建成環境對捷運轉乘運具選擇的影響：臺北捷運南港線之實證研究

BUILT ENVIRONMENT IMPACTS ON THE MODE CHOICE OF METRO ACCESS: AN EMPIRICAL STUDY OF THE NANGANG METRO LINE IN TAIPEI

本研究目的在了解旅次起迄點周圍建成環境對捷運轉乘運具選擇之影響，以臺北捷運南港線4個車站的進出乘客為研究對象，首先透過文獻回顧與個案訪談提出先驗關係，繼而經由問卷調查獲得樣本資料，接著運用個體選擇模式進行實證分析，最後檢討跟先驗關係不相符的實證結果的可能原因與意義。進站旅次的實證結果發現，人口密度、路口密度與銀行數量對選擇非機動運具為正影響；及業密度、運具多樣性指標與捷運站距離對選擇非機動運具為負影響；及業密度、街廓規模、自行車道長度與道路長度對選擇公車為正影響，而人口密度、建物密度與停車空間對選擇公車為負影響。出站旅次的實證結果發現，人口密度、建物密度、公車站距離與商店數量對選擇非機動運具為正影響，而及業密度、道路車輛密度、土地混合使用比例熵值、運具多樣性指標與道路長度對選擇非機動運具為負影響；及業密度與道路密度對選擇公車分別為正向與負向的影響。概似比檢定結果顯示，建成環境變數在解釋捷運轉乘運具選擇上，具有顯著的重要性。

This study aimed at empirically analyzing the impacts of built environments around trip ends on the mode choice of metro access by using the passengers of four stations along the Nangang Metro Line in Taipei as study sample. The study began with literature reviews and interviews to develop hypothetical relation- ships. To verify the proposed hypothetical relationships, questionnaire survey and discrete choice models were employed to obtain and analyze sample data. Finally, the possible reasons and meanings of the inconsistency between hypothetical relationships and empirical results were carefully examined. The empirical results of entering station trips show that population density, intersection density and number of banks increase non-motorized mode uses; employee density, transport mode diversity and distance to metro station decrease non-motorized transport mode uses; employee density, block size, bikeway length and road length increase bus uses; and, population density, building density and parking space decrease bus uses. For leaving station trips, population density, building density, distance to bus stops and number of shops increase non-motorized transport mode uses; employee density, vehicle density, mixed land uses, transport mode diversity and road length decrease non- motorized transport mode uses; employee density increases bus uses; and, vehicles density decreases bus uses. Based on likelihood ratio tests, built environment variables significantly contributed the goodness-of-fit of mode choice models for metro access.