Spectral Analysis of Borehole Skin and Wellbore Storage Effects

Spectral analysis of water level time series from observation wells can provide unique opportunities to characterize a variety of hydrogeological processes, including identification of system inputs, outputs, and aquifer parameters that control hydraulic responses. This classic technique has provided novel methods for interpreting available information in terms of dynamic systems behavior. This paper explores the transient interconnection between atmospheric pressure and/or solid earth tidal loadings on water level responses in a well with borehole skin effects. A new analytical solution is derived that describes periodic groundwater fluctuations in a well completed in an idealized, two-layer confined aquifer system. The first layer is located immediately adjacent to the wellbore and represents a skin effect of finite thickness. This layer is generally associated with either a filter pack (i.e., a negative skin effect) or to formation damage resulting from mechanical alterations and/or invasion of drilling fluids (i.e., a positive skin effect). The second layer is located radially outside the first layer and represents an undisturbed, idealized, homogeneous aquifer of uniform thickness as characterized in Ritzi et al. (1991, WRR, 27(5), 873-893). The solution presented here also applies to both small and large diameter wells because wellbore storage effects are included. This solution is presented in terms of non-traditional type-curves that are related to the spectral (SRF) and phase (PRF) response functions over a range of dimensionless aquifer transmissivity values. Results from these analyses can be used to characterize the effects of skin type, thickness, and hydraulic properties compared to those in the undisturbed aquifer. Here we focus on positive skin effects since they are of obvious concern in monitoring well applications. These results suggest that positive skin effects may significantly distort both the SRF and PRF over selected ranges of dimensionless transmissivity. However, both of these functions are relatively insensitive to variations in storage coefficient. This approach is illustrated by several real-world examples utilizing hourly water level records from typical monitoring wells.