<p>Loess slopes are highly susceptible to landslides during rainfall events. Therefore, analyzing water migration on loess slopes during rainfall is crucial for clarifying the infiltration mechanism. In this study, the spatiotemporal effects of rainfall and water infiltration on a loess slope under different rainfall conditions are investigated. Based on the findings, an optimized infiltration model is proposed and validated by examining the relationship between the infiltration depth and rate, the migration of the wetting front, and rainfall duration under varying rainfall intensities. The results reveal three distinct phases of the volumetric water content during rainfall events: a stable phase before the sensor responds, a slow increase phase, and a stable phase. Rainfall intensity and slope height significantly influence the volumetric water content, with the rainfall infiltration rate rising as these parameter values increase. Furthermore, the optimized infiltration model, which incorporates stratification, slope gradient, and slope reinforcements, demonstrates the closest alignment with the on—site measurements. In comparison, the model that only considers stratification and slope gradient follows in terms of accuracy. However, the traditional model exhibits the largest error, which tends to increase with the infiltration depth and rainfall intensity. Consequently, the model considering the influence of the transition layer in the unsaturated zone, exhibits a larger infiltration rate during the free infiltration stage and a smaller rate during the ponding infiltration stage than the traditional model. This research provides theoretical and technical support for stability assessment and early warning systems of loess slopes.</p>