Models based on StorAge Selection (SAS) functions are useful tools for modelling and understanding the temporal dynamic of storage/release of water and solutes at a catchment scale. SAS functions actually describe the sampling of different ages in catchment storage that produces river discharge. This sampling generally varies over time, for instance according to soil moisture, and so one possible strategy for a modeller is to infer different shapes of a given SAS function over time. In this work, we use the model states of a conceptual rainfall-runoff model (GR) to vary the parameters of the SAS function. This modelling strategy is applied for two OZCAR observatories: the French ORACLE observatory (Orgeval catchment, 104 km², in the East of Paris) and the French AgrHys observatory (Coët-Dan catchment, 5 km², in Brittany region). Chloride concentrations are used as a conservative tracer for both catchments. Because the GR model has many different states and internal fluxes, different modelling strategies are investigated to understand which one is the best proxy for modelling transit time dynamics. Previous studies over the Orgeval catchment, which is influenced by deep and large aquifers, highlighted that the simulated intercatchment groundwater flow (IGF) is the most informative for understanding the water age of the river. We further investigate this assumption in the context of more shallow groundwater, as in the Coët-Dan catchment. This study thus opens perspectives to better define how to adapt the general SAS function framework to different hydrological contexts.