Publication Date:
2024-03-06
Description:
The groundwater pressure response to the ubiquitous Earth and atmospheric tides provides a largely untapped opportunity to passively characterize and quantify subsurface hydro-geomechanical properties. However, this requires reliable extraction of closely spaced harmonic components with relatively subtle amplitudes but well-known tidal periods from noisy measurements. The minimum requirements for the suitability of existing groundwater records for analysis are unknown. This work systematically tests and compares the ability of two common signal processing methods, the discrete Fourier transform (DFT) and harmonic least squares (HALS), to extract harmonic component properties. First, realistic conditions are simulated by analyzing a large number of synthetic data sets with variable sampling frequencies, record durations, sensor resolutions, noise levels and data gaps. Second, a model of two real-world data sets with different characteristics is validated. The results reveal that HALS outperforms the DFT in all aspects, including the ability to handle data gaps. While there is a clear trade-off between sampling frequency and record duration, sampling rates should not be less than six samples per day and records should not be shorter than 20 days when simultaneously extracting tidal constituents. The accuracy of detection is degraded by increasing noise levels and decreasing sensor resolution. However, a resolution of the same magnitude as the expected component amplitude is sufficient in the absence of excessive noise. The results provide a practical framework to determine the suitability of existing groundwater level records and can optimize future groundwater monitoring strategies to improve passive characterization using tidal signatures.
Description:
H2020 Marie Sklodowska-Curie Actions
http://dx.doi.org/10.13039/100010665
Description:
Swedish Research Council
Keywords:
ddc:551.49
;
Tidal subsurface analysis
;
Tidal constituents
;
Signal analysis
;
Harmonic least squares
;
Non-uniform sampling
Language:
English
Type:
doc-type:article
Permalink
