This paper proposes a spatial-temporal topology from a macroscopic view to analyze the performance of airport taxi network operations. Through a macroscopic modelling of arrival and departure aircraft taxi processes in the airport taxi network, we establish a system of taxi network performance indicators (TNPIs) consisting 5 categories and 26 indicators, which includes the surface instantaneous flow indicators (SIFIs), surface cumulative flow indicators (SCFIs), aircraft queue length indicators (AQLIs), slot resource demand indicators (SRDIs) and aircraft taxi time indicators (ATTIs). Then, we analyze the correlation among different TNPIs. By identifying the key factors affecting aircraft taxi time such as takeoff and landing queue length, we provide models for predicting aircraft taxi time based on multiple regression analysis. The real-world case study in Shanghai Pudong airport demonstrates significant correlations among some of the proposed TNPIs, and the results also show the significantly improved accuracy of the proposed prediction models over some conventional models, which brings significant benefits to analyze the performance of airport taxi network and support decision making in airport operations.