IMPACTS OF NON-IDEAL FIELD CONDITIONS ON VERTICAL WATER VELOCITY ESTIMATES FROM STREAMBED TEMPERATURE RECORDS

Fluxes between streams and groundwater can be quantified directly from temperature time series data in streams and their streambeds using analytical solutions to the one-dimensional heat transport equation. The solutions rely on three underlying assumptions: (1) purely vertical flow, (2) no thermal gradient with depth in the streambed and (3) sinusoidal temperature signals. Here, VS2DH, a USGS numerical water and heat transport model, was used to generate synthetic temperature-time series data under conditions in the streambed that violate the aforementioned assumptions. The synthetic temperature records were used to evaluate the impact of violations of model assumptions on vertical water velocity estimates from the analytical model. The analytical model is very robust to violations of model assumptions, particularly when flow is predominantly vertical, and therefore may be a more accurate and more consistent method of estimating streambed flux, as compared to Darcy flux calculations. When using the amplitude ratio (A_r) to compute vertical water velocity, thermal gradients in the streambed and non-sinusoidal temperature signals generated percent errors of velocity estimates that were less than 10%. The percent error of velocity estimates from A_r was 30% or less when the vertical velocity was at least half the horizontal velocity. Analytical methods using A_r to derive velocity are less prone to error than methods that use lag time under non-ideal field conditions. The greatest source of error is non-vertical flow in the streambed. Errors from analytical solutions for velocity are comparable to or smaller than errors inherent to Darcy-based flux estimates and therefore the use of temperature data to quantify flux across the streambed is a promising alternative to more commonly used approaches, such as Darcy flux calculations.