AUTOMATED CALCULATION OF VERTICAL PORE-WATER FLUX FROM REAL-WORLD TEMPERATURE TIME SERIES USING THE VFLUX METHOD AND COMPUTER PROGRAM

Heat is a useful tracer for quantifying vertical water flux to or from surface water bodies. Recent developments in instrumentation and analytical modeling have improved the process of estimating groundwater-surface water exchange using temperature time series; however, analyzing large amounts of real-world thermal data still confronts researchers with many practical challenges. We present a new method for processing raw temperature time series and calculating vertical water flux in shallow subsurface-water systems. The step-by-step process, named VFLUX, synthesizes several recent advancements in signal processing into a novel workflow, and adds new techniques for calculating flux rates with large numbers of temperature records from high-resolution sensor profiles. The method has been incorporated into a flexible and robust computer program written in the MATLAB language, which automates the entire process of calculating vertical flux rates from natural temperature time series collected in the field. The program synchronizes and resamples temperature data from multiple sensors in a vertical profile, isolates the diurnal signal from each time series and extracts its amplitude and phase angle information using Dynamic Harmonic Regression (DHR), and calculates vertical water flux rates between many sensor pair combinations using heat transport models. Flux rates are calculated every one-to-two hours using four previously-developed analytical methods. One or more “sliding analysis windows” can be used to automatically identify any number of variably spaced sensor pairs for flux calculations, which can be especially useful when a single vertical profile contains many sensors, such as in a high-resolution fiber-optic distributed temperature sensing (DTS) profile. We demonstrate the utility of the VFLUX program by processing two real-world temperature time series datasets collected using discrete iButton Thermochron temperature sensors and a continuous high-resolution DTS profile. These examples illustrate the challenges associated with modeling real-world time series, and how they can be efficiently overcome using the VFLUX method and program.