With the rapid development of the low-altitude economy, particularly represented by urban air mobility, low-altitude airspace is expected to present characteristics such as high frequency, high density, heterogeneity, and dynamic operations. Traditional operation safety assessment systems, which are predominantly rule-driven and offline verification-based, are no longer sustainable in terms of reliability and real-time performance. There is an urgent need to reshape the entire system through a new methodology centered on digital twin technologies. To this end, this paper systematically elaborates on a cloud-edge-terminal collaborative architecture of the digital twin of low-altitude transportation systems. Within this architecture, the cloud layer serves as the twin hub, constructing a global digital environment; the edge layer covers key nodes, completing the aggregation and fusion of multi-source sensing data; and the device layer, primarily composed of aircraft, acts as the sensed objects while also undertaking responsibilities such as data reporting and command execution. Focusing on this architecture, this paper sorts out collaborative and non-collaborative perception methods within advanced low-altitude sensing systems, which form the data foundation supporting digital twin modeling, mapping, and scenario driving. For high-fidelity mapping of complex elements in low-altitude transportation, two core objects, the digital twin environment and the digital twin aircraft, were defined, and twin simulation platforms and tools for system testing and algorithm validation were summarized. It also sorts out the standardized operations safety assessment system based on digital twin and reviews quantitative operations safety assessment technologies, providing a scientific basis for the design, certification, and policy-making of low-altitude transportation systems.