Differentiation of Multi-Parametric Groups of Groundwater Bodies through Discriminant Analysis and Machine Learning

Abstract

In order to facilitate the monitoring of groundwater quality in France, the groundwater bodies (GWB) in the Provence-Alpes-Côte d’Azur region have been grouped into 11 homogeneous clusters on the basis of their physico-chemical and bacteriological characteristics. This study aims to test the legitimacy of this grouping by predicting whether water samples belong to a given sampling point, GWB or group of GWBs. To this end, 8673 observations and 18 parameters were extracted from the Size-Eaux database, and this dataset was processed using discriminant analysis and various machine learning algorithms. The results indicate an accuracy of 67% using linear discriminant analysis and 69 to 83% using ML algorithms, while quadratic discriminant analysis underperforms in comparison, yielding a less accurate prediction of 59%. The importance of each parameter in the prediction was assessed using an approach combining recursive feature elimination (RFE) techniques and random forest feature importance (RFFI). Major ions show high spatial range and play the main role in discrimination, while trace elements and bacteriological parameters of high local and/or temporal variability only play a minor role. The disparity of the results according to the characteristics of the GWB groups (geography, altitude, lithology, etc.) is discussed. Validating the grouping of GWBs will enable monitoring and surveillance strategies to be redirected on the basis of fewer, homogeneous hydrogeological units, in order to optimize sustainable management of the resource by the health agencies.

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

“…The consequences, i.e., local faecal contamination that varies over time [35,36], are very frequently observed for surface water [37] but also by the health agencies responsible for monitoring water quality and correspond to the main reasons for non-compliance reported [34,38,39]. Moreover, a similar mechanism has already been observed in other regions based on extracts from the Sise-Eaux database [17][18][19][20][22][23][24], particularly in Mediterranean climates where late summer storms can be violent and favour run-off. It should be noted, however, that faecal contamination is carried by at least the first three factorial axes, which are orthogonal to each other and therefore reflect independent mechanisms.…”

“…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 ). The grouping method works on regions of variable size, i.e., medium-sized and fairly contrasted (Provence-Alpes-Côte d'Azur) or larger but moderately contrasted (Occitanie) from a lithological, altitudinal, environmental, or land-use point of view.…”

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 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]. A study conducted in the vast Auvergne-Rhône-Alpes region established a typology of quality parameters based on structures or associations between these parameters, differing in terms of spatial extent, seasonality, or long-term behavior [14].…”

“…A study conducted in the vast Auvergne-Rhône-Alpes region established a typology of quality parameters based on structures or associations between these parameters, differing in terms of spatial extent, seasonality, or long-term behavior [14]. Finally, quantifying the information initially contained in the datasets and lost during the grouping into homogeneous water bodies validated the proposed analysis method [13,15] on scales ranging from small to large regions (8000 to 80,000 km 2 ). This method works across regions of variable size, ranging from moderately contrasting (Provence-Alpes-Côte d'Azur) to larger but moderately contrasting in lithological, altitudinal, environmental, or land use aspects.…”