An interesting accidental case study is the Thousand Springs area in south-central Idaho.

There were some naturally occurring springs there before farming and irrigation got going. The springs were substantially enhanced when a big network of surface canals were built in the late 19th century, and then substantially depleted as there was a transition to using well water and more efficient irrigation (edited to add: using sprinklers rather than flood irrigation). As butiwouldrathernot mentioned, the discharge and recharge happens on a timescale of decades.

When I lived in eastern Idaho a decade or so ago, there was a lively debate about whether we should deliberately run water through the canals all winter when the farmers were not using it, to recharge the aquifer and keep downstream wells from running dry.