资源环境科技发展态势分析平台

India's land transport GHG emissions are small in international comparison, but growing exponentially. Understanding of geographically-specific determinants of GHG emissions is crucial to devise low-carbon sustainable development strategies. However, previous studies on transport patterns have been limited to socio-economic context in linear and stationary settings, and with limited spatial scope. Here, we use a machine learning tool to develop a nested typology that categorizes all 640 Indian districts according to the econometrically identified drivers of their commuting emissions. Results reveal that per capita commuting emissions significantly vary over space, after controlling for socioeconomic characteristics, and are strongly influenced by built environment (e.g. urbanization, and road density), and mobility-related variables (e.g. travel distance and travel modes). The commuting emissions of districts are characterized by unique, place-specific combinations of drivers. We find that income and urbanization are dominant classifiers of commuting emissions, while we explain more fine-grained patterns with mode choice and travel distance. Surprisingly the most urbanized areas with highest population density are also associated with the highest transport GHG emissions, a result that is explained by high car ownership. This result contrasts with insights from OECD countries, where commuting emissions are associated with low-density urban sprawl. Our findings demonstrate that low-carbon commuting in India is best advanced with spatially differentiated strategies.