Temporal evolution of age data under transient pumping conditions

While most age data derived from tracers have been analyzed in steady-state flow conditions, we determine their temporal evolution under transient pumping conditions. Starting pumping in a well modifies the natural flow patterns induced by the topographical gradient to a mainly convergent flow to the well. Our study is based on a set of models made up of a shallowly dipping aquifer overlain by a less permeable aquitard. These settings are characteristic of the crystalline aquifer of Plœmeur (Brittany, France) located in a highly fractured zone at the contact between a granite and micaschists. Under a pseudo steady-state flow assumption (instantaneous shift between two steady-state flow fields), we solve the transport equation with a backward particle-tracking method and determine the temporal evolution of the concentrations at the pumping well of the four atmospheric tracers CFC 11, CFC 12, CFC 113 and SF6. We show that apparent ages deduced from these concentrations evolve both because of the flow patterns modifications and because of the non-linear evolution of the atmospheric tracer concentrations. Flow patterns modifications only intervene just after the start of pumping, when the initially piston-like residence time distribution is transformed to a broader distribution mixing residence times from a wide variety of flow lines. Later, while flow patterns and the supplying volume of the pumping well still evolve, the residence time distributions are hardly modified and apparent ages are solely altered by the non-linear atmospheric tracer concentrations that progressively modifies the weighting of the residence time distribution. These results are confirmed by the observations at the site of Plœmeur in the pumping area. First, long term chloride observations confirm the quick evolution of the flow patterns after the start of pumping. Second, posterior and more recent evolutions of apparent ages derived from CFCs are consistent with the modeling results revealing in turn the marginal effect of the 20-year pumping on the first 70 years of the residence time distribution. We conclude that the temporal evolution of apparent ages should be used with great care for identifying the temporal evolution of the flow patterns as the apparent age evolution can have two sources - the transient flow patterns and transient tracer atmospheric concentrations. We argue that both evolutions either controlled by transient flow patterns or by transient tracer atmospheric concentrations provide key information that can be further used for the characterization of the hydrogeological system. This study illustrates that the temporal evolution of apparent ages could be used for models segregation and slightly compensate for the small number of tracers.