Green roofs are valuable solutions for rain management improvements in urban areas as they can partly store and delay rainfall water. Here, we point out that green roofs cannot be considered as static systems, which performances remain constant over time. This work is based on the cross-use of a lab-scale experiment and a modelling approach to evaluate the hydraulic performances of a green roof substrate over time. Experiments were conducted on new and aged materials, after 30 months of in situ aging. The experimental device was designed and implemented to simulate irrigation or rainfall and accurately monitor hydraulic fluxes. An estimation of hydraulic parameters was obtained by using inverse modelling of experimental data with HYDRUS-1D software. Comparisons between measured and modelled data demonstrated the reliability of the model for simulating the hydraulic behavior of the green roofs. Considering an incoming water event – which mimics a heavy rainfall – of 43 mm h−1 and similar water content initial conditions, our simulations indicate that the retention capacity and the delay effect were always higher for the new substrate than for the aged one. Both of these performance indicators strongly vary with the initial water content of the substrates. Whereas the relation is linear for the retention capacity ranging from 100% of retention for the drier conditions to 0% for the saturated substrates, it is more complex for the delay effect. Such performances were comparable to analogous data from the existing literature. Furthermore, this comparison confirmed that green roofs are submitted to an early aging in terms of structure, i.e. porosity. In our study, the aged substrate presented less favorable performances thus highlighting the key role of the composition of green roof substrate not only on the initial performances but also on their sustainability.