Assessing the influence of the low frequency variability on the groundwater trends across metropolitan France

The groundwater levels in metropolitan France display various dynamics depending on physical and morphometric properties of aquifers. Due to the capability of catchments and aquifers to act as a low-pass filter, the low-frequency variability, originating from the large-scale climate variability, is often well-imprinted in groundwater levels. Consequently, the low-frequency variability can widely influence the trend assessment of groundwater levels, and leading to misleading interpretations of trends.

Therefore, we aimed to evaluate the influence of the low-frequency variability on groundwater trends as part of the TACTIC project. To conduct this analysis, the multiannual (~7-yr) and decadal (~17-yr) variabilities were filtered out groundwater levels to assess their respective influence on trends via the maximum overlap discrete wavelet transform. The ~7-yr and ~17-yr variabilities were targeted because they are the predominant low-frequency variabilities in groundwater levels across metropolitan France. Then, the Sen’s slope was estimated on monthly averages (the original signal), the ~7-yr filtered time series, and the ~17-yr filtered time series to be compared.

The figure 1 provides an example of the analysis for the borehole of Blacqueville monitoring the Seno-Turonian chalk of Normandy over the 1976-2019 period.

Figure 1. Assessment of Sen’s slope (green) on the monthly averages of groundwater levels of Blacqueville and on the filtered time series of either the ~7-yr or ~17-yr variability.

For the example of Blacqueville borehole, we can observe a decreasing trend of monthly averages of groundwater levels. When the ~7-yr variability is filtered out groundwater levels, the trend is accentuated. Consequently, it means that the ~7-yr variability attenuates the downward trend. Conversely, when the ~17-yr variability is filtered out groundwater levels, the trend is attenuated. Consequently, it means that the ~17-yr variability accentuates the downward trend.

In this case, the downward trend of groundwater levels at Blacqueville seems to be primarily the result of the weakening amplitude of the low-frequency variability since the end of the 2000’s (Fig.1; Monthly averages). The amplitude modifications of low-frequency variabilities in groundwater levels over time (which are also noticeable in precipitation signals) are induced by the large-scale atmospheric and oceanic circulation and these modifications are purely stochastic. Due to the low-pass filter effect of aquifers, these amplitude differences are exacerbated in groundwater levels compared to precipitation signals. Therefore the downward trend in groundwater levels of Blacqueville must not be interpreted as a trend induced by the climate change. This trend is primarily carried by the ~17-yr variability displaying a weakening amplitude since the end of the 2000’s that is visible on the middle plot of the figure 1.

We generalized the analysis to 215 non-influenced boreholes by pumping monitoring sedimentary, bedrock, volcanic and alluvial aquifers in metropolitan France. Results of this analysis were submitted to Journal of Hydrology.