Chemometrics of the Environment: Hydrochemical Characterization of Groundwater in Lioua Plain (North Africa) Using Time Series and Multivariate Statistical Analysis

Abstract

This study aims to analyze the chemical composition of Lioua’s groundwater in order to determine the geological processes influencing the composition and origin of its chemical elements. Therefore, chemometrics techniques, such as multivariate statistical analysis (MSA) and time series methods (TSM) are used. Indeed, MSA includes a component analysis (PCA) and a cluster analysis (CA), while autocorrelation analysis (AA), supplemented by a simple spectral density analysis (SDA), is used for the TMS. PCA displays three main factors explaining a total variance (TV) of 85.01 %. Factors 1, 2, and 3 are 68.72%, 11.96%, and 8.89 % of TV, respectively. In the CA, total dissolved solids (TDS) and electrical conductivity (EC) controlled three groups. The elements SO₄²⁻, K⁺, and Ca²⁺ are closely related to TDS, the elements Na⁺, Cl⁻, and Mg²⁺ are closely related to CE, while HCO³⁻ and NO³⁻ indicate the dissociation of other chemical elements. AA shows a linear interrelationship of EC, Mg²⁺, Na⁺, K⁺, Cl⁻, and SO₄²⁻. However, NO₃⁻ and HCO₃⁻ indicate uncorrelated characteristics with other parameters. For SDA, the correlograms of Mg²⁺, Na⁺, K⁺, Cl⁻, and SO₄²⁻ have a similar trend with EC. Nonetheless, pH, Ca²⁺, HCO₃⁻ and NO₃⁻ exhibit multiple peaks related to the presence of several distinct cyclic mechanisms. Using these techniques, the authors were able to draw the following conclusion: the geochemical processes impacting the chemical composition are (i) dissolution of evaporated mineral deposits, (ii) water–rock interaction, and (iii) evaporation process. In addition, the groundwater exhibits two bipolar characteristics, one recorded with negative and positive charges on pH and Ca⁺ and another recorded only with negative charges on HCO₃⁻ and NO₃⁻. On the other hand, SO₄²⁻, K⁺, Ca²⁺, and TDS are the major predominant elements in the groundwater’s chemical composition. Chloride presence mainly increases the electrical conductivity of water. The lithological factor is dominant in the overall mineralization of the Plio Quaternary surface aquifer waters. The origins of HCO₃⁻ and NO₃⁻ are as follows: HCO₃⁻ has a carbonate origin, whereas NO₃^– has an anthropogenic origin. The salinity was affected by Mg²⁺, SO₄²⁻, Cl⁻, Na⁺, K⁺, and EC. Ca²⁺, HCO₃⁻, and NO₃⁻ result from human activity such as the usage of fertilizers, the carbonate facies outcrops, and domestic sewage.

1. Introduction

Water is an essential compound of everyday life, it is so familiar that we often forget its role, importance, and necessity as a vital resource for nature and living beings [1,2]. In developing countries with arid climates, groundwater is crucial since it is often the only source of drinking water [3]. Studies on water management have prioritized water conservation due to an increased reliance on groundwater [4].

Due to chemical interactions with the geological layers through which it flows and, to a lesser extent, due to other factors such as air, surface water, and anthropogenic activity, groundwater contains a vast variety of dissolved solids in varying concentrations [5,6]. Groundwater constitutes a memory that brings to the surface indications of the deeper reservoir [7]. It is clear that the chemistry of groundwater depends, particularly, on the lithological composition of the layers that have been crossed and on the contact time with those layers. Consequently, the elements, which are in the solution, are informative regarding the nature of the aquifer they have crossed [8]. Nonetheless, water quality, which is a combination of chemical, physical and biological parameters, is a somewhat subjective term since its real value depends on the specifics of a particular use [9]. An increased understanding of the hydrogeochemical systems in the water table and the geochemical evolution of water in arid and semi-arid regions may be necessary for the efficient management of water resources and their sustainable exploitation. It is vital to identify the variables impacting water quality using appropriate evaluation techniques. Before beginning any associated applied research projects, a feasibility analysis for a certain region is essential.

Presently, researchers have been using linear models and multivariate statistical analysis for sustainable groundwater resource management [10,11]. Chemometric techniques and saturation indices have been employed to better characterize the hydro-geochemistry in the literature. [12,13,14]. Chemometrics has long been considered useful in obtaining information from environmental data that could be interpreted to uncover useful correlations [15,16,17,18]. Principal component analysis (PCA), cluster analysis (CA), and geographic information system (GIS) techniques help in the interpretation of the big dataset to provide a better understanding of the geochemical dynamics and causes of the pollution in the alluvial aquifers [17].

To study the chemical changes in groundwater, linear modeling should be mentioned. These include time series, multivariate analysis, and geostatistical techniques [5,19,20,21] as well as applied multivariate analysis using cluster analysis (CA) as well as principal component analysis (PCA) [22,23]. We used factor analysis (FA) to discuss mineralization, geochemical evolution and finally, groundwater contamination. In addition, according to Roubil et al. [24], CA has been used to interpret hydrochemical data. This technique has been widely utilized by researchers to examine the chemical development of water along groundwater flow [25]. This method enables the verification of spatial and temporal variations caused by anthropogenic and natural factors [26].

Concerning time series analysis, this method is considered one of the most useful techniques applied in modeling and forecasting water quality [27]. At present, time series analyses are used in various disciplines, such as economics, natural sciences, physics, and engineering. Water resources engineering also belongs to this category since many characteristics of streams, water bodies, and groundwater resources, as well as seas and lakes, are defined using time series [28]. Therefore, it can be useful in modeling and in developing an understanding of the process or a phenomenon, as well as in forecasting future values based on past observations and data [29,30]. Autocorrelation analysis determines the linear relationship between subsequent values over a time period and the autocorrelation function is stated as the variance and auto covariance of time series data [31]. At the same time, a simple spectral density analysis completes the autocorrelation analysis [32].

Regarding Algeria, while perennial agriculture of date palms or greenhouse crops requires much more water from irrigation, the development of irrigated crops in the south of the country (a region characterized by an arid climate and located beyond the isohyet of 150 mm/year) has been based primarily on groundwater pumping. This approach has increased farmer income, but has also caused water quality to deteriorate (salinization), especially with regard to the region’s groundwater resources. The soils of the Saharan regions of Algeria are rich in soluble salts, which accumulate and often present calcareous or gypseous or calcareous–gypseous crusts. The presence of these gypsum, limestone, and saline accumulations, in general, poses many problems for the physicochemical quality of groundwater via water–soil interaction. This is the case in the Lioua region.

The aim of this study is to examine the chemical composition of groundwater while determining the origin of the chemical elements found in the water using chemometric techniques, such as time series and multivariate statistical analysis methods. The techniques mentioned will be used for all the samples taken from a potential underground water resource, massively exploited, located in the arid region of Algeria’s Lower Saharan region. This is the surface aquifer of the Plio Quaternary aquifer, intended for drinking water supply and irrigation of the Lioua region.

This article also elucidates the use of the chemometric technique for the analysis and interpretation of water quality data for the rational management of groundwater resources in arid regions. The results of this study will provide important information on the quality of groundwater in the study area and help adopt an appropriate remedial management approach in other agricultural regions similar to Lioua of the lower Sahara of Algeria.

2. Materials and Methods

2.1. Study Area

The province of Biskra is located in the southeast of Algeria and, more precisely, in the south of the Aurès, constituting its natural limit in the north. It extends to the southeast to the Chotts area (Chott Melghir) and to the southwest to the beginning of the great eastern erg. Geo-physically, the wilaya extends over an area of 12,755 km² and is made up of large geographical units: to the north, a mountain and high plateau area, to the southeast, an area of the Chotts to the east and to the south, an area of plateaus and steppe plains on the El Outaya–Doucen axis in the center. The majority of the lands of the wilaya have a large flat expanse followed by a low-pressure area, that of Chott Melghir, which is shown in Figure 1.

Lioua is a commune of Biskra Province that is located about forty kilometers south of Biskra (Figure 1). Located on Oued Djedi; it extends over an area of approximately 242.1 km² at an altitude of 94 m. It is part of the Zab Gharbi palm groves in the Oued Djedi watershed. The study area belongs to the lower arid bioclimatic level (200 < p > 100 m) with a strong continental influence. Rainfalls usually fall as heavy downpours. The average temperature throughout the year is 22.4 °C, with a high seasonal variation. The maximum temperature can reach up to 50 °C during the presence of superheated air coming from the south.

The region of Lioua has an agricultural vocation par excellence where intercropping market gardening and fruit trees are practiced intensively in the oases of date palms. The region’s agricultural land is estimated to be 7495 Ha; in accordance with the agricultural services of the wilaya of Biskra, there are 50 wells and 1160 boreholes there that irrigate 116 Ha and 4902 Ha, respectively. In arid areas and faced with the virtual absence of surface water resources, the exploitation of groundwater remains the only way to meet needs. Agriculture is a productive sector characterized by the greatest demand for water, mainly due to farming needs that are higher than rainfall contributions. In addition, the generalization of irrigation is being spread over large agricultural perimeters, where the productive potential of the soil is closely linked to the availability of water during the summer period [8].

2.2. Geological and Hydrogeological Setting

The region of Biskra is mainly characterized by sedimentary terrains, ranging from Barremian (Cretaceous) at the base to Quaternary at the top. The lower Sahara is, in fact, a vast backfilling plain, which has slowly subsided since the upper Cretaceous to the Quaternary. This basin is filled with post-Eocene continental Tertiary deposits, made up of agglomerated sands intercalated with clay layers and clay-sand banks [33]. The Neogene deposits, which are mainly the product of the dismantling of the Atlas chain, completely mask the underlying folded structures. In addition, it is only thanks to geophysics and drilling logs that this structure has been updated [30].

Litho-stratigraphically, the Lioua region is made up of formations of ages ranging from Secondary to Quaternary. Secondary formations (Cretaceous) consist of limestone, crystalline limestone, dolomites sandstone, gypsum, anhydrite, clays, and marls, while Tertiary formations (Paleogene and Neogene) consisting of limestone, marls, gypsum, clays, gravel and sand red. The Quaternary formations consist of scree, pebbles, gravel, sand, gypsum limestone, sandy alluvium, and clay alluvium (Figure 2).

For the hydrogeological context, geological and hydrogeological studies have made it possible to highlight the existence of several aquifer reservoirs of very distinct importance in terms of their lithological constitution, their geological structure, and the ease with which they are exploited. These aquifers belong to the Quaternary, Mio-Pliocene, Lower Eocene and Upper Senonian (Maastrichtian), and Albian. The Albian aquifer, also called the continental intercalary (CI) aquifer, is by far the most important reservoir in the region since it covers most of the northern Saharan territory. Thus, the groundwater belongs to a complex hydrogeological basin whose main aquifer reservoirs are shown in Figure 3 and Figure 4.

The water table of Oued Djedi particularly represents the superficial water table. The thickness of this water table varies between 50 and 70 m, with an average flow of 20 L/s. The quality of the water extracted is poor to average; only south of Lioua is it is of good quality. The aquifer reservoir is heterogeneous and consists of detrital materials (pebbles, gravel, and sand). The substratum is made up of a thick clay formation; sometimes, it appears in the form of lenses of sand in discordance with the layers of clay. The water table is fed by rainwater, seepage from the Oueds, and irrigation water.

The Mio Pliocene sands table is essentially constituted by an alternation of sands, gravels, and clays. It is strongly exploited by a very important number of drillings intended essentially for the irrigation of agricultural lands. The thickness of this aquifer varies from 80 to 140 m with a flow rate of 5 to 15 L/s. The aquifer is made up of several producer levels with a heterogeneous composition: detrital materials, gravel, and sand wrapped in a clay matrix. At depth, the formation becomes predominantly sandy clay and rests on an impermeable formation composed of gypsum marl and anhydrite of the middle Eocene. The sands sheet is covered by a shallow alluvial deposit or a Quaternary gypsum sand layer. In places, the Mio Pliocene outcrops bring this aquifer into direct contact with the surface, thus ensuring its supply from surface water [34].

The Lower Eocene limestone aquifer is the most solicited aquifer in the Ziban region (Zab El-Gharbi). It has been well known for a long time for its artesianism and its natural outlets, which are the north-east springs of the Lioua region (Oumache, M’lili, Megloub). The thickness of the aquifer ranges from 150 to 250 m with a flow of 5 to 40 L/s. This aquifer contains important reserves, which are linked to the facies, the state of fissuring of the rock, and to the underground recharge from the Saharan Atlas. Its roof is constituted in the north by sandy clay formations of the Mio Pliocene and in the South by marls with gypsum of the middle Eocene, contributing to its loading [35]. The reservoir of the limestone nappe is essentially made up of limestone of the lower Eocene, upper Senonian, and Turonian.

The continental intercalary aquifer (Albo-Barremian) is a very deep nappe (more than 2000 m), often called Albian. It is characterized by warm waters whose temperature can exceed 60 °C and have a poor chemical quality. The thickness of this water table varies from 250 to 300 m with a flow of more than 25 to 120 L/s and it is characterized by its artesianism. It is made up of sandstone, limestone, and clay. It is not solicited in the region, except in Ouled Djellal and Sidi Khaled where the Albian or Barremian sandstone formations are found at depths of 1500 to 2500 m (Figure 3).

2.3. Groundwater Sampling and Analysis

Groundwater hydrochemical analysis is an important aspect of hydrogeological research, which guides the sustainable use and management of groundwater resources. Thirty-one (31) water samples were collected from different electrically pumped wells at different locations in the study area to determine the chemical parameters of the water samples. These included 12 samples collected by the authors at the beginning of the rain season (January 2021) and the rest were collected at the end of the rain season (May 2021). To determine the location sampling points, a global positioning system (GPS) (Model: GPS map 76 CSx) was used, as presented in Figure 2. It is assumed that during/after the rainy season, pollutants may have been subjected to downward leaching and thus contaminating the underlying aquifers [36]. After stabilizing the water temperature and pumping for about 15 min to remove groundwater that had been stored in the hydraulic structure, samples were collected using two polypropylene (PP) bottles that had been washed with acid in accordance with the American Public Health Association’s method (APHA) [37]. Each sample was promptly filtered locally using cellulose acetate 0.45-m filters. Filtrate was put into 100 cm³ polyethylene (PE) bottles for cations studies, where it was instantly acidified with ultrapure nitric acid to pH 2. (5 mL 6N HNO₃) (Germany’s Merck). Samples for anions analysis were put into 250 cm³ plastic bottles without being acidified. Our samples were kept cold, at around 4 °C, in an ice chest; then, they were transferred to the laboratory and analyzed.

The physicochemical parameters (electric conductivity, temperature, and pH) were measured using a WTW multiparameter (P3 MultiLine pH/LF-SET), Welheim- Germany. The WTW Multi-Line P3/LF-SET multi-parameter meter is equipped with a simultaneous connection of a pH/Redox electrode and a conductivity cell, which makes it possible to measure all three parameters at the same time with a linear and non-linear temperature compensation function for ultra-pure and natural waters. Chemical analyses were carried out at the Water Control and Quality Laboratory of the Biskra Unit of the Algerian Water Company (ADE Biskra). The main dissolved chemical components of the water samples were analyzed. HCO₃⁻ is measured by titration endpoint methyl orange immediately after sampling. Ca²⁺ and Mg²⁺ were determined by complex titration. We determined the chloride concentration by titrating and precipitating AgCl until silver chromate appeared. SO₄²⁻ was determined using spectrophotometry using a Mecasys UV/VIS Spectrophotometer 5U5707-14021-00. Na⁺ and K⁺ were analyzed by flame photometer and NO₃⁻ was determined by cadmium column reduction method. The TDS was estimated by weighing and drying at 103–105 °C in an oven. Nitrates were determined using the cadmium column reduction method. All samples were analyzed in triplicate with analytical uncertainty of less than 4%. To check the correctness of the analysis, the cations–anions balance was used, where it was within ±5%, indicating the reliability of the chemical analysis. This was achieved by applying standard methods of the APHA [37]. The respective ionic balance is generally around 5%. Hydrochemical results of all samples were statistically analyzed by using DIAGRAMMES version 5.8. The results of the time series analysis and the multivariate statistical techniques were produced by STATISTICA (version 14). Geographic Information System (GIS) was used to create distribution maps of the principal (anion and cation) ions in the area under study, specifically with the software ArcGIS (version 10.5.X).

2.4. Methodology

2.4.1. Data Treatment and Multivariate Statistical Analysis

Chemometric analyses, such as PCA and CA, have been widely employed as objective approaches for the retrieval of crucial information from the hydrochemical dataset in order to comprehend the origins of the primary ions and the geochemical processes impacting the quality of the groundwater [17,38,39]. In order to extract the most important components and to decrease the data with the least amount of information lost, principal component analysis (PCA) techniques are typically used to examine the correlations between various sets of groundwater hydrochemical data [1]. The chemometric analyses help to simplify and organize large data sets to provide meaningful insight. PCA is a statistical technique for reducing the initial inputs of data variables into a small number of principal components for better interpretation of the data [40,41]. The first principal component (PC1) accounts for the majority of the variance in the data set, with each succeeding component accounting for a decreasingly variance [42]. Standardization of the chemical parameters is done before the PCA is applied (scale z). In order to remove any potential biases toward a specific parameter of the distinct unit at large concentrations, this standardization makes each of the parameters dimensionless [43,44]. Principal components (PCs) were extracted using the varimax rotation approach, with eigenvalues greater than 1 considered important for interpretation [45]. In this research, the PCA method was applied to hydrochemical data obtained from the Lioua plain to attain significant information in order to understand the geochemical processes influencing the chemical composition of the groundwater and quality in the studied area.

Using cluster analysis (hierarchical clustering), one can categorize vast amounts of data into groups based on a predetermined set of traits. To identify real data groups, CA combines several multivariate approaches [46]. Based on similarities and differences between objects, the main goal of CA is to locate relatively homogenous groupings or clusters of those objects [47]. First, related objects are grouped together; then, when the degree of similarity increases, all subgroups are combined into a single cluster. In the end, this can help in identifying patterns that might be significant [48]. It is well known that in clustering, items are grouped together so that they belong to the same class [49]. Using Ward’s approach and Euclidean distance as a measure of similarity, cluster analysis (CA) is conducted on the standardized data set to produce a dendrogram.

Therefore, Ward’s method as a linkage rule for classifying the hydrogeochemical data is applied in the present study. The Euclidean distance was used as a measure of distance between samples, which is one of the most commonly adopted measures [50].

2.4.2. Time Series Analysis

The main application of a time series analysis applied to environmental systems is to understand seasonal changes and/or trends over time. Nevertheless, understanding and modeling the correlational structure in the time series is another objective that is frequently of utmost relevance. On stationary processes, this kind of study is typically performed. Any process that remains consistent across time is said to be stationary. In other words, a stationary time series is one that has no periodic variations and no systematic change in its mean or variance [51].

A short overview of the mathematical expressions of autocorrelation and spectral density is presented. The theory of correlation has been highlighted in detail in [52] and [53]. According to [54], a simple autocorrelation analysis provides quantitative information for linearly dependent successive values in time. Autocorrelation function r(K) is expressed using autocovariance C(K) and variance C(0) of the time series, such as in (Equation (1)):

$$\mathrm{r}\left(\mathrm{k}\right)=\raisebox{1ex}{$\mathrm{C}\left(\mathrm{k}\right)$}\!\left/ \!\raisebox{-1ex}{$\mathrm{C}\left(0\right)$}\right. \mathrm{c}\left(\mathrm{K}\right)=\frac{1}{\mathrm{n}}{\displaystyle \sum }_{\mathrm{t}=1}^{\mathrm{n}-\mathrm{k}}\left({\mathrm{x}}_{\mathrm{t}}-\overline{\mathrm{X}}\right)\left({\mathrm{x}}_{\mathrm{t}+\mathrm{k}}-\overline{\mathrm{X}}\right)$$

(1)

where k is a time lag (k = 0 − m), n is the number of events, ${\mathrm{x}}_{\mathrm{t}}$ is a single event, X is the mean of events and m is the cutting point. The analysis interval and the current situation are typically used to identify the cutting point. The autocorrelation function shows a slowly diminishing slope along with the non-zero values over a long lag in situations when time series are highly correlated and also involve a long-memory impact. However, when uncorrelated, such as when it rains, the autocorrelation function rapidly declines and shortly reaches zero [55]. The autocorrelation study is complemented by a straightforward spectral density analysis. By performing a Fourier transformation on the autocorrelation function, the spectral density function denotes a change from a time mode to a frequency mode [56]. The spectral density function, S(f), can be interpreted by identifying the different peaks, which stand for periodical phenomena. S(f) characterizes the system as in (Equation (2)):

$$\mathrm{S}\left(\mathrm{f}\right)=2\left[1+2{\displaystyle \sum }_{\mathrm{k}=1}^{\mathrm{m}}\mathrm{D}\left(\mathrm{k}\right)\mathrm{r}\left(\mathrm{k}\right)\cos \left(2\mathsf{\pi }\mathrm{fk}\right)\right]\mathrm{D}\left(\mathrm{k}\right)=\frac{\left(1+\cos \mathsf{\pi }\frac{\mathrm{k}}{\mathrm{m}}\right)}{2}$$

(2)

where f is the frequency and D(k) ensures that estimated values S(f) are not biased.

3. Results

Thirty-one groundwater samples from a highly agricultural plain in the Lower Sahara of Algeria were experimentally obtained and analyzed. A descriptive statistical summary of the analyzed groundwater samples’ parameters, along with the drinking water standards of the World Health Organization [57], is provided in

Table 1. Descriptive analysis results of the groundwater samples in the study area.

Parameters	Min	Mean	Max	St Dev	Coef. of Variation (CV, %)	WHO (2017) Second Addendum 2021
Ca2+	252	452.71	640	117.89	0.26	150
Mg2+	52.8	172.61	439.2	84.66	0.49	70
Na+	47.73	373.16	1358.54	260.41	0.7	200
K+	5.01	12.02	25.63	4.66	0.39	12
HCO3−	129.32	187.6	283.04	43.32	0.23	500
SO42−	571.43	1636.22	3028.57	647.01	0.4	250
Cl−	106.5	462.07	1462.6	277.65	0.6	250
NO3−	0.2	37.13	153.44	30.31	0.82	50
pH	7.03	7.64	8.16	0.3	0.04	6.5–8.5
EC	2000	4370.97	10000	1617.88	0.37	1500
TDS	1232.00	3333.45	7145.00	1269.12	0.38	1000

Remark: units in mg/L except for pH (unitless) and EC (µS/cm); St Dev for standard deviation.

. The spatial distribution map of major ions (Ca²⁺, Mg²⁺, Na⁺, K⁺, HCO₃⁻, SO₄²⁻, Cl⁻, NO₃⁻) of the groundwater samples in the study region are summarized and discussed herein (Figure 5 and Figure 6).

3.1. Hydrochemical Water Types and Groundwater Natural Evolution Mechanisms

The hydrochemical activities taking place in the aquifer may have an impact on the chemistry of groundwater [58]. The Piper diagram [59] (Figure 7) was used in this work to represent how key ions regulate the hydrochemical groundwater types. According to the cation–anion pairings, the plotting position reveals the relative composition of groundwater [60]. The Piper diagram’s diamond-shaped field demonstrates the existence of one of the main water types: Ca-Mg-SO₄.

For a better understanding of hydrochemistry and to compare the water types, Chadha’s diagram [61] was plotted (Figure 8). In this study, all the samples are plotted in the sixth field, representing Ca²⁺–Mg²⁺–SO₄²⁻ type, where Alkaline earths exceed alkali metals and strong acidic anions exceed weak acidic anions.

Gibbs plots [62] were used to investigate the main mechanisms influencing the development of water and the many hydrogeochemical processes regulating groundwater chemistry in the research region. Evaporation is a significant impact in the evolution of groundwater chemistry, as shown by the Gibbs plots (Figure 9), which show that all of the analyzed groundwater samples lie in the top half of the diagrams.

Other approaches are needed to reveal the origin and the relationship between the major elements, including: [Ca²⁺ + Mg²⁺] versus [SO₄²⁻ + HCO₃⁻] (Figure 10a), [SO₄²⁻] versus [Mg²⁺/Ca²⁺] (Figure 10b), [Ca²⁺ + Mg²⁺] versus [HCO₃⁻ − SO₄²⁻] (Figure 10c), [Mg²⁺/Ca²⁺] versus [Mg²⁺/Na⁺] (Figure 10e), [Mg²⁺/Na⁺] versus [Ca²⁺/Na⁺] and [HCO₃⁻/Na⁺] versus [Ca²⁺/Na⁺] (Figure 10f) [63]. In order to recognize the ion exchange process between groundwater and aquifer minerals, both chloroalkaline indices were applied (CAI-I and CAI-II) (Equations (3) and (4); Figure 10d).

$$\mathrm{CAI}-\mathrm{I}=\frac{{\mathrm{Cl}}^{-}-({\mathrm{Na}}^{+}+{ \mathrm{K}}^{+})}{{\mathrm{Cl}}^{-}}$$

(3)

$$\mathrm{CAI}-\mathrm{II}=\frac{{\mathrm{Cl}}^{-}-({\mathrm{Na}}^{+}+{ \mathrm{K}}^{+})}{{\mathrm{SO}}_4^{2-}+{ \mathrm{HCO}}_3^{-}+{ \mathrm{CO}}_3^{2-}+{ \mathrm{NO}}_3^{-}}$$

(4)

3.2. Statistical Analysis

3.2.1. Correlation Analysis and Elementary Statistics

Furthermore, correlations between various chemical components and TDS and EC have been used extensively in studies of the acquisition of salinity mechanisms. When the samples’ physicochemical data were applied, Pearson’s correlation matrix showed a moderate to high correlation value (0.62–0.96) between EC, Ca²⁺, Mg²⁺, Cl⁻, Na⁺, K⁺, and SO₄²⁻ (Supplementary Materials). This demonstrates that these substances are the salinity’s primary constituents.

Coefficient of variation (Cv) is usually used to characterize the stability of the variable, which represents the ratio of the standard deviation (SD) to the mean. Where 0 < Cv < 10 percent for weak variation; 10% < Cv < 100% for moderate variability; and Cv > 100% strong variation [64]. In this study, the statistical analysis results of 31 groundwater samples from the study area are presented in Table 1.

3.2.2. Principal Component Analysis (PCA)

In order to analyze the data from the 31 samples and 10 variables, PCA was used. More than 89.6% of the total variance was represented following the analysis on three factors, and Figure 11 shows the parameter weights for the three components from the PCA of the dataset. The percentage of variance, expressed by the first factor (68.72%), shows that there is a fairly good structure in the sampling carried out. This proves that many factors that affect the structure of the samples are linked to each other. This was reflected in the correlation matrix (Supplementary Materials), where there was a significant correlation between Na⁺, Cl⁻, Mg²⁺, K⁺, SO₄²⁻, Ca²⁺, and conductivity/TDS. These correlations highlight the origin of the salinity of the water of the Plio Quaternary aquifer of Lioua.

3.2.3. Cluster Analysis (CA)

According to [65], to find clusters, we can use two alternative techniques, such as R- or Q-modes. R-mode is typically used with water quality factors to demonstrate how they interact (Figure 12). In the current study, the hydrogeochemical data are categorized using Ward’s approach as a linkage rule. One of the most often used metrics of the distance between samples is the Euclidean distance, which was utilized [50].

HCA was used to calculate the degree of similarity between the groundwater samples using a combination of Ward’s linkage technique and Euclidean distance. Figure 12 displays the spatial HCA dendrogram that was created. All variables were log-transformed and roughly matched normally distributed data for statistical reasons. The variables were then normalized to their respective standard scores (z-scores), as per [66]. The only criterion for choosing the groups in the Dendrogram is eye inspection because there is no specific test to establish the ideal number of groups in the dataset.

3.2.4. Time Series Analysis

Autocorrelation

The autocorrelation functions of selected hydrochemical parameters such as pH, Ca²⁺, Mg²⁺, HCO₃⁻, SO₄²⁻, Cl⁻, Na⁺, K⁺, NO₃⁻ and EC are shown in Figure 13. The experimental variograms for each of the ten variables as well as the auto variogram were generated to evaluate the autocorrelation effect. Variogram functions were estimated overall (Figure 13).

Spectral Density

Spectral density functions of all the hydrochemical variables are displayed in Figure 14. The functions behave similarly, and their greatest peaks often have just one or two points. The correlograms of Mg²⁺, Na⁺, K⁺, Cl⁻, and SO₄²⁻ follow a pattern that is almost identical to that of EC.

4. Discussion

Several natural and anthropogenic factors can affect the evolution of groundwater chemistry. Thus, the application of statistical tools such as descriptive statistics, multivariate prospecting techniques, and time series has allowed us to understand the mechanisms of groundwater mineralization acquisition in the Lioua plain. The obtained results show that the mean temperature of the water was 22.7 °C with a minimum of 11.4 °C and maximum of 49.2 °C. The pH value in the study area ranges from 7.03 at F-03 to 8.16 at F-28 with a mean value of 7.64, indicating a weakly alkaline environment, as shown in Figure 5c. These values were found to be within the drinking water standards prescribed by the World Health Organization (WHO) [57], ranging from 6.5 to 8.5 (Table 1). The presence of calcium carbonate and magnesium carbonate contributes carbonate ions to the buffering system. Alkalinity is commonly related to hardness, as the main alkalinity source often comes from carbonate rocks (limestone), made up mostly of CaCO₃ [67,68]. It is well known that the pH range of 6.5 to 7.5 is where the buffering of calcite occurs most frequently [69].

The estimation of total dissolved solids (TDS) is crucial for understanding the relationship between the environment and groundwater chemistry [70]. The suitability of groundwater with a TDS value above 3000 mg/L is often considered poor, while a TDS level above about 1000 mg/L exceeds the guideline value for human consumption, according to [57]. The average TDS for all subterranean water samples was 3333.45 mg/L, with F-17 recording the highest TDS at 7145 mg/L and F-26 the lowest at 1232 mg/L (Table 1). According to the classification of [71], the samples from the study area were classified as unsuitable for consumption and irrigation. The TDS spatial variation map (Figure 5a) shows that the TDS values increased in the southwestern and northeastern limits, which could be due to the geological characteristics and anthropogenic factors of the study area. The increase in the TDS coincides with the course of the Oueds of the study region Oued Bou-Mlih and Oued El-Ouzenn in the south-west and Oued Djedi, which crosses the plain from the south-west to the northeast. This configuration leads us to think that the runoff water towards the low points (Oueds) leads to their enrichment in dissolved salts by leaching of the saline soils before their infiltration in the basement via the Quaternary matrix. Likewise, it is known that excessive irrigation leads to increased water salinity, especially since the region is known for its intense agricultural activity.

Electrical conductivity (EC) is the measure of water’s ability to carry electrical current; it makes it possible to evaluate the total mineralization of the water. EC shows extremely variable values ranging from 10000 µS/cm measured at point F-30 in the southwest to 2000 µs/cm measured at point F-1 in the south-east, with an average value of 4370.97 µS /cm and a coefficient of variation (CV) of 37% (Table 1). However, all the samples’ values have been found to be above the guideline value for human consumption, as restricted to 1500 µS/cm according to [57]. Regarding the spatial variation of EC, the higher values were observed in the southwest and in a small area in the northeast of the study region (Figure 5b). This significant variation in electrical conductivity from one point to another shows heterogeneity in the distribution of the mineral load in the groundwater of the surface aquifer of the Plio Quaternary of the Lioua region, as in Figure 5b. This suggests the predominance of the phenomenon of the dissolution of salt minerals and the transport of agricultural inputs through the leaching of the land, which consequently leads to an increase in the mineral load in these waters. In the study area, soil salinization is serious and thus requires considerable attention.

Calcium is the preferred element of carbonate rocks. The most prevalent source of calcium in groundwater is limestone, which is created when precipitates of CaCO₃ dissolve in sedimentary rocks during groundwater recharge [72,73]. The concentration of this element in groundwater is mainly controlled by the solubility of certain minerals such as carbonate minerals (calcite and dolomite), gypsum, or silicates [74]. Calcium ion concentration in our groundwater samples varied from 252 to 640 mg/L, with an average concentration of 452.71 mg/L (Table 1). The calcium values in all samples were above the guide value of 150 mg/L for drinking water, which was prescribed by [57]. The spatial variation map of Ca²⁺ (Figure 6a) indicates that the highest contents were located in the west and the east center of the study area. These concentrations indicate that these waters are influenced by the dissolution of gypsum formations (Equation (5)).

$${\mathrm{CaSO}}_4+2\left({\mathrm{H}}_2\mathrm{O}\right)={\mathrm{Ca}}^{2+}+{\mathrm{SO}}_4^{2-}+2\left({\mathrm{H}}_2\mathrm{O}\right)$$

(5)

Magnesium comes from the dissolution of carbonate and salt formations. The ion exchange of ferromagnesium minerals in rocks and the water-induced dissolution of dolomite and soils can be used to explain the primary source of magnesium (Mg) in groundwater [75,76]. In the case of magnesium, the Mg²⁺ ions in water samples ranged from 52.8 to 439.2 mg/L (Figure 6b), with an average value of 172.61 mg/L in all groundwater samples of the study area (Table 1). The highest concentration was found in the F-11 (439.2 mg/L). For human consumption, the recommended magnesium content is 70 mg/L [57]. According to the spatial map of magnesium, 42% of our samples fall under the guide value for drinking water (Figure 6b). The waters of the Lioua plain are rich in magnesium, which leads us to consider that they were enriched following the dissolution of evaporites and salt formations such as clays—which are rich in Mg²⁺—during the recharge of the water table and the leaching of rich soils in magnesium minerals (Debdaba region).

Sodium is found in evaporites (halite NaCl, mirabilite Na₂SO₄ (10(H₂O)). The presence of sodium is linked to the rapid dissolution of evaporitic formations rich in halite. The weathering of rock-forming minerals and agricultural activities may be the main sources of Na in groundwater [76]. Na⁺ concentration is in the range of 47.73 to 1358.54 mg/L, with an average of 373.16 mg/L. Only 29% of the samples have a Na concentration less than the value guide of drinking water quality of 200 mg/L (Table 1). As the spatial distribution map of sodium (Figure 6f), the highest Na concentration was located in the northeast (Debdaba region) parts of the study area, exactly in the F-30. These concentrations testify to a salt supply from the evaporites. Nevertheless, it is probable that during their journey, the groundwater may undergo mineralization in contact with the clays constituting the matrix of the Plio Quaternary aquifers of the region per the phenomenon of base exchange by fixing a Ca²⁺ ion after the release of two Na⁺ ions.

In general, potassium rarely exceeds 10 or 15 mg/l in natural waters. The weathering of potash feldspar minerals is likely what causes the greatest potassium concentrations [77]. The minimum and maximum concentrations of potassium in the groundwater samples of the study area varied from 5.01 to 25.63 mg/L, with an average of 12.02 mg/L (Table 1). Accordingly, [57] suggests the value guide of K in drinking water as 12 mg/L. Since the spatial distribution map of K (Figure 6e) revealed that 45% of samples in the study region fall below the value guide of [57] (Figure 6e). These potassium levels come from the alteration of potassic clays and the dissolution of chemical fertilizers (NPK), which are used massively by farmers in the region.

The alkalinity of groundwater is mostly caused by the presence of HCO₃⁻ and CO₃²⁻ ions. The weathering of silicate rocks such as feldspar and the dissolution of carbonate rocks such as dolomite (CaMg(CO₃)₂) and calcite (CaCO₃) in aquifers [78] and as described in Supplementary Materials are two significant sources of HCO₃⁻ ions in groundwater.

HCO₃⁻ concentrations are in the range of 129.32 to 283.04 mg/L, with an average of 187.6 mg/L. According to the spatial distribution map of bicarbonate in the study area (Figure 6c), all samples have HCO₃⁻ concentrations which exceed the value guide of 120 mg/L [57]. The research areas F-31 in the north-eastern portion of the study area had the greatest bicarbonate concentration (283.04 mg/L), while F-16 in the southwest portion of the study area had the lowest bicarbonate concentration (129.32 mg/L) (Figure 6c).

The presence of sulfate ions in water is linked to the dissolution of gypsum formations, the degradation of organic matter in the soil, and anthropogenic input (agricultural origin), expressed in Equation (6):

$${\mathrm{CaSO}}_4+2{\mathrm{H}}_2\mathrm{O}={\mathrm{Ca}}^{2+}+{\mathrm{SO}}_4^{2-}+2{\mathrm{H}}_2\mathrm{O}$$

(6)

Corrosion may be impacted by groundwater with high concentrations of Cl⁻ and sulfate network systems for distributing water and phenomena [79]. The values of sulfates in water samples varied from 571.43 to 3028.57 mg/L, with an average of 1636.22 mg/L (Table 1). According to the spatial variation map (Figure 6g), the most significant concentrations of SO₄²⁻ were observed in the southwestern parts of the study area. Moreover, all groundwater samples have SO₄²⁻ concentrations that are higher than the drinking water quality guidelines offered in [57] (250 mg/L). The elevated SO₄²⁻ contents, as presented in the spatial distribution map (Figure 6g), may be attributed to the dissolution of gypsum minerals and other anthropogenic influences.

The origin of chlorine is mainly linked to the dissolution of salt formations. Evaporites are the main source of this element. The effect of endorheic basins, the salting of roads, and wastewater discharges can also be the cause of this element [5]. In unconfined aquifers, the chloride concentration is directly related to the chloride content of precipitation. The concentrations measured in these systems depend on the lithology. In confined aquifers, high chloride contents are to be expected in the presence of evaporitic formations rich in chlorine [80]. Chloride concentrations for all groundwater samples ranged from 106.5–1462.6 mg/L, with an average value of 462.07 mg/L (Table 1). The highest contents (1462.6 mg/L) were registered in F-11 in the west-center part of the study area (Figure 6d). Of the groundwater samples, 74% had a Cl⁻ concentration above the drinking water quality guidelines outlined in [57]. The high chloride concentrations in the groundwater samples, in particular of the west-center part of the study area, may be due to salt inputs from evaporitic formations and probably from the agri-food industry in the region or linked to urbanization through wastewater discharges.

Nitrates can have several origins. The main anthropogenic activities that affect the high concentration of NO₃⁻ include return flow irrigation, extensive irrigation with excessive inorganic nitrogenous fertilizer application, and domestic and industrial discharges [5,81]. A high concern exists around nitrate pollution since it may harm ecosystems [82]. NO₃⁻ ion is a familiar pollutant in water [83]. The maximum value was obtained at F-6 at 153.44 mg/L and the lowest at 0.2 mg/L at F-12, with an average value of 37.13 mg/L in the groundwater of the study area (Table 1). For the spatial variation map (Figure 6h), the highest contents were observed in the central parts of the study area, which coincides with village Lioua and agricultural areas.

Cations are displayed in zones B and A in Figure 7, demonstrating that the groundwater in the study region is primarily of the calcium type and does not have a dominating type. Samples for anions are typically found in zone F, where the sulfate type is highly prevalent. All groundwater samples in the study area are plotted in zone 1, such that SO₄²⁻, Ca²⁺, and Mg²⁺ are the major ions. The examination of the molar concentrations of different elements in the area shows that the cations evolve as follows: Ca²⁺ > Mg²⁺ > Na⁺ > K⁺ while the anions evolve in the following manner: SO₄²⁻ > Cl⁻ > HCO₃⁻ > NO₃⁻. The chemical profile is calcium and magnesium sulfate facies that appear due to the dissolution of evaporitic formations. The region’s lithology has a significant impact on the distribution of the principal ions (Ca²⁺, Mg²⁺, and SO₄²⁻), which are caused by anthropogenic factors such as irrigation water quality and unrestricted fertilization. The breakdown of calcium or magnesium sulfates can yield both calcium and magnesium. Dilution following mixing or precipitation of one of the ions can be used to change from one dominating ratio to another. The decomposition of organic matter in the soil and the addition of leachable sulfates to fertilizers in the intensively farmed parts of the Lioua plain are the first two clearly identified sources of SO₄²⁻, while the dissolution of gypsum is the second [84]. The presence of the evaporate sequence allows the dissolution of gypsum according to Equation (6) [85].

The results obtained from Chadha’s diagram are similar to the results obtained from the piper plot (Figure 7). Such water has permanent hardness and acquires its mineralization in the process of reverse ion exchange. Therefore, when irrigated, they do not deposit residual sodium carbonate.

Figure 9 shows that the Lioua region is in an arid climate zone and has a high evaporation rate, which causes the plots to diverge into the evaporation dominance zone (Figure 9). On the other hand, the Gibbs diagram cannot be used to describe the hydrochemical evolution processes of groundwater, which are also influenced by human activities [86]. However, it is clear that, with the enhancement of water levels induced by the intensification of irrigation and effective rainfall associated with the shallow water depth in the region, evaporation has become the main driver of ion concentration. In addition, the upper unsaturated but moist matrix especially at the approach of the saturated zone is rich in evaporites, which leads to the precipitation of evaporites by evaporation that is eventually leached into the saturated zone. Consequently, all of this leads to an increase in salinity (TDS max = 7145 mg/L), as the water level rises, groundwater evaporation becomes more severe and sulfate groundwater is more affected by evaporation than bicarbonate groundwater [5]. Other studies that have verified the sulfate type of groundwater experiencing intense evaporation in the alluvial plain lend weight to this conclusion [87].

As shown in Figure 10a, both reverse ion exchange and ion exchange affect the aquifer chemistry in the study area of Lioua. The samples are close to the 1:1 line, indicating that the dissolutions of calcite, dolomite, and gypsum are the prevalent reactions in the system of the study region. Those under the 1:1 line exhibit the processes of ion exchange, where Ca²⁺ is retained in the soil and the Na⁺ is returned in the groundwater. Inversely, those above line 1:1 exhibit the reverse ion exchange, where the Ca²⁺ is emitted into the groundwater and the Na⁺ is retained in the soil. The samples above line 1:1, are enriched in calcium and magnesium. Therefore, this suggests that the dissolution of Ca²⁺ and Mg²⁺ from evaporites is greater than from carbonates. The Ca²⁺, Mg²⁺, and SO₄²⁻ contents are more related to the dissolution of evaporites, essentially gypsum, anhydrite, and magnesium sulfate. According to Appelo and Postma [88], the presence of sulfates in large amounts in groundwater could also be attributed to the dissolution of the anhydrite and pyrite. On average, the contribution of direct ion exchange and reverse ion exchange reactions is almost equal.

Figure 10b shows the ratio of Ca²⁺/SO₄²⁻ and Mg²⁺/SO₄²⁻. A high correlation between Ca²⁺ and SO₄²⁻ was found, with a value correlation coefficient of 0.74 and that of Mg²⁺ and SO₄²⁻ is 0.67. In the scatter plot between SO₄²⁻ and Ca²⁺ (Figure 10b), most of the samples were below the equiline (1:1), indicating the gypsum and anhydrite dissolution in groundwater [89]. The samples that showed results far from the 1:1 line indicate another source of SO₄²⁻, such as mineral weathering, ion exchange reactions, and agricultural activities.

In addition, the ion exchange mechanism commonly used to determine the occurrence of cation exchange processes is studied by plotting (Ca²⁺ + Mg²⁺ − HCO₃⁻ − SO₄²⁻) versus (Na⁺ − Cl⁻) [63]. It is agreed that the excess (Ca²⁺ + Mg²⁺) could be related to sources other than carbonate and gypsum. Most water samples are close to the y = −X line (Figure 10c) and only a few points deviate from this relationship, indicating that cation exchange plays a nontrivial role in controlling the hydrochemical components of groundwater. Excess Ca²⁺ can exchange Na⁺ from aquifer minerals, resulting in increased Na⁺ in the groundwater.

Chloroalkalinity has been used to identify the ion exchange mechanism between groundwater and aquifer minerals. In general, Ca²⁺ and Mg²⁺ in the aquifer matrix exchange with Na⁺ and K⁺ in groundwater. Both chloroalkaline indices are negative when there is an ion exchange between Ca²⁺ or Mg²⁺ in the groundwater and Na⁺ and K⁺ in the aquifer material, and positive when there is a reverse ion exchange. Na⁺ and K⁺ levels in the groundwater fall in the Lioua system, indicating direct ion exchange. Only seven of the groundwater samples had positive CAI values in CAI-1 and CAI-2 (Figure 10d). Therefore, the main cation exchange process involves the exchange of Ca²⁺ and Mg²⁺ with Na⁺ and K⁺.

Our research has revealed that magnesium is a minor component of soil salts and does not precipitate during the early stages of low silica water evaporation because its concentrations in water are only predicted to increase as magnesium is released from ferromagnesium minerals in the bedrock during chemical weathering [90,91,92]. As a result, the influence of soil salt leaching and salt precipitation during the initial stages of water evaporation will not modify the concentration of magnesium. The groundwater in the research area is one rocky zone plot according to the Mg²⁺/Cations pattern (Figure 10e). The samples were found in the rock-dominated zone and had average Mg²⁺/Na⁺ and Mg²⁺/Ca²⁺ ratios (Figure 10e). The plot shows that the major ions in the study area are different from the interactions between the water and the rocks, whereas soil–salt leaching may be a subdominant process and evaporation is only a minor process.

Mineral dissolution is likely to be the primary natural activity in charge of the principal solutes in natural water [92]. A Na-normalized molar ratio has been suggested by Gaillardet et al. [93] as a way to represent various hydrochemical processes under non-mixed conditions. The Na-normalized Ca²⁺ versus Mg²⁺ plot and Na-normalized Ca²⁺ versus HCO₃⁻ diagrams demonstrate how water–rock interactions, including silicate weathering and mild carbonate dissolution, affect the natural water samples in the Lioua plain. Figure 10f demonstrates that the main method used to regulate the concentration of groundwater solutes was silicate weathering.

When there are fewer components, PCA results are more effective [94]. Of the variance, 68.72% is explained by factor 1, which has highly negative loadings in EC, TDS, Cl⁻, Na⁺, Mg²⁺, Ca²⁺, K⁺, and SO₄²⁻, according to the factorial analysis, which were −0.98, −0.97, −0.94, −0.93, −0.91, −0.89, −0.70 and −0.82, respectively (Supplementary Materials). High EC and TDS scores are the outcome of lithologic variables and numerous hydrogeochemical processes controlling them (mineralized water). On the F1–F2 factorial plan (Figure 11), the factor F1 is determined negatively by TDS, EC, Cl⁻, Na⁺, Mg²⁺, SO₄²⁻, K⁺, and Ca²⁺, which highlight the origin of the salinity by weathering of silicate, limestone, halite, and gypsum, pyrite dissolution and various ion exchange processes in the water system [95]. Therefore, the factor can be termed a salinization factor.

Factor 2 has substantial negative loadings in NO₃⁻ and HCO₃⁻, which were −0.66 and −0.64 (Figure 11), and accounts for 11.96% of the dataset’s overall variance. This indicates that the element has no effect on the overall mineralization of the water, predicts the origin of HCO₃⁻ as a result of weathering of carbonates, and may indicate the impact of acid–base equilibrium conditions on the chemistry of groundwater [96].

On the factorial plane F1–F3, factor 3 explains 8.89 % of the total variance of the dataset (Figure 11) and shows significant negative loading in NO₃⁻ (−0.59), which predicts the association of this factor with chemical fertilizers, animal waste, crop residues and mineralization of soil and non-agricultural sources such as septic tanks or deep water mixing with surface water. As we know, long-term applications of huge volumes of fertilizer, including urea and commercial composite, which is a significant component of the return flow from the development of irrigation, have been made. The nitrification process quickly converts NH₄⁺, the primary component of fertilizers, to NO₃⁻ in hazardous conditions [97], as reported in Equation (7).

$${\mathrm{NH}}_4^{+}+2{\mathrm{O}}_2={ \mathrm{NO}}_3+{ \mathrm{H}}_2\mathrm{O} +2{\mathrm{H}}^{+}$$

(7)

The dendrogram of the ten physicochemical parameters (Ca²⁺, Mg²⁺, Na⁺, K⁺, Cl⁻, HCO₃⁻, SO₄²⁻, NO₃⁻, EC and TDS) can be divided into three main groups (Figure 12) and reveals that TDS and EC can be the major elements of distinction between parameters. G1 reveals a close association with SO₄²⁻, K⁺, Ca²⁺, and TDS, which reflects the major predominance of these elements in the chemical composition of groundwater in the region (sulfates and anhydride, calcium sulfates). This is clearly visible by the calcium sulfate chemical facies. Similarly, group G1 shows a distant association between K⁺ and Ca²⁺, which reflects the probably different origins of these two elements. G2 reveals a close association between Na⁺, Cl⁻, Mg²⁺, and EC, which reflects the major participation of salts and chlorides in the electrical conductivity of water. The G1 and G2 together reflect the dominance of the lithological factor in the overall mineralization of the waters of the Plio Quaternary surface aquifer of Lioua. G3 shows the dissociation of bicarbonates and nitrates from other chemical elements present in groundwater, which reflects the different origins of these two elements, carbonated for HCO₃⁻ and anthropogenic for NO₃⁻. It should be noted that these analyses corroborate well with the results of the analysis of the correlation matrix between variables.

The autocorrelation functions of pH recorded a high number. This is an indicator of the pH’s weak relationship to various other factors [98]. On the other hand, the autocorrelation functions of Mg²⁺, SO₄²⁻, Cl⁻, Na⁺, K⁺, and EC displayed a sine-wave pattern. Therefore, throughout the study period, specific factors had significant associations. However, mean autocorrelation coefficients decrease slowly from 0.7 to −0.3 for pH and Ca²⁺. Obtained results show a mean linear relationship of Mg²⁺, EC, K⁺, Na⁺, SO₄²⁻, and Cl⁻ autocorrelation. Furthermore, the parameters NO₃⁻ and HCO₃⁻ are an indicator of the uncorrelated characteristics with other parameters. In the study area, the NO₃⁻ resulted from human activity—fertilizer and domestic sewage. However, HCO₃⁻ resulted from the carbonate facies outcrops, [99] also report these results.

Spectral density functions of all the hydrochemical variables have shown that the salinity was affected by these elements and that EC showed a more important role in the quality of groundwater. The preparation of activated carbon from wooden or other materials has helped to receive efficient adsorbents for the removal of pollutants from water [100,101]. pH, Ca²⁺, HCO₃⁻ and NO₃⁻ exhibit multiple peaks, which means that several distinct cyclic mechanisms are present [102]. The multiple peaks of these elements initiated the anthropogenic activity of fertilizer, the carbonate facies domestic sewage, and outcrops [103,104].

5. Conclusions

In this study, the hydrochemistry of Lioua’s groundwater in Algeria’s Lower Sahara was experimentally explored using chemometrics methods. The aim was to deduce the geochemical processes influencing the chemical composition and to determine chemical elements’ origins. The multivariate statistical analysis and time series approaches were used. Principal component analysis (PCA) and cluster analysis (CA) were applied when autocorrelation analysis supplemented by simple spectral density analysis was used as time series methods. The main finding obtained is that PCA showed the three most important parameters proving a total overall variance (TV) of 85.01 %. Factors 1, 2, and 3 are 68.72, 11.96 and 8.89 % of TV. Those results indicate that the dissolution of evaporated mineral deposits, evaporation, and water–rock interaction processes are processes of a geochemical nature that influence the chemical composition. Within the CA, tree parameters are regulated by the EC and TDS. G1 showed a high correlation with K⁺, SO₄²⁻, TDS, and Ca²⁺, and G2 showed a high correlation between Cl⁻, Na⁺, EC, and Mg²⁺. G3 showed the bicarbonates dissociation and NO₃⁻ formation from other geochemical processes. These results indicate that the groundwater also exhibits two bipolar characteristics, one recorded with negative and positive charges on pH and on Ca⁺ and another recorded only with negative charges on HCO₃⁻ and NO₃⁻. The autocorrelation analysis showed a linear relationship of Mg²⁺, EC, K⁺, Na⁺, SO₄²⁻, and Cl⁻. While HCO₃⁻ and NO₃⁻ indicated that no correlation with characteristics of parameters from other compounds were seen. In addition, SDA shows that the correlograms of Mg²⁺, K⁺, Na⁺, SO₄²⁻, and Cl⁻ had a similar correlation toward EC. However, Ca²⁺, pH, NO₃⁻ and HCO₃⁻ showed multiple high values, possibly due to the existence of various distinct factors. The main participation of chlorides and other salts was due to the water’s electrical conductance. The combination of autocorrelation analysis and simple spectral density analysis results confirm the following findings: firstly, the prevalence of factors such as lithology in the mineralization of surface aquifer waters from the Plio Quaternary; secondly, that the presence of NO₃⁻ and HCO₃⁻ are from various sources—specifically, that carbonatization for HCO₃⁻ originates from carbonate while NO₃^– originates from human activities; thirdly, that the salinity levels were varied by SO₄²⁻, Mg²⁺, Na⁺, Cl⁻, EC and K⁺; and, finally, that HCO₃⁻, Ca²⁺, and NO₃⁻ enter the system from the agricultural use of fertilizers, and the carbonate facies from domestic sewage and outcrops. The results shown in the current study help elucidate the possibilities of the chemometrics methods when applied to hydrochemistry.