[Purposes] Under the influence of complex geological conditions, extreme rainfall climate, and engineering activities, landslides on highway slopes have occurred frequently. The landslides not only caused critical infrastructure damage such as subgrade settlement and pavement collapse, but also often lead to ecological issues like soil erosion and vegetation destruction. It is crucial to understand the characteristics and processes of rainfall-induced slope instability for comprehending slope failure mechanisms and developing rational stability analysis methods. [Methods] Based on the fiber Bragg grating (FBG) humidity sensors, time series data of water content during rainfall infiltration was obtained. The internal seepage field of the subgrade slope during rainfall infiltration was reconstructed. A flow-solid coupling finite element model of the subgrade slope was established by using the finite difference method. The physical and mechanical responses of the slope during rainfall infiltration were revealed by combining with the results of model test results. [Findings] The monitoring results indicated that the front continuously advances towards the deeper part with the duration of rainfall, and the maximum depth of the wetting front reached 45 cm, while the maximum failure depth extended to 30 cm in the subgrade slope model test. While a continuous high water content zone forms at the toe of the slope, with the maximum water content approximately 46%, and the unsaturated permeability coefficient increases periodically. The numerical analysis results shows that rainfall infiltration leads to the attenuation of matrix suction and the rise of pore water pressure, with the absolute value of negative pore pressure being approximately 17.9 kPa. The safety factor decreased from approximately 1.612 to approximately 1.027, the shear strain increment zone gradually penetrated and extended towards the toe of the slope, and the slope approaches the critical instability state. [Conclusions] Fiber Bragg grating (FBG) humidity sensors can effectively characterize the spatiotemporal evolution of water content under rainfall infiltration. It complements the existing dynamic monitoring of seepage fields of slope, and enables long-term real time monitoring of the stability of the rainfall subgrade slope.