Abstract
The need for quality water in Africa for agriculture cannot be overemphasized amidst the current global water crises. The focus of this study sought to evaluate the quality of groundwater for irrigation purposes while unearthing the emerging challenges in the study area. In total, 202 groundwater samples were collected, and several parameters were tested. The study employed QGIS and multi-criteria decision analysis to examine zones of suitable groundwater quality for agriculture. Findings from the study indicate that the primary water types were Na–HCO3 and Ca–HCO3. The overall accuracy (OV) of the land use land cover (LULC) map using the Random Forest (RF) algorithm was 94.5%. The analysis revealed that SpH influences GpH at p = 0.02891 (p < 0.05) and SOM and SOC influence GNO3 at p = 0.044 (p < 0.05). The overlay analysis spatially classified the groundwater in the study area into three categories of suitability with ariel coverage showing areas of good quality (1,534.34 km2), moderate quality (1,933.35 km2), and poor quality (1,815.21 km2). The results from this study uncovered that 72.33% of the samples were within the desirable limits and it can be concluded that the quality of groundwater in the area is acceptable for irrigation.
HIGHLIGHTS
This paper examines different surficial factors.
This study presents several cases (quality versus LULC, slope, and soil).
The paper examines the suitable areas for irrigation as per the available groundwater quality.
INTRODUCTION
Groundwater is an essential resource, but its availability for domestic, irrigation, and industrial uses has been jeopardized by several factors, the most serious of which is climate variability (Havril et al. 2018). Several factors such as subsurface geochemical processes, precipitation, recharge water, and inland surface water contribute to the occurrence of poor groundwater quality (Chegbeleh et al. 2020). The quality of groundwater has attracted global attention due to the high-quality water demand for domestic and irrigation purposes. Groundwater quality parameters are classified into three types: physicochemical parameters, bacteriological parameters, and trace metals.
In the increasing communities of sub-Saharan Africa (SSA), groundwater usage has increased significantly for both irrigation and household use since the 1980s (Giordano 2006). Despite the huge global interest in developing groundwater for various uses within the sub-Saharan terrain, there remains a growing concern about the immediate quality. Because of this interest, routine quality analysis is required to determine suitability for consumption and irrigation purposes in Ghana, Africa, SSA, and the world at large.
In the last three decades, approximately 14,500 boreholes were drilled throughout the Northern-Voltaian belt of Ghana, with nearly 52% serving the rural population (Abdul-Ganiyu & Prosper 2021). The remaining 48% are unserviceable and can be attributed to poor quality and quantity. However, more boreholes have been drilled recently as the Ghana government has put in efforts to meet the Sustainable Development Goals (SDGs) 6 and 14 by providing clean and potable water to inaccessible communities. The question now is why some rural communities lack access to quality water and why some irrigation schemes have failed. These can be associated with the lack of data on viable areas to explore, lack of awareness, fragmented data, and lack of geospatial research on groundwater quality in the study area. For that reason, this study fundamentally aims to assess the quality of groundwater using geospatial and statistical methods following past studies while developing new methods for future use (Gidey 2018; Moharir et al. 2019; Rawat et al. 2019).
Adimalla et al. (2020) evaluated the quality of groundwater for drinking and irrigation purposes in Central Telangana, India. It was found that 96% of the samples were within the required quality standards and, therefore, were recommended for irrigation. The absence of a groundwater quality suitability map for irrigation was not presented in the study. Chegbeleh et al. (2020) and Gidey (2018) in separate studies also evaluated groundwater quality for mainly irrigation purposes but developed spatial thematic layers and suitability maps complementing the previous studies. This present study sought to evaluate the groundwater quality in the Savannah Region of Ghana, however, this study produced groundwater quality suitability maps for irrigation as well as finding the correlation between soil and groundwater quality which was lacking in the previous studies. This study, however, adopted an integrated approach combining statistical and geospatial methods to examine groundwater quality for irrigation purposes.
DESCRIPTION OF STUDY AREA
MATERIALS AND METHODS
Sample collection and physiochemical analysis
Soil sampling and analysis
From Figure 1, it was observed that soil samples were collected at random grids (farm sites, towns, and industrial areas) to assess the impact of soil properties on groundwater quality. A random sampling grid was produced for sampling using the grid tool in QGIS 3.26.2. Auger/chisel holes were dug at a depth of 25 cm. From Figure 1, the different sampling points are shown across the study area. The soil samples collected for laboratory analysis were air-dried and sieved to pass through a 2-mm sieve. The soil pH was determined using a glass electrode at a soil-to-water ratio of 1:1 (Mclean 2015) at the Soil Laboratory at the University for Development Studies.
Spatial analysis of groundwater quality for irrigation
Several datasets were acquired to assess the impacts of surficial factors on groundwater quality for irrigation. These datasets include slope, geology, soil, and satellite imagery of the study area, except for the satellite imagery all other spatial datasets were acquired from the University for Development Studies Soil Laboratory. The datasets were provided in a vector format but were later rasterized in the GIS environment (QGIS) using the conversion tools.
Land use land cover classification
The first and most important pre-analysis procedures in using Landsat datasets are cloud cover, shadows, and no data removal. However, this study had a dataset with cloud cover ranging from 0 to 3%, therefore, doing this cloud cover removal was deemed unnecessary. The satellite imagery used for the land use land cover (LULC) was Landsat 8–9 (using bands 2, 3, 4, and 5). The QGIS 3.26.3 software program was used for the LULC, the essence of using this program was to produce a LULC map with higher spatial accuracy. The study employed the Random Forest (RF) algorithm which has been reported in recent studies to be higher performing compared to other classifiers like ISODATA and minimum distance (Naghibi et al. 2017; Kpiebaya et al. 2022). The putative land use type was selected using the Garmin 62s GPS device, and the selected coordinates were cross-referenced using Google Earth Pro data.
Geostatistical analysis of groundwater quality parameters
The difference between classical statistics and geostatistics is the assumption of spatial dependency. Several studies have adopted different interpolation methods to show the spatial distribution of groundwater parameters across the globe (Gidey 2018; Rawat et al. 2019). The essence of geospatial analysis is to visualize areas of low to high concentrations which might not be suitable for exploration purposes. Some of these interpolation methods used in previous studies include IDW (Inverse Distance Weight), Kriging, Zoning, and Spline. The smart-map plugin was activated in QGIS and selected groundwater quality parameters (EC, pH, TDS, SAR, MH, RSC, , PS, KR, TH, and Cl−) were interpolated using ordinary kriging. The main benefit of using kriging over alternative interpolations is that an un-sampled point attribute value is the weighted average of known values in the area, and the weights are proportional to the distances between the forecast site and the sampled locations (Zandi et al. 2011; Moharir et al. 2019). The main outputs from the analysis were the range, standard error, root mean square, partial sill, and lag size which are mathematically known and expressed in past and recent studies (Reza et al. 2017). The soil pH was also interpolated using geostatistics; however, the ‘Extract Multi Values to Points’ in the processing toolbox was used to extract the cell values from the soil pH layer and the groundwater pH layer for the statistical analysis.
Statistical analysis
The data obtained were first subjected to descriptive statistical analysis and variability expressed in terms of range, standard deviation, and coefficient of variation. Also, an analysis of variance (ANOVA) was conducted to assess the impact of soil pH on groundwater pH. The essence of these was based on the assumption that the study area had high groundwater pH, however, this was observed in agricultural areas and the relationship between soil pH and groundwater was considered prominent in this study. The groundwater parameters were selected to spatially determine the efficacy of groundwater quality for agriculture EC, pH, TDS, SAR, MH, RSC, , PS, KR, and TH. These parameters were selected with reference to Gidey (2018), Rawat et al. (2018), and Pandey et al. (2020). The selected parameters were exported from the kriged format to a GeoTIFF file showing areas of low and high variations.
Factors . | SAR . | TDS . | EC . | MH . | RSC . | . | pH . | PS . | KR . | Cl− . | NPE (%) . | Weights . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
SAR | 1.00 | 1.07 | 1.00 | 1.07 | 1.16 | 1.16 | 1.13 | 1.07 | 1.00 | 1.00 | 10.62 | 0.11 |
TDS | 0.94 | 1.00 | 1.07 | 1.13 | 1.13 | 1.11 | 1.11 | 1.11 | 1.00 | 1.00 | 10.56 | 0.11 |
EC | 1.00 | 0.94 | 1.00 | 1.13 | 1.13 | 1.13 | 1.07 | 1.13 | 1.00 | 1.00 | 10.51 | 0.11 |
MH | 0.94 | 0.88 | 0.88 | 1.00 | 1.07 | 1.11 | 1.13 | 1.11 | 1.00 | 1.00 | 10.06 | 0.10 |
RSC | 0.86 | 0.88 | 0.88 | 0.94 | 1.00 | 1.11 | 1.11 | 1.11 | 1.00 | 1.00 | 9.83 | 0.10 |
0.86 | 0.90 | 0.88 | 0.90 | 0.90 | 1.00 | 1.11 | 1.13 | 1.00 | 1.00 | 9.65 | 0.10 | |
pH | 0.88 | 0.90 | 0.94 | 0.88 | 0.90 | 0.90 | 1.00 | 1.13 | 1.00 | 1.00 | 9.52 | 0.10 |
PS | 0.94 | 0.90 | 0.88 | 0.90 | 0.90 | 0.88 | 0.88 | 1.00 | 1.00 | 1.00 | 9.28 | 0.09 |
KR | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 9.99 | 0.10 |
Cl− | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 9.99 | 0.10 |
Factors . | SAR . | TDS . | EC . | MH . | RSC . | . | pH . | PS . | KR . | Cl− . | NPE (%) . | Weights . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
SAR | 1.00 | 1.07 | 1.00 | 1.07 | 1.16 | 1.16 | 1.13 | 1.07 | 1.00 | 1.00 | 10.62 | 0.11 |
TDS | 0.94 | 1.00 | 1.07 | 1.13 | 1.13 | 1.11 | 1.11 | 1.11 | 1.00 | 1.00 | 10.56 | 0.11 |
EC | 1.00 | 0.94 | 1.00 | 1.13 | 1.13 | 1.13 | 1.07 | 1.13 | 1.00 | 1.00 | 10.51 | 0.11 |
MH | 0.94 | 0.88 | 0.88 | 1.00 | 1.07 | 1.11 | 1.13 | 1.11 | 1.00 | 1.00 | 10.06 | 0.10 |
RSC | 0.86 | 0.88 | 0.88 | 0.94 | 1.00 | 1.11 | 1.11 | 1.11 | 1.00 | 1.00 | 9.83 | 0.10 |
0.86 | 0.90 | 0.88 | 0.90 | 0.90 | 1.00 | 1.11 | 1.13 | 1.00 | 1.00 | 9.65 | 0.10 | |
pH | 0.88 | 0.90 | 0.94 | 0.88 | 0.90 | 0.90 | 1.00 | 1.13 | 1.00 | 1.00 | 9.52 | 0.10 |
PS | 0.94 | 0.90 | 0.88 | 0.90 | 0.90 | 0.88 | 0.88 | 1.00 | 1.00 | 1.00 | 9.28 | 0.09 |
KR | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 9.99 | 0.10 |
Cl− | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 | 9.99 | 0.10 |
NPE, Normalize Principal Eigenvector.
Here, GQI is the suitability index, Xi is the normalized weight of the ith feature of the thematic layer, Wj is the normalized weight of the jth thematic layer, m is the sum of themes, and n is the sum of classes in a theme. This trend, however, occurs in ArcGIS using the overlay tool, whereas Equation (4) is applicable in QGIS using the raster calculator. As shown in Table 1, SAR had the highest weight of 10.65% while PS had the lowest (9.28%).
RESULTS AND DISCUSSION
Physiochemical analysis
The essence of physiochemical parameters in this study cannot be underplayed because it is the primary determinant for drinking, domestic, agricultural, and industrial uses (Ewida et al. 2020). Presented in this study is a summary of physiochemical results depicting the chemical composition of groundwater in the study area. From Table 2, pH ranged from 5.69 to 8.89, with an average value of 8.06, indicating that the groundwater in the study area is slightly alkaline which agrees with the saline groundwater type in Abu Dhabi, UAE reported by Batarseh et al. (2021).
Quality parameters . | Units . | Min . | Max . | Mean . | STD . | CV . | Acceptable limits . | Number of qualified samples . | Number of qualified samples (%) . |
---|---|---|---|---|---|---|---|---|---|
pH | 5.69 | 8.89 | 8.06 | 0.80 | 0.10 | 6.5–8.5 | 105 | 51.98 | |
TDS | mg/l | 8.73 | 1,367.83 | 370.44 | 250.27 | 0.68 | 1,000 | 198 | 98.02 |
EC | S/m | 2.00 | 245.00 | 67.50 | 38.70 | 0.57 | 1,000 | 175 | 86.63 |
TH | mg/l | 0.49 | 267.50 | 74.17 | 55.00 | 0.74 | 500 | 148 | 73.27 |
mg/l | 0.00 | 108.61 | 3.59 | 9.73 | 2.71 | 10 | 184 | 91.09 | |
Cl− | mg/l | 0.89 | 747.53 | 33.29 | 80.81 | 2.43 | 250 | 197 | 97.52 |
mg/l | 0.09 | 596.90 | 37.44 | 67.91 | 1.81 | 250 | 197 | 97.52 | |
mg/l | 0.00 | 21.00 | 4.39 | 4.80 | 1.09 | 200 | 202 | 100.00 | |
mg/l | 4.27 | 685.03 | 292.07 | 173.13 | 0.59 | 250 | 86 | 42.57 | |
Na+ | mg/l | 1.56 | 434.47 | 110.52 | 88.01 | 0.80 | 200 | 169 | 83.66 |
K+ | mg/l | 1.25 | 59.35 | 8.00 | 7.03 | 0.88 | 30 | 200 | 99.01 |
Ca2+ | mg/l | 0.04 | 102.52 | 15.68 | 14.73 | 0.94 | 200 | 202 | 100.00 |
Mg2+ | mg/l | 0.01 | 27.95 | 6.14 | 6.86 | 1.12 | 70 | 202 | 100.00 |
SAR | mg/l | 0.15 | 42.17 | 8.20 | 8.28 | 1.01 | 10–25 | 64 | 31.68 |
RSC | mg/l | 3.00 | 656.21 | 274.65 | 175.31 | 0.64 | 125–250 | 47 | 23.27 |
PS | mg/l | 1.39 | 749.35 | 38.07 | 81.26 | 2.13 | 4.23–8.23 | 18 | 8.91 |
KR | 0.06 | 101.05 | 11.09 | 16.38 | 1.48 | 1 | 42 | 20.79 | |
MH | % | 0.46 | 60.00 | 24.97 | 13.78 | 0.55 | 50 | 194 | 96.04 |
Quality parameters . | Units . | Min . | Max . | Mean . | STD . | CV . | Acceptable limits . | Number of qualified samples . | Number of qualified samples (%) . |
---|---|---|---|---|---|---|---|---|---|
pH | 5.69 | 8.89 | 8.06 | 0.80 | 0.10 | 6.5–8.5 | 105 | 51.98 | |
TDS | mg/l | 8.73 | 1,367.83 | 370.44 | 250.27 | 0.68 | 1,000 | 198 | 98.02 |
EC | S/m | 2.00 | 245.00 | 67.50 | 38.70 | 0.57 | 1,000 | 175 | 86.63 |
TH | mg/l | 0.49 | 267.50 | 74.17 | 55.00 | 0.74 | 500 | 148 | 73.27 |
mg/l | 0.00 | 108.61 | 3.59 | 9.73 | 2.71 | 10 | 184 | 91.09 | |
Cl− | mg/l | 0.89 | 747.53 | 33.29 | 80.81 | 2.43 | 250 | 197 | 97.52 |
mg/l | 0.09 | 596.90 | 37.44 | 67.91 | 1.81 | 250 | 197 | 97.52 | |
mg/l | 0.00 | 21.00 | 4.39 | 4.80 | 1.09 | 200 | 202 | 100.00 | |
mg/l | 4.27 | 685.03 | 292.07 | 173.13 | 0.59 | 250 | 86 | 42.57 | |
Na+ | mg/l | 1.56 | 434.47 | 110.52 | 88.01 | 0.80 | 200 | 169 | 83.66 |
K+ | mg/l | 1.25 | 59.35 | 8.00 | 7.03 | 0.88 | 30 | 200 | 99.01 |
Ca2+ | mg/l | 0.04 | 102.52 | 15.68 | 14.73 | 0.94 | 200 | 202 | 100.00 |
Mg2+ | mg/l | 0.01 | 27.95 | 6.14 | 6.86 | 1.12 | 70 | 202 | 100.00 |
SAR | mg/l | 0.15 | 42.17 | 8.20 | 8.28 | 1.01 | 10–25 | 64 | 31.68 |
RSC | mg/l | 3.00 | 656.21 | 274.65 | 175.31 | 0.64 | 125–250 | 47 | 23.27 |
PS | mg/l | 1.39 | 749.35 | 38.07 | 81.26 | 2.13 | 4.23–8.23 | 18 | 8.91 |
KR | 0.06 | 101.05 | 11.09 | 16.38 | 1.48 | 1 | 42 | 20.79 | |
MH | % | 0.46 | 60.00 | 24.97 | 13.78 | 0.55 | 50 | 194 | 96.04 |
STD, standard deviation; CV, coefficient of variation.
However, TDS in the study area ranged from 8.73 to 1,367.83 mg/l, with an average of 370.44 mg/l, it was found that 98.02% of the samples were within the acceptable limits (WHO). These results correspond to Ewida et al. (2020) because of similar geology. TH in groundwater samples in this study ranged from 0.49 to 267.50 mg/l. Sajil Kumar et al. (2014) reported some devastating effects of high TH in groundwater including; corrosiveness on metallic pipes and altering groundwater chemical composition. It was noticed that 73.27% of the analyzed samples in this study were within the acceptable WHO standards of TH (Table 2).
Na+ was found to be the dominant cation in the groundwater samples analyzed, ranging from 1.56 to 434.47 mg/l with a mean value of 110.52 mg/l (Table 2). The importance of Na+ in this study is equally important to other quality parameters but it is an essential parameter in computing SAR and KR. It was observed that 86.63% of samples analyzed for EC were within the desirable limits ranging from 2 to 245 mg/l. Continually, was found to be within the range of 4.27 and 685.03 mg/l, with an average of 292.07 mg/l suggesting its adaptability for irrigation purposes. The wide variation in the physical parameters can mainly be attributed to the existing geology and climatic variation. Also, physical parameters are exacerbated by anthropogenic activities.
In the present study, Na+ > Ca2+ > K+ > Mg2+ was the major cation concentration in that order in the groundwater collected. However, the occurrence of Na+ in groundwater indicates the influence of sedimentary rocks such as sandstone on the aquifer (Singh et al. 2012). This is because sandstone contains high levels of salts and hence during the rainy season the dissolvable salts are then infiltrated into the aquifer through percolation. The occurrence of these cations in this study is consistent with studies of Yidana et al. (2012), and Yidana et al. (2013) because of the same study regime and methodology.
From Table 2, Cl− was the major occurring anion (Cl− > > > > ) in the study area. The cation–anion interaction shows two major groundwater types, namely: low mineralized water (Mg2+– and K+–) and high mineralized water (Na+– and Ca2+–) which are consistent with results from the piper trilinear diagram. The essence of unearthing the groundwater type is to have a general perceptive of the influence of hydrogeochemical processes and anthropogenic activities on groundwater in the study area.
The results showed that high mineralized water (Na+– and Ca2+–) can be attributed to the water–rock interactions, specifically weathering of aluminosilicates, dissolution of carbonate minerals, and cation exchange reactions within the Voltaian supergroup of the Bimbilla formation and the Panabako sandstone (Figure 1) (Yidana et al. 2012; Chegbeleh et al. 2020). In summary, 72.33% of the groundwater samples analyzed were within the desired quality requirements (WHO), implying that the groundwater in the study area is moderately suitable for both irrigations.
Impact of LULC on groundwater quality for irrigation
The influence of LULC in groundwater quality assessment cannot be undermined because it is a major determinant of the high concentration of some parameters. However, the recent global climatic issues have exacerbated the impact and led to poor quality. Several studies have reported the effects of LULC on groundwater potential and quality across the globe, indicating that the high levels of some parameters can be linked to anthropogenic factors, climatic change variability, and land use change (Goyette et al. 2019; Yan et al. 2022). From the LULC statistics, shrublands had the largest area of about 46.96% and the least was the rivers which had an area of 0.02%. The codominant LULC classes were settlements, croplands, riparian, dams, open woodlands, and close woodland which had an area of 10.5, 5.68, 1.35, 0.34, 11.44, and 23.7%, respectively. Furthermore, analysis from the post-classification statistics yielded an overall accuracy (OV) of 94.27% which is recognized as a higher accuracy. Munyati (2019) in a study reported the accuracies of different classifiers and found RF to have an accuracy of 96% compared to Support Vector Machine (SVM) which had 93% indicating that RF is better performing.
From Table 3, p = 0.02891 (at p < 0.05) which shows a 2.9% probability of soil pH impacting groundwater quality (pH). Also, Table 4 presented p = 0.044 (at p < 0.05) suggesting a 4.4% chance of soil organic carbon and organic matter influencing the concentration of nitrate in groundwater. The essence of adopting this statistical approach is to confirm findings from the spatial analysis as seen in the LULC map. This probability of groundwater pollution can mainly be attributed to the impact of land use activities in and partly spatio-temporal variation as reported in some studies (Alawi 2023; Shrestha & Shakya 2023).
Source of variation . | SS . | df . | MS . | F . | P-value . | F-crit . |
---|---|---|---|---|---|---|
Between groups | 1.808649 | 1 | 1.808649 | 4.988558 | 0.02891 | 3.986269 |
Within groups | 23.92892 | 66 | 0.362559 | |||
Total | 25.73757 | 67 |
Source of variation . | SS . | df . | MS . | F . | P-value . | F-crit . |
---|---|---|---|---|---|---|
Between groups | 1.808649 | 1 | 1.808649 | 4.988558 | 0.02891 | 3.986269 |
Within groups | 23.92892 | 66 | 0.362559 | |||
Total | 25.73757 | 67 |
SS, sum of squares; df, degree of freedom; MS, means of squares.
Source of variation . | SS . | df . | MS . | F . | P-value . | F-crit . |
---|---|---|---|---|---|---|
Between groups | 43.00026 | 2 | 21.50013 | 3.116201 | 0.048702 | 3.08824 |
Within groups | 683.0475 | 99 | 6.89947 | |||
Total | 726.0478 | 101 |
Source of variation . | SS . | df . | MS . | F . | P-value . | F-crit . |
---|---|---|---|---|---|---|
Between groups | 43.00026 | 2 | 21.50013 | 3.116201 | 0.048702 | 3.08824 |
Within groups | 683.0475 | 99 | 6.89947 | |||
Total | 726.0478 | 101 |
SS, sum of squares; df, degree of freedom; MS, means of squares.
Impact of elevation on groundwater quality
Spatial variability of groundwater quality parameters
Spatial suitability of groundwater quality for irrigation use
The essence of this analysis is to inform irrigators and agriculturists about areas wholesome for exploration for agricultural purposes. Analysis from this study is similar and consistent with studies of Rakotondrabe et al. (2018) in East Cameroon around a mining site, which found groundwater quality favorable for irrigation. Findings from the studies revealed that groundwater quality around the surrounding communities can be explored for irrigation because the largest classes were found to have good quality and the least was found to be very close to the mines. However, studies by Rakotondrabe et al. (2018) conflict results reported by Semy & Romeo (2021) who found mining communities to have poor water quality for irrigation, this can be attributed to the type of mining and processes involved. The main finding from this work showed that more than 3,467.69 km2 of the entire study area have groundwater quality useful for irrigation and the remaining 1,815.21 km2 should be given more attention to reducing the possible effect on irrigation as seen in Figure 5.
Hydro-chemical facies (piper trilinear analysis)
Spatial dependence of groundwater quality parameters (cross-validation)
Geostatistics is an important spatial tool that has been used in recent years to evaluate groundwater quality for different purposes by unearthing the spatial structure of these parameters. The best-fit model used in this study is depicted in the Supplementary data (Figure S1(a)–1(k)). Five different isotropic semivariogram models were used, namely Stable, Circular, Spherical, Gaussian, and K-Bessel (Table 5). The isotropic models were chosen because the spatial interdependence among variables developed in the same way in all orientations. The root mean squared error (RMSE) and root mean squared standardized error (RMSS) were used to validate spatial layers used in the overlay analysis (Clark 2010) study showed that the higher the nugget value (between 0 and 1) the smoother and more accurate the spatial layer. The higher nugget value was 0.795 (circular model) which was used to develop the MH layer, suggesting a smoother layer compared to TH (K-Bessel) with 0.035. This study showed that the Stable, Gaussian, Circular, and Spherical semivariances were the best-fit semivariance. Also, RMSE has reported in several studies that it should be close to 1 to exhibit a higher accuracy (Setianto & Triandini 2015). The presented results showed that all quality parameters were close to 1 as seen in Table 5 indicating higher spatial dependent layers that can be used for further analysis.
Parameter . | Model . | Nugget . | Range . | Partial sill . | Lag SIZE . | Mean . | RMS . | RMSE . | Regression equation Y=mx+c . |
---|---|---|---|---|---|---|---|---|---|
EC | Stable | 0.379 | 0.320 | 0.750 | 0.027 | −0.035 | 29.208 | 1.139 | 0.57x + 27.92 |
TDS | Gaussian | 0.486 | 0.265 | 0.620 | 0.025 | 1.660 | 191.350 | 1.061 | 0.53x + 171.56 |
SAR | Stable | 0.285 | 0.322 | 1.046 | 0.027 | −0.178 | 6.245 | 1.195 | 0.57x + 3.412 |
MH | Circular | 0.795 | 0.286 | 0.272 | 0.024 | −0.405 | 12.829 | 0.978 | 0.17x + 20.56 |
RSC | Circular | 0.417 | 0.356 | 0.847 | 0.030 | −0.347 | 128.800 | 1.011 | 0.57x + 115.83 |
HCO3 | Circular | 0.444 | 0.272 | 0.628 | 0.027 | −2.449 | 130.010 | 0.985 | 0.54x + 130.73 |
pH | Spherical | 0.479 | 0.292 | 0.609 | 0.024 | 0.016 | 0.623 | 0.991 | 0.40x + 4.866 |
PS | Stable | 0.379 | 0.198 | 0.646 | 0.017 | −3.206 | 75.299 | 1.088 | 0.1x + 27.77 |
KR | Spherical | 0.339 | 0.293 | 0.790 | 0.032 | −0.367 | 13.800 | 1.245 | 0.32x + 6.034 |
TH | K-Bessel | 0.035 | 0.002 | 1.183 | 0.0002 | −0.750 | 53.770 | 0.935 | 0.007x + 74.32 |
Cl | Stable | 0.504 | 0.025 | 0.431 | 0.012 | −5.614 | 74.986 | 1.143 | 0.027x + 23.32 |
Parameter . | Model . | Nugget . | Range . | Partial sill . | Lag SIZE . | Mean . | RMS . | RMSE . | Regression equation Y=mx+c . |
---|---|---|---|---|---|---|---|---|---|
EC | Stable | 0.379 | 0.320 | 0.750 | 0.027 | −0.035 | 29.208 | 1.139 | 0.57x + 27.92 |
TDS | Gaussian | 0.486 | 0.265 | 0.620 | 0.025 | 1.660 | 191.350 | 1.061 | 0.53x + 171.56 |
SAR | Stable | 0.285 | 0.322 | 1.046 | 0.027 | −0.178 | 6.245 | 1.195 | 0.57x + 3.412 |
MH | Circular | 0.795 | 0.286 | 0.272 | 0.024 | −0.405 | 12.829 | 0.978 | 0.17x + 20.56 |
RSC | Circular | 0.417 | 0.356 | 0.847 | 0.030 | −0.347 | 128.800 | 1.011 | 0.57x + 115.83 |
HCO3 | Circular | 0.444 | 0.272 | 0.628 | 0.027 | −2.449 | 130.010 | 0.985 | 0.54x + 130.73 |
pH | Spherical | 0.479 | 0.292 | 0.609 | 0.024 | 0.016 | 0.623 | 0.991 | 0.40x + 4.866 |
PS | Stable | 0.379 | 0.198 | 0.646 | 0.017 | −3.206 | 75.299 | 1.088 | 0.1x + 27.77 |
KR | Spherical | 0.339 | 0.293 | 0.790 | 0.032 | −0.367 | 13.800 | 1.245 | 0.32x + 6.034 |
TH | K-Bessel | 0.035 | 0.002 | 1.183 | 0.0002 | −0.750 | 53.770 | 0.935 | 0.007x + 74.32 |
Cl | Stable | 0.504 | 0.025 | 0.431 | 0.012 | −5.614 | 74.986 | 1.143 | 0.027x + 23.32 |
CONCLUSION
Groundwater has special economic significance globally and locally because it is the largest hydro-structure with freshwater. Groundwater plays an important role in the supply of freshwater for domestic and agricultural use in rural communities around the world. Based on the analyzed samples, it can be concluded that the range of physiochemical parameters was within the acceptable limits for irrigation and therefore can be adopted for these purposes. The output from the overlay analysis showed that regions of good groundwater quality for irrigation had a total area of 1,534.34 km2. Regions of moderate groundwater quality for irrigation had an area of 1,933.35 km2 while zones of poor quality were found to be 1,815.21 km2. Based on these findings, it is recommended that:
Groundwater in the study area showed a degree of viability for agricultural purposes based on the analyzed samples. However, it is essential to subject future water samples to thorough quality analysis before use.
Further research should be done to reveal the extent of the impact of LULC and some surficial factors on the groundwater. This will aid policymakers to develop strategies to combat and reveal groundwater pollution.
The irrigation water requirements for crops should be strictly adhered to in order to avoid over-irrigation and subjecting crops to excess leaching of available nutrients. Also, agroforestry methods should be practiced within farmland to reduce the effect of aridity on crop production.
Groundwater exploration is an expensive commodity in SSA, hence thorough geospatial modeling should be done in examining suitable areas for irrigation. Some of these models include artificial intelligence, deep learning, fuzzy overlay, machine learning, and many more.
ACKNOWLEDGEMENTS
Much gratitude to all of our colleagues who advised us on how to modify and remodify this research so that it could be published today. We would like to express our appreciation to the anonymous editor and reviewers for their invaluable contributions to this paper. The opinions expressed in this paper do not necessarily reflect the views of the University for Development Studies (UDS).
DATA AVAILABILITY STATEMENT
Data cannot be made publicly available; readers should contact the corresponding author for details.
CONFLICT OF INTEREST
The authors declare there is no conflict.