainous, residential mountainous, and residential non-mountainous. The MGWR results show that road growth coefficients, average longitudinal slopes, and road lengths significantly influence station performance. For instance, the average longitudinal slope substantially affects employment in mountainous stations, particularly during the morning peak. The analysis reveals that the average longitudinal slope exerts a stronger negative effect on morning peak inbound passenger flow at employment mountainous stations (-0.949), indicating that commuters are more sensitive to travel time during the morning peak. In contrast, the evening peak inbound passenger flow is less impacted (-0.409), suggesting that evening commuters face fewer time constraints. These findings offer strategic insights for zoning transit stations to support transit-oriented development(TOD).