Characterizing water level responses to barometric pressure fluctuations from seconds to days

Abstract

A regression deconvolution method using distributed lags for the efficient calculation of barometric/loading response functions in the presence of Earth tides and background trends was applied to water pressure measurements collected every second. This method was able to simultaneously deconvolve multiple stresses that can be used for both signal decomposition (removal of noise and isolation of signals) and the determination of impulse response functions that can be interpreted for hydraulic properties and subsurface characterization. We also demonstrate how the static barometric efficiency is a deficient parameter when applied to wells having delayed responses and needs to be applied with care. The distributed lag method combined with high frequency monitoring has implications for improved site characterization, determining the continuity of subsurface features, parameter estimation, understanding the degree of confinement, and signal decomposition. The methods for this paper can be found in the R hydrorecipes package.