Geospatial Distributions of Groundwater Quality in Gedaref State Using Geographic Information System (GIS) and Drinking Water Quality Index (DWQI)

The observation of groundwater quality elements is essential for understanding the classification and distribution of drinking water. Geographic Information System (GIS) and remote sensing (RS), are intensive tools for the performance and analysis of spatial datum associated with groundwater sources control. In this study, groundwater quality parameters were observed in three different aquifers including: sandstone, alluvium and basalt. These aquifers are the primary source of national drinking water and partly for agricultural activity in El Faw, El Raha (Fw-Rh), El Qalabat and El Quresha (Qa-Qu) localities in the southern part of Gedaref State in eastern Sudan. The aquifers have been overworked intensively as the main source of indigenous water supply in the study area. The interpolation methods were used to demonstrate the facies pattern and Drinking Water Quality Index (DWQI) of the groundwater in the research area. The GIS interpolation tool was used to obtain the spatial distribution of groundwater quality parameters and DWQI in the area. Forty samples were assembled and investigated for the analysis of major cations and anions. The groundwater in this research is controlled by sodium and bicarbonate ions that defined the composition of the water type to be Na HCO3. However, from the plots of piper diagram; the samples result revealed (40%) Na-Mg-HCO3 and (35%) Na-HCO3 water types. The outcome of the analysis reveals that several groundwater samples have been found to be suitable for drinking purposes in Fa-Rh and Qa-Qu areas.


Introduction
Groundwater is a noble resource for water in arid and semiarid areas [1][2][3][4][5][6]. Accessibility to water is an important global goal whose effects are abundantly felt in developing countries. The benefit of DWQI was calculated by adopting weighted arithmetical index methods considering thirteen water quality parameters (pH, TDS, Ca +2 , Mg +2 , Na + , K + , Fe +2 , Cl − , HCO 3 − , SO 4 −2 , F − , NO 3 − , and E.C) in order to assess the degree of groundwater contamination and suitability for drinking purposes.
For the better understanding of geological units in this project, the thin sections of rock samples have been generated. With this ability, the rock mineral contents have been determined much better. This study has great importance; due to the plan for obtaining drinking water from the groundwater sources to Fw-Rh and Qa-Qu localities. However, this investigation is helpful in understanding groundwater environments and its suitability for human uses, especially in arid and semi-arid regions.

Geology
Geologically Figure 1; the lower Proterozoic rocks of the basement complex (Mainly Granitic Gneisses) [34], Syn-orogenic Granit and Syn-orogenic gabbro underlain the sandstone of Gedaref formation, Tertiary (Oligocene) basalt [35], Umm Rawaba formation, sand sheets and recent alluvium and wadi deposits. The groundwater of the area was taped at the sandstone of Gedaref formation sequence, alluvium soil, and fractures of the Oligocene basalt aquifers with depths ranging from 14 to 64 m. DWQI was calculated by adopting weighted arithmetical index methods considering thirteen water quality parameters (pH, TDS, Ca +2 , Mg +2 , Na + , K + , Fe +2 , Cl − , HCO3 − , SO4 −2 , F − , NO3 − , and E.C) in order to assess the degree of groundwater contamination and suitability for drinking purposes.
For the better understanding of geological units in this project, the thin sections of rock samples have been generated. With this ability, the rock mineral contents have been determined much better. This study has great importance; due to the plan for obtaining drinking water from the groundwater sources to Fw-Rh and Qa-Qu localities. However, this investigation is helpful in understanding groundwater environments and its suitability for human uses, especially in arid and semi-arid regions.

Geology
Geologically Figure 1; the lower Proterozoic rocks of the basement complex (Mainly Granitic Gneisses) [34], Syn-orogenic Granit and Syn-orogenic gabbro underlain the sandstone of Gedaref formation, Tertiary (Oligocene) basalt [35], Umm Rawaba formation, sand sheets and recent alluvium and wadi deposits. The groundwater of the area was taped at the sandstone of Gedaref formation sequence, alluvium soil, and fractures of the Oligocene basalt aquifers with depths ranging from 14 to 64 m.

Hydrogeological Setting
The groundwater was studied by using the data collected from forty boreholes drilled in Fw-Rh and Qa-Qu area as seen in Table 1. The hydrogeological characteristics of rock units were investigated, and aquifer systems were determined depending on field investigations and previous studies. Therefore, the hydrogeological map of the Fw-Rh and Qa-Qu area was settled adopting ArcGIS v. 10.5 software, based on characteristics of the lithological units Figure 2. According to these evaluations, the aquifer types were described as sandstone, alluvium, and fracture basalt.

Hydrogeological Setting
The groundwater was studied by using the data collected from forty boreholes drilled in Fw-Rh and Qa-Qu area as seen in Table 1. The hydrogeological characteristics of rock units were investigated, and aquifer systems were determined depending on field investigations and previous studies. Therefore, the hydrogeological map of the Fw-Rh and Qa-Qu area was settled adopting ArcGIS v.10.5 software, based on characteristics of the lithological units Figure 2. According to these evaluations, the aquifer types were described as sandstone, alluvium, and fracture basalt.  Alluvium  23  16  425  409  Ellewatah3  Sandstone  27  16  424  408  Abu Kalbo  Sandstone  64  37  423  386  Um Rakuba  Sandstone  48  27  425  398  Um Tireaza  Sandstone  21  11  444  432  Um Tireaza 2  Sandstone  60  13  554  505  Macancana  Sandstone  48

Spatial Interpolation and Groundwater Quality Mapping
Spatial interpolation is a procedure of predicting the value of attributes at unsampled sites from measurements made at point locations within the same area [36]. There are two main groupings of interpolation techniques: deterministic and geostatistical. Deterministic interpolation techniques create surfaces from measured points, based on either the extent of similarity (e.g., Inverse Distance Weighted) or the degree of smoothing (e.g., radial basis functions). Geostatistical interpolation techniques (e.g., kriging) utilize the statistical properties of the measured points.
In this study, we found that the Kriging (Ordinary and Simple) interpolation method is the most suitable method. Thus, the histograms and normal QQplots were plotted to examine the normality distribution of the observed data for each water quality element in both Fw-Rh and Qa-Qu localities.

Drinking Water Quality Index DWQI
DWQI has been determined based on the standards of drinking water quality as counseled by WHO. Therefore, thirteen chemical parameters (pH, TDS, Ca, Mg, Na, K, Cl, HCO 3 , F, NO 3 , Fe, and E.C.) were used for the calculation. To apply DWQI in the current study, the study area was divided into two parts, Fw-Rh, and Qa-Qu localities. The water quality parts were generated by a weighting factor and then formerly aggregated by using the simple mean calculations. To estimate the water quality in this project, the quality rating (Q i ) for all elements was estimated through the following equation; where, Q i = Quality ranking of the element form a total number of water quality elements, V a = Real amount of the water quality element taken from laboratory study, V i = Ideal rate of the water quality element can be realized from the standard Tables. V i for pH = 7 and for other elements it is equaling to zero. V s standard = Value of WHO standard. Then, the Relative weight (W r ) was studied from inversed proportional of recommended standard (S i ) for the corresponding parameter using the following expression; Here W r = Relative (unit) weight for specific element; Si = Standard allowable amount for certain element; I = Proportionality constant.
Assuredly, the total DWQI was determined using the assemblage equations of the quality rating with the unit weight linearly as the following: where Qi = Quality rating; Wr = Relative weight.
In general, DWQI is determined for particular and intended uses of water. In this work, the DWQI was estimated for human consumption, and the maximum DWQI value for the drinking purposes was regarded as 100 scores.
The methodology ideas in this work have been done through several steps Figure 3.

Results
Several factors may control the groundwater geochemistry such as the type of rock forming the aquifer, residence time of water in the hosted aquifer, the origin of the groundwater and the flow directions of groundwater. Hydro-chemical properties of the groundwater of the area are shown in Table 2. The water pH ranges between 7.5 and 8.9, indicate an alkaline chemical reaction in both sandstone and basaltic aquifers. The electrical conductivity (E.C) varies from 345 to 3342 μS/cm. (WHO)= The world health organization standard

Interpolation and Elements Distribution Maps
The quality of interpolation is described by the difference of the interpolated value from the true value. Thus, the Anderson-Darling test, which is an ECDF (empirical cumulative distribution function) based test, tests the prospect that the value of a parameter falls within a particular range of values (confidence level 95%). The data points are relatively close to the fitted normal distribution line. The p-value is greater than the significance level of 0.05. Subsequently, the scientist fails to reject the null hypothesis that the data follow a normal distribution.

Results
Several factors may control the groundwater geochemistry such as the type of rock forming the aquifer, residence time of water in the hosted aquifer, the origin of the groundwater and the flow directions of groundwater. Hydro-chemical properties of the groundwater of the area are shown in Table 2. The water pH ranges between 7.5 and 8.9, indicate an alkaline chemical reaction in both sandstone and basaltic aquifers. The electrical conductivity (E.C) varies from 345 to 3342 µS/cm. (WHO)= The world health organization standard.

Interpolation and Elements Distribution Maps
The quality of interpolation is described by the difference of the interpolated value from the true value. Thus, the Anderson-Darling test, which is an ECDF (empirical cumulative distribution function) based test, tests the prospect that the value of a parameter falls within a particular range of values (confidence level 95%). The data points are relatively close to the fitted normal distribution line. The p-value is greater than the significance level of 0.05. Subsequently, the scientist fails to reject the null hypothesis that the data follow a normal distribution.
According to this test, in Fw-Rh area we found that the parameters (Na and K) showed a normal distribution when the other elements (Ca, Mg, HCO3, Cl, SO4 and TDS) showed a more or less abnormal distribution in Figures 4  According to this test, in Fw-Rh area we found that the parameters (Na and K) showed a normal distribution when the other elements (Ca, Mg, HCO3, Cl, SO4 and TDS) showed a more or less abnormal distribution in Figures 4 and 5.  The same test has been performed in (Qa-Qu) area, which showed that the (Mg and SO4) parameters reflected normal distribution while the other variables (Ca, K, Na, HCO3, Cl, and TDS) present non-normally distributions.
Generally, most of the collected elements in both Fw-Rh and Qa-Qu localities were skewed. However, the transformations (Log & BoxCox), have been used to make the data normally distributed and satisfy the assumption of equal variability for the data.
For the maps prediction, several kinds of semivariogram models were examined for each water quality parameter to obtain the preferable one, as seen in Figure 6 as an example. Predictive performances of the fitted models were checked on the basis of cross-validation tests. The values of The same test has been performed in (Qa-Qu) area, which showed that the (Mg and SO 4 ) parameters reflected normal distribution while the other variables (Ca, K, Na, HCO 3 , Cl, and TDS) present non-normally distributions.
Generally, most of the collected elements in both Fw-Rh and Qa-Qu localities were skewed. However, the transformations (Log & BoxCox), have been used to make the data normally distributed and satisfy the assumption of equal variability for the data.
For the maps prediction, several kinds of semivariogram models were examined for each water quality parameter to obtain the preferable one, as seen in Figure 6 as an example. Predictive performances of the fitted models were checked on the basis of cross-validation tests. The values of mean error (ME), mean square error (MSE), root mean error (RMSE), average standard error (ASR) and root mean square standardized error (RMSSE) were estimated to ascertain the performance of the developed models. After conducting the cross-validation procedure, maps of kriged estimates were created that provided a visual representation of the distribution of the groundwater quality parameters in the Fw-Rh and Qa-Gu areas. mean error (ME), mean square error (MSE), root mean error (RMSE), average standard error (ASR) and root mean square standardized error (RMSSE) were estimated to ascertain the performance of the developed models. After conducting the cross-validation procedure, maps of kriged estimates were created that provided a visual representation of the distribution of the groundwater quality parameters in the Fw-Rh and Qa-Gu areas. Kriging (Ordinary and Simple) interpolation method is the most suitable method in the studied areas. The value range of the better interpolation models were observed and reported in Table 3. If the RMSE is close to the ASE, the prediction errors were assessed correctly. If the RMSE is smaller than the ASE, then the variability of the predictions is overestimated; conversely, if the RMSE is greater than the ASE, then the variability of the predictions is underestimated. The same could be deduced from the RMSSE statistic. It should be close to one. If the RMSSE is greater than one, the variability of the predictions is underestimated; also, if it is minimal than one, the variability is overestimated. After generating the cross-validation procedure, estimated maps of kriging were created, which gives a visual representation of the distribution of the groundwater quality parameters.  Kriging (Ordinary and Simple) interpolation method is the most suitable method in the studied areas. The value range of the better interpolation models were observed and reported in Table 3. If the RMSE is close to the ASE, the prediction errors were assessed correctly. If the RMSE is smaller than the ASE, then the variability of the predictions is overestimated; conversely, if the RMSE is greater than the ASE, then the variability of the predictions is underestimated. The same could be deduced from the RMSSE statistic. It should be close to one. If the RMSSE is greater than one, the variability of the predictions is underestimated; also, if it is minimal than one, the variability is overestimated. After generating the cross-validation procedure, estimated maps of kriging were created, which gives a visual representation of the distribution of the groundwater quality parameters.  The hydro-chemical of Fw-Rh area; the sodium concentration patterns in Figure 7a show similar trends to the potassium in Figure 7b

Correlation Matrix
The correlation matrix provides the assessment of the correlation coefficients "r" between groundwater quality elements. These coefficients are applied to suppress the strength of the linear relationship between the variables. It has been used to estimate both positive and negative correlations. The project area describes three examples of groundwater aquifers; (1) sandstone

Correlation Matrix
The correlation matrix provides the assessment of the correlation coefficients "r" between groundwater quality elements. These coefficients are applied to suppress the strength of the linear relationship between the variables. It has been used to estimate both positive and negative correlations.    (Table 5).

Groundwater Facies
Groundwater facies were defined by applying a Piper plot and Durov diagrams as seen in Figures 11 and 12. The descriptions reveal that the area consists of eight groups of groundwater types Table 6, Na-Mg-HCO 3 , Na-HCO 3 , Na-Ca-HCO 3 , Na-Ca-Mg-HCO 3 , Mg-Na-Ca-Cl, Mg-Na-HCO 3 , Na-Ca-Mg-HCO 3 -Cl, and Na-Mg-Ca-HCO 3 . The analytical results achieved from the samples when plotted on Piper's plot, explained that the alkalis (Na + , K + ), appear considerably over the alkaline elements (Ca +2 , Mg +2 ), and the weak acidic (HCO 3 − ) appear considerably over strong acidic anions According to the plotting from the Durov diagram, most of the elements of water plotted within the HCO3·Na zone, except some other samples that were fell in HCO3 Cl-Na, SO4·Cl·HCO3-Na, or HCO3·Cl-Na·Mg types.  Table 6. Groundwater facies distribution in the study area.

Drinking Water Quality Index (DWQI)
To gain a comprehensive representation of the quality of the drinking groundwater, drinking water quality index (DWQI) is one of the useful tools. It supplies a single amount to a state's overall water quality at a specific location and time, based on a number of water quality parameters. DWQI in Table 7, was calculated by adopting weighted arithmetical index method considering thirteen water quality parameters (pH, TDS, Ca +2 , Mg +2 , Na + , K + , Fe +2 , Cl − , HCO 3 − , SO 4 −2 , F − , NO 3 − , and E.C) in order to assess the degree of groundwater contamination and suitability.

101-200
Very Poor 3 (7.5%) 2 1 >>200 Unsuitable for drinking 3 (7.5%) 2 1 Thirteen thematic layers of water quality parameters were used in the ArcGIS environment to acquire the output of drinking water quality index DWQI maps for Fw-Rh Figure 13a, and Qa-Qu Figure 13b, localities. The water quality index was reclassified into five classes in order to characterize the quality of groundwater in the studied localities.

101-200
Very Poor 3 (7.5%) 2 1 >>200 Unsuitable for drinking 3 (7.5%) 2 1 Thirteen thematic layers of water quality parameters were used in the ArcGIS environment to acquire the output of drinking water quality index DWQI maps for Fw-Rh Figure 13a, and Qa-Qu Figure 13b, localities. The water quality index was reclassified into five classes in order to characterize the quality of groundwater in the studied localities.

Discussion
In this study, most of the boreholes were recently drilled (2015-2017), no other boreholes were available. Due to the few observation points, limited previous investigations and few hydrogeological data, using geospatial distributions, GIS and DWQI provide support in groundwater studies. As far as we know, no other study was conducted using the techniques in Fw-Rh and Qa-Qu areas.
The chemical composition and elements concentration of groundwater, are related to the rocks lithology and time residence of the water in the aquifers. To identify the effects of the reaction between the groundwater and the (geological units) aquifer, the bivariate diagrams were applied to explain the chemical changes in ionic concentrations in the host rocks and groundwater. The reaction between water and the surrounding surface/soil from agricultural fields can change groundwater chemistry. The bivariate diagram of NO3 vs. TDS and E.C, Figure 14a,b, records that six samples of alluvium aquifer and two samples of sandstone aquifer, plots along the 1:1 aquiline, show the highest

Discussion
In this study, most of the boreholes were recently drilled (2015-2017), no other boreholes were available. Due to the few observation points, limited previous investigations and few hydrogeological data, using geospatial distributions, GIS and DWQI provide support in groundwater studies. As far as we know, no other study was conducted using the techniques in Fw-Rh and Qa-Qu areas.
The chemical composition and elements concentration of groundwater, are related to the rocks lithology and time residence of the water in the aquifers. To identify the effects of the reaction between the groundwater and the (geological units) aquifer, the bivariate diagrams were applied to explain the chemical changes in ionic concentrations in the host rocks and groundwater. The reaction between water and the surrounding surface/soil from agricultural fields can change groundwater chemistry. The bivariate diagram of NO 3 vs. TDS and E.C, Figure 14a,b, records that six samples of alluvium aquifer and two samples of sandstone aquifer, plots along the 1:1 aquiline, show the highest correlation among TDS/NO 3 . The appearance of NO 3 associated with the fertilizers activities in the agricultural farms [37][38][39]. The elements Na + versus K + Figure 14c, at both sandstone and alluvium aquifers, reflects a linear relationship at (r = 0.99) suggesting the reactions of water with sodium feldspar (Albite) and potassium feldspar (Orthoclase) in equations five and six respectively. The Mg +2 strongly correlated with Ca +2 and SO 4 −2 Figure 14d,e, especially in the basaltic aquifer and some other boreholes in the alluvium locations, show a high response of water with a group of minerals (i.e., Pyroxene, Olivine and Biotite) in equations seven, eight, and nine respectively. For the better understanding of geological units in research areas, the thin sections of rock samples have been generated and studied to identify the main mineral composition for each rock sample, as seen in Figure 15. With this ability, the rock mineral contents have been determined much better.

Conclusions
This study explains the geospatial distribution, adopting statistical methods with GIS to characteristics and mapped the groundwater quality in the different hydrogeological units such as sandstone, alluvium, and basaltic aquifers, which are located in eastern Sudan (the southwestern part of Gedaref State). Forty water boreholes samples from different locations were collected, analyzed and estimated.
Aqua Chem v.2014.2 software has been used for groundwater quality elements analysis, while The main mechanism for the dissolution of rock minerals that releases the element such as: (Ca, Mg, Na, K and HCO 3 ); into the groundwater, have been indicated in the following reactions:

Conclusions
This study explains the geospatial distribution, adopting statistical methods with GIS to characteristics and mapped the groundwater quality in the different hydrogeological units such as sandstone, alluvium, and basaltic aquifers, which are located in eastern Sudan (the southwestern part of Gedaref State). Forty water boreholes samples from different locations were collected, analyzed and estimated.
Aqua Chem v.2014.2 software has been used for groundwater quality elements analysis, while ArcGIS software was chosen for the interpretation and spatial mapping, so that groundwater quality estimation studies have been completed successfully. This study envisions the significance of graphical illustrations, i.e., Piper, Bivariate, Dendrogram, and Durov diagrams plot, to determine variation in hydro-chemical facies and to understand the evolution of hydro-chemical processes in Qa-Qu and Fw-Rh areas.
The hydrogeochemical evaluation outcomes and distribution of groundwater cations (Na + , Ca +2 , K + , Mg +2 ) and anions (HCO 3 − , Cl − , SO 4 −2 , F − ) in both the Qa-Qu and Fw-Rh areas, shows that the groundwater is chemically affected by aquifer lithology. According to the plotting from the Durov diagram, most of the elements of water plotted within the HCO 3 ·Na zone, except some other samples that fell in HCO 3 Cl-Na, SO 4 ·Cl·HCO 3 -Na, or HCO 3 ·Cl-Na Mg types. With the exclusion of a few elements, the quality of groundwater is mostly suitable for drinking purposes and other domestic uses. The groundwater in this project is controlled by sodium and bicarbonate ions, which define the composition of the water type to be Na HCO 3 . According to this investigation, three potential aquifers (sandstone, alluvium, and basalt); have been identified in the research areas. The DWQI was used to determine the groundwater quality and its suitability for drinking purposes. According to this investigation, 20% of groundwater samples represent "excellent water", 50% indicate "good water", 15% represent "poor water", 7.5% shows "very poor water", and 7.5% appear as "unsuitable for drinking". The drinking water quality index that was produced for this study reveals that the northwest and southeast parts of Fw-Rh and the southwest part of Qa-Qu locations has the poorest water quality, which is classified as "unsuitable for drinking".
It should be noted, that the actual variations in spatial interpolations, can considerably diverge from the values predicted by spatial interpolation, it may lead to probable limitations of Kriging especially when data is scarce and unequally distributed. Thus, it is essential to know the number of data locations and the geographical extent of the region containing those data locations. In this case, one of the crucial steps is estimating the variogram model, which is more difficult with a small number of data locations. In this study, the transformations (Log & BoxCox), have been used to make the data normally distributed and satisfy the assumption of equal variability for the data. Several types of semivariogram models were tested in Table 3, for all water quality parameters to achieve more reliable results.