Springs sourced by shallow aquifers often exhibit a temperature signal that mimics the seasonal harmonic air temperature signal, albeit significantly damped and lagged. If conduction does not dominate, it seems possible that numerical models could be employed to gain an understanding of shallow subsurface flow regimes from the degree of damping or lagging in the spring temperature. I suspect that it could be an unconstrained problem unless either the aquifer dimensions or thermal properties were known.
This would be somewhat analogous to how the damping of the stable isotope ratio seasonality is used as a tracer in catchment hydrology. In the present case, the groundwater discharge temperature would have to be recorded prior to mixing with the surface water.