QUANTIFYING GROUNDWATER DISCHARGE USING STATISTICAL ANALYSIS OF TEMPERATURE TIME-SERIES AND RESISTIVITY METHODS

Although surface-water and groundwater are often referred to as separate domains, they are intimately related. Declines in surface water flows during dry periods often result in contributions from groundwater. Even small-scale seeps can act as pathways for surface water-groundwater (SW-GW) interaction and thus act as contaminant transport pathways in the hydrologic cycle. Therefore a comprehensive understanding of the SW-GW relationship is essential for improved management practices. The objective of this study is to develop a new statistical method to better identify and quantify the SW-GW fluxes in an area characterized by low hydraulic gradients and conductivity sediments with possibly diffusive and small-scale flow paths. These characteristics provide a challenge when attempting to quantify SW-GW fluxes utilizing traditional methods. Traditional methods used to quantify hydraulic gradients such as piezometers are hampered by these low gradients and permeabilities making it difficult to quantify discharge rates. To overcome these challenges, annual time-series thermal records were collected in Oso Creek at four different depths acting as proxies for conventional methods of estimating small scale SW-GW fluxes. These data were analysed using Seasonal Trend Decomposition by Loess (STL) to identify small-scale exchanges. Through STL analysis the annual time-series data were decomposed into daily and annual cyclic components and a random or non-cyclic component. By examining the random component thermal records for the various depths and periods of SW-GW interaction were intuitively discernable. Cross correlation of the random thermal components provided an average groundwater discharge flux rate of 0.21 m*d­^-1 for all levels. VFLUX was then used to validate flux rates using the daily cyclic components. VFLUX results were inconclusive and provided a range of fluxes from 0.002 to 6.9 m*d­^-1. These inconsistencies were the result of sensor spacing and thermal properties of the matrix. Further analysis will incorporate calculating flux rates from amplitude and phase shift methods using the random components which is in stark contrast to the VFLUX method. The flux rates will further be compared to time-lapse resistivity-derived fluxes.