Exploring Multiscale Variability in Groundwater Quality: A Comparative Analysis of Spatial and Temporal Patterns via Clustering

Abstract

Defining homogeneous units to optimize the monitoring and management of groundwater is a key challenge for organizations responsible for the protection of water for human consumption. However, the number of groundwater bodies (GWBs) is too large for targeted monitoring and recommendations. This study, carried out in the Provence-Alpes-Côte d’Azur region of France, is based on the intersection of two databases, one grouping together the physicochemical and bacteriological analyses of water and the other delimiting the boundaries of groundwater bodies. The extracted dataset contains 8627 measurements from 1143 observation points distributed over 63 GWB. Data conditioning through logarithmic transformation, dimensional reduction through principal component analysis, and hierarchical classification allows the grouping of GWBs into 11 homogeneous clusters. The fractions of unexplained variance (FUV) and ANOVA R2 were calculated to assess the performance of the method at each scale. For example, for the total dissolved load (TDS) parameter, the temporal variance was quantified at 0.36 and the clustering causes a loss of information with an R2 going from 0.63 to 0.4 from the scale of the sampling point to that of the GWB cluster. The results show that the logarithmic transformation reduces the effect of outliers and improves the quality of the GWB clustering. The groups of GWBs are homogeneous and clearly distinguishable from each other. The results can be used to define specific management and protection strategies for each group. The study also highlights the need to take into account the temporal variability of groundwater quality when implementing monitoring and management programs.

Groundwaters in the Auvergne-Rhône-Alpes Region, France: Grouping Homogeneous Groundwater Bodies for Optimized Monitoring and Protection

“…The data used in this study were extracted from the Sise-Eaux database (https: //data.eaufrance.fr/concept/sise-eaux, accessed in 15 March 2021). For further details regarding this national database, previous works by our research group [8][9][10][11][12][13][14][15], as well as the basic references [25,26], can be consulted. The extraction conducted for a 30-year period from July 1990 to September 2020 for our study area resulted in a sparse matrix of 114,033 observations (water samples) distributed across 3146 sampling points (Figure 3a), with 21 measured parameters.…”

“…The grouping of homogeneous groundwater bodies (i.e., waterbodies with similar compositions and similar mechanisms leading to this composition) after reducing the data hyperspace dimensionality significantly improved this issue. The presence of extreme values in the dataset exaggerated the impact of certain parameters, which was addressed by logarithmic data conditioning [10,11]. Discriminating spatial and temporal variance helped identify seasonal mechanisms or long-term trends [12,13].…”

“…Only the pH, which already corresponds to the logarithmic transformation of the chemical activity of H 3 O + , was retained without conditioning. The goal was to align the distributions of each parameter with a normal distribution, but more importantly, to limit the influence of extreme values that could mask certain processes responsible for the variability in water quality within the dataset [11,12]. 2.…”

Multi-Parameter Analysis of Groundwater Resources Quality in the Auvergne-Rhône-Alpes Region (France) Using a Large Database

“…Initial work carried out in the Provence-Alpes-Côte d'Azur region highlighted the fact that the heterogeneity of natural environments on a regional scale masks some of the major processes involved in acquiring quality [17]. The development of groups of homogeneous groundwater bodies after reducing the dimensionality of the data hyperspace has considerably improved this point [18], but the presence of extreme values in the dataset exaggerates the impact of certain parameters, which was resolved by logarithmic conditioning of the data [19,20]. Discriminating spatial and temporal variance has made it possible to highlight seasonal mechanisms or long-term trends [20][21][22].…”

“…The development of groups of homogeneous groundwater bodies after reducing the dimensionality of the data hyperspace has considerably improved this point [18], but the presence of extreme values in the dataset exaggerates the impact of certain parameters, which was resolved by logarithmic conditioning of the data [19,20]. Discriminating spatial and temporal variance has made it possible to highlight seasonal mechanisms or long-term trends [20][21][22]. Finally, quantification of the information initially contained in the datasets and lost when grouped into homogeneous bodies of water made it possible to validate the proposed analysis method [23,24] on the scale of small to large regions (8000 to 80,000 km 2 ).…”