Using Wavelet Coherence to Characterize Surface Water Infiltration into a Low‐Lying Karst Aquifer

Abstract

Karstified carbonate aquifers may receive significant recharge contributions from losing streams, hence, the knowledge about surface water-groundwater (SW-GW) interactions is crucial with regard to water management (e.g., source protection zone delineation). The dynamics of SW-GW interactions may depend on factors such as the relative water levels between streams and aquifers, resulting in a temporal variation of exchange, which imposes complexity to the understanding of such dynamics. This study highlights the use of high-resolution time series and multiresolution analysis to help to gain insights into such complex dynamics. Wavelet coherence is applied on hourly time series of rainfall, stream, and spring discharges of a low-lying karstified spring catchment to yield a correlation in the time-frequency domain. This analysis provides comprehensive information on the overall impact of the river on the spring, which is supported by the cross-correlation function, as well as by more detailed information, including time-variant influences such as a threshold level of influence. Field observations of turbidity sampling at the spring appear to support this interpretation. This innovative approach relies on basic hydrological parameters, water level, or discharge, and is therefore applicable to many other systems with such existing time series. Article impact statement: Wavelet coherence yields highresolution information on surface water-groundwater interaction.

Flow Cytometry and Fecal Indicator Bacteria Analyses for Fingerprinting Microbial Pollution in Karst Aquifer Systems

“…Spring I revealed more consistent TCC results across the year with some slight elevated values during periods of zero effective rainfall in the summer. This small catchment (3.4 km 2 ) has no visible karst features of interest at the surface, and recent hydrological studies of the catchment have shown that under moderate recharge conditions the aquifer has a damped response to rainfall, but during higher sustained recharge conditions the spring becomes directly connected to a nearby river via a high level conduit, providing a much faster response to rainfall events (Schuler, 2020 ; Schuler et al., 2021 ).…”

Inland Tidal Oscillations Within the Yucatan Peninsula

“…This work has also exhibited the advantages of wavelet analysis over other types of signal analysis, particularly for temporally varying events. Recently, other studies such as Schuler et al (2021); Rodríguez-Murillo and Filella (2020); Lee and Kim (2019) have explored wavelet analysis and wavelet coherence as a tool for characterizing hydrological measurements (see Supporting Information). Future work could also include extending sensor coverage across the peninsula to create a more granular map of lag times.…”

An evaluation of semidistributed-pipe-network and distributed-finite-difference models to simulate karst systems

“…Due to the time-invariant limitations of some transfer function techniques when trying to reconstruct the temporal structure of relatively infrequent flow events such as floods or low water periods (Kuczera and Mroczkowski 1998;Labat et al 2000a), different numerical methods have been used to develop nonparametric transfer functions in order to capture nonlinear and nonstationary dynamics more accurately. These have included approaches more rooted in the spectral (frequency) domain (Larocque et al 1998;Labat et al 2000aLabat et al , 2002Bailly-Comte et al 2008) as well as time/frequency domains using wavelets (Labat et al 2000b(Labat et al , 2001Schuler et al 2020b). Machine learning approaches, for example artificial neural networks, have also been applied to karstic aquifers (Beaudeau et al 2001;Kurtulus and Razack 2007;Hu et al 2008).…”

“…Given the limitations highlighted in the previous models about the uncertainty in the diffuse recharge process, two catchments with long-term, high-resolution spring discharge time series were chosen, namely Ballindine and Manorhamilton (see section 'Materials and methods') in order to characterise the different flow components better by use of numerical techniques in both time and frequency domains on output signals (Schuler et al 2020b). The quantification of the relative contribution of these components, as well as their numerical representation, was attempted by analysing the three recharge components in the time and frequency domain.…”