Space-Terrestrial Integrated Networks (STIN) have an important influence on Service Function Chains (SFCs), broadening their service scope and enhancing application performance. However, STIN features a dynamic terrestrial access layer and an inter-satellite layer; it's complex to orchestrate the SFC in such a dynamic network. In this paper, we provide the Virtual Multi-Resource Storage Time Aggregated Graph (VMR-STAG) model and SFC orchestration algorithms for SFC orchestration in STIN. Inspired by the Virtual Topology (VT) method, VMR-STAG aggregates the dual-layer dynamic topology and time-varying resources into a single virtual graph, thus reducing the complexity of the STIN model. Based on VMR-STAG, we propose the Offline Request Orchestration (ORO) and Online Request Orchestration (OLRO) algorithms, designed to minimize SFC migration frequency, service delay, service jitter, and enhance load balancing. Leveraging resource distribution across multiple time slots, these algorithms schedule the long-lasting, high-performance SFC within STIN. Evaluation and simulation results demonstrate that our proposed model and algorithms significantly outperform conventional solutions, achieving significant improvements in model complexity, SFC migration frequency, workload balancing, service delay, and jitter.