Water temperature at springs generally provides useful information concerning aquifer geometry and recharge. Temperature monitoring at 25 springs and cave streams in southeastern Minnesota has shown four distinct thermal patterns that can be interpreted in terms of heat exchange effectiveness along a flow path and the nature of recharge. The patterns provide information about the size of the flow path, recharge type and duration, and aquifer depth.
Water temperature is generally an interactive tracer, where heat exchange rapidly occurs when water and aquifer rock are at different temperatures. In a multi-tracer experiment at Freiheit Spring, MN, uranine, chloride, and δD breakthrough curves were essentially identical and conservative. In contrast, the water temperature interacted with the aquifer as it moved along the flow path, producing a damped, lagged thermal signal at the spring. However, both the conservative and nonconservative tracers provide useful geometrical information. By summing discharge between the initial increase in stage produced by a pressure pulse and the chloride peak, the conduit volume is estimated as 51 m<super>3</super>. Using a heat transport simulation to reproduce the modified thermal signal requires a planar flow path with a hydraulic diameter of 7 cm. Both methods together suggest a bedding plane flow path that is 3.5 cm high by 10 m wide, in agreement with the observed spring geometry. The different tracers provide complementary information and stronger constraints on flow path geometry than could be obtained using a single tracer.
Finally, numerical simulations were run to determine variables controlling thermal retardation in karst conduits. The lag of a thermal peak in the water is proportional to a conduit's length; is proportional to the square root of recharge duration, rock thermal conductivity, rock specific heat, and rock density; and is inversely proportional to a conduit's hydraulic diameter, velocity, water specific heat, and water density. These individual relationships were then combined to form one collective function, which, when plotted against thermal peak lag produced a line in log-log space. The relationship between the thermal peak lag and the combined function potentially enables estimates of conduit geometry using thermal peak lag data.