Introduction

With more than half of the global population residing and making a living in towns and cities, the sustainability of water supplies is a major concern. This is more pronounced in coastal areas, where net migration is much higher than the national averages (Wong et al. 2014). Therefore, groundwater has emerged as a vital water resource for meeting the domestic, industrial, agricultural, and environmental demands owing to its relatively consistent yield of good-quality water. In many coastal regions, already scarce fresh groundwater resources are under an unprecedented amount of stress from different drivers and pressures, including urbanization and land-use change (Nlend et al. 2018; Idowu et al. 2020; Ajibade et al. 2021), population growth (Okello et al. 2015), increased domestic and commercial demand, and climate change (Aladejana et al. 2020; Stein et al. 2020). Fresh groundwater resources are dynamic in nature and are affected by these factors; hence, monitoring and conserving aquifers are crucial for sustainable groundwater management.

In coastal settlements, demographic change has intensified the rate of groundwater abstraction and dependence on groundwater, thereby resulting in groundwater vulnerability to depletion and pollution. When groundwater levels in aquifers deplete faster than they can recharge, it results in seawater intrusion (SWI). The direct negative impact of aggravated SWI is that it diminishes the quality and the availability of the water for potable use and other needs. Several studies in different parts of the globe have shown that increased levels of seawater intrusion have far-reaching adverse impacts on humans and the environment. It reduces groundwater quality below the quality thresholds for drinking water (Naseem et al. 2018; Mahammad and Islam 2021), negatively impacts irrigation practices and agriculture (Heydarirad et al. 2019; Sarkar et al. 2021), proves detrimental to human health and living conditions (Siyal 2018; Shammi et al. 2019), and causes adverse effects on coastal ecology and freshwater marsh (Herbert et al. 2018; Siyal 2018) and a host of other undesirable effects.

The theoretical background on seawater intrusion can be found in many studies (Verruijt 1968; Darnault and Godinez 2008; Vengadesan and Lakshmanan 2019). Sea level rise has been identified as one of the factors impacting sea water intrusion (Idowu and Home 2015; Ketabchi et al. 2016; Vu et al. 2018). More exhaustive details on the anthropogenic-driven factors including land-use change and climatic factors impacting the seawater intrusion (SWI) are provided in many published texts (Bear et al. 1999; Barlow 2003; Kogo et al. 2019; Kumar et al. 2019). The multifaceted nature and wide range of factors affecting SWI and the quality of a coastal aquifer make it a dynamic problem. For instance, hydrogeological factors are relatively stable over decades and sometimes centuries, but land use land cover changes are not.

Fresh groundwater quality assessments have often taken a multi-approach where two or more techniques are combined in one study (Krishna-Kumar et al. 2015; Bouderbala et al. 2016; Akoteyon et al. 2018). The focus of groundwater quality assessments depends on the objectives, e.g., irrigation purpose, municipal use, and most commonly drinking/potable water suitability. One of the widely used techniques applied for drinking water purposes is the Water Quality Index (WQI) where the water quality in the aquifer of interest is assessed against known water quality standards like the World Health Organization (WHO) guidelines and national standards (Ketata-Rokbani et al. 2011). Several studies have applied the WQI either singly or in combination with other analytical techniques (Alastal et al. 2015; Krishna-Kumar et al. 2015; Bouderbala et al. 2016; Akoteyon et al. 2018; Hamlat and Guidoum 2018).

Groundwater vulnerability assessments take on three broad approaches—statistical, process-based, and index/overlay methods (Al-Abadi 2017). In the case of index/overlay methods, factors known to play influential roles on pollutant movement are mapped, assigned weights, and importance ratings based on significance to the pollutant movement. Typical examples of these factors include geology, hydraulic conductivity, slope, and soil type. These factors are then overlaid and a resultant vulnerability index map that aggregates the impact of all the factors is obtained. The groundwater vulnerability map obtained is usually qualitative and specific to the study area. Groundwater vulnerability is described as either intrinsic or specific, where intrinsic vulnerability, is based on the physical attributes of environmental factors while specific vulnerability factors are concerned with the transport behavior of the contaminant or group of contaminants through the porous medium (Witkowski et al. 2014). Hence, most index and overlay methods measure intrinsic vulnerabilities.

Some existing overlay methods include DRASTIC (Aller et al. 1987), SINTACS (Civita 1994), GOD (Foster 1987), AVI (Stempvoort et al. 1993), EPIK (Doerfliger and Zwahlen 1997) the ISIS (Civita and Regibus 1995), and the GALDIT index (Chachadi and Lobo-Ferreira 2005; Lobo-Ferreira et al. 2005). Comparisons of the performance of these methods have been done in many studies as in the case of Luoma et al. (2017), where GALDIT, AVI, and SINTACS were compared in a single study, or Bouderbala et al. (2016) where AVI, GALDIT, and WQI were utilized for assessing the quality and the vulnerability of the coastal aquifer in Tipaza, North Algeria, and a comparison between GALDIT and DRASTIC indices in Kapas Island, Malaysia (Kura et al. 2015). Commonly used indices are DRASTIC, for determining the vulnerability of groundwater to anthropogenic pollution from surface sources, and SINTACS—a modified DRASTIC method—with more weight-strings and additional factors associated with human activities and watercourses. However, the application of DRASTIC and SINTACS to coastal aquifer studies is limited because they do not account for seawater intrusion contamination. Hence, the GALDIT index which utilizes a similar system of rating and weightages to DRASTIC, but focuses more on detecting the vulnerability of a coastal aquifer to SWI was developed (Chachadi and Lobo-Ferreira 2005; Lobo-Ferreira et al. 2005).

Alongside the GALDIT overlay index, analytical approaches focusing on seawater intrusion have also been developed. Typical examples include the seawater mixing ratios (Lee et al. 2015); Stuyfzand classification based on total hardness (Stuyfzand 1993); and a robust improvement on seawater mixing ratio developed by Tomaszkiewicz et al. (2014), known as the seawater intrusion groundwater quality index GQISWI. GQISWI helps generate a representative seawater intrusion index not only based on seawater mixing ratio but also piper analytical system. With the impressive advances in GIS and geospatial analysis, the point values obtained from these analyses can be spatially shown as maps using spatial interpolation techniques.

Many of the groundwater quality and vulnerability studies have either leaned more towards geospatial analyses/overlay indices than statistical/analytical analyses (Eguaroje et al. 2015; Kura et al. 2015; Bouderbala et al. 2016; Ezekiel et al. 2016; Luoma et al. 2017; Seenipandi et al. 2019; Wei et al. 2021) or vice versa (Bouderbala 2017; Idowu et al. 2017; Aminiyan and Aminiyan 2020; Hajji et al. 2020). Some studies managed to create an even balance between geospatial and statistical analyses but only covered narrower scopes (Ayed et al. 2018; Vaiphei et al. 2020). There is also a growing interest in the application of machine learning techniques to water quality and seawater intrusion studies (Etsias et al. 2020; Bordbar et al. 2021; El-Bilali et al. 2021). Although the application of machine learning algorithms like artificial neural network (ANN) and support vector machine (SVM) in coastal groundwater is currently in its fledgling stage, this research direction is likely to expand in the nearest future.

Coastal groundwater pollution studies require an integrated approach due to the multi-faceted nature of the environmental and anthropogenic-driven processes impacting the quality and pollution status of the groundwater. The application of a single approach to the study of groundwater geochemistry tends to produce unreliable or skewed results (Werner et al. 2013). Therefore, the main aim of this study was to integrate overlay index and analytical and geospatial techniques—three different approaches to coastal groundwater studies—and use the aggregated results for creating a holistic understanding of the water quality and SWI status of a coastal aquifer. The water quality focus is on the suitability of the groundwater for drinking water purposes based on established standards. Furthermore, little emphasis has hitherto been placed on the role of efficient geodatabases in creating detailed and balanced spatio-statistical analyses of coastal aquifers. Hence, the secondary objective of this study was to provide a transferable template for the creation and management of geodatabases used in the study of coastal aquifers. In this study, the GALDIT overlay index was combined with GQISWI, WQI, the Stuyfzand classification analytical approaches, and geospatial analyses for assessing the effect of seawater intrusion and water quality extent in the coastal aquifer of Mombasa North Coast. The combination of these approaches can be an effective and consistent way of assessing the amount and extent of seawater intrusion, and the water quality condition of the aquifer in the bid to meet water quality standards. The methodologies introduced in this study are not location specific and therefore applicable to other areas for short and long-term monitoring of coastal aquifers. The findings of the study provide insights to environmental managers and policy makers for sustainable aquifer management practices like protecting susceptible areas and implementing adaptation plans for coastal aquifers.

Study area

The study area is in the northern part of Mombasa, Kenya, covering an area of 74.2 km2 and lying between latitudes 3° 95″ and 4° 07″ and longitudes 39° 68″ and 39° 72″ south of the equator and east of the Greenwich meridian, respectively. It is bounded by the Indian Ocean on the East, creeks on the northern and southern ends, and elevated hills on the west (Fig. 1). A 2019 national census shows that it has a combined population of 508,507 encompassing two administrative sub-counties—Kisauni and Nyali (KNBS 2019). This is a 34% population increase in the last decade where the combined population in 2009 was 380,055 (KNBS 2010). The climate of the study area is largely tropical and characterized by high humidity and warm temperatures. The difference between the day and night temperatures could range from 6 to 8 °C with temperatures reaching as high as 30 °C from January to March (the warmer months) while night temperatures hover around 20 °C throughout the year (Kenya Coast 2011). Monsoons and the passage of the biannual intertropical convergence zone (ITCZ) also impact the climate of the area with the north-eastern and south-eastern monsoons experienced from January to March and June to October, respectively. Most of the rainfall is experienced between these two monsoon winds with annual precipitation averaging 1072.7 mm (Climatemps 2017).

Fig. 1
figure 1

Study area

Full size image

The geological formation is of the Pleistocene age where the coastal belt was formed as a result of sea-level fluctuations and is mainly composed of sedimentary deposits, corals, coral breccia, and wind-blown sands (Caswell 2007). The lithology is heterogeneous, comprising sandstone, limestone, and shale while the geomorphology is characterized by sandy beaches, creeks, rock-strewn shores, coral reefs, and muddy tidal flats (Abuodha 2004). The topography is relatively flat with low elevations mostly below 42 m above mean sea level (MSL), except for the elevated Nguu Tatu hills located on the western end of the study area (Idowu et al. 2017). The study area’s aquifer is unconfined and is part of the 15,000 km2 transboundary Karoo sedimentary aquifer located between the southern and northern coasts of Kenya and Tanzania, respectively (Igrac 2017). The water table in the study area is relatively shallow but varies widely across the study area (Idowu 2017). The major sources of piped freshwater include the Marere and Mzima springs, Tiwi boreholes, and Baricho wellfields all of which are located outside the administrative boundaries of the study area. A recent study highlights that piped systems only supply a quarter of the city’s daily water needs and the rest are being supplied by groundwater from private boreholes and wells (Kithiia and Majambo 2020). This implies that groundwater is the principal source of freshwater in the city and the dependence on groundwater is rapidly increasing due to the exponential population increase. The impact of this continued dependence on groundwater has largely been in the form of increased pollution. An extensive review of the groundwater of the study area describes the recent pollution-related studies and highlights the presence of high groundwater salinity which varies seasonally, and is usually more severe in the drier months (Idowu and Lasisi 2020).

Methodology

Data description

Geological information, water quality parameters from the field, laboratory test results, and geospatial data were obtained in achieving the objectives of the study. The geological data were obtained from the Mines and Geology department of the Ministry of Environment and Natural Resources of the study area (Caswell 2007). The hydrochemical data is based on extensive post-monsoon groundwater quality data. Post-monsoon data was considered because groundwater recharge is relatively lower and abstraction rates are higher due to the absence of rainfall and rainwater. The data include the groundwater heads above mean sea levels determined by subtracting static groundwater level measurements obtained using water level meters from the digital elevation model (DEM) of the study area. Other parameters include in situ measurements of the electrical conductivity (EC), total dissolved solids (TDS), and pH all measured using a portable Hanna HI 99,300 conductivity meter. Subsequently, the major cations and anions were analyzed in the laboratory from the samples obtained in the boreholes and wells. The concentrations of cations Na+ and K+ were determined using the flame emission spectroscopy while Mg2+ and Ca2+ concentrations were determined using the flame atomic absorption spectroscopy. The anions HCO3− and SO42− concentrations were determined using acidimetric neutralization, and turbidimetric method using a spectrophotometer, respectively, while and Cl− concentrations were determined through the colorimetric determination of chlorine using mercuric thiocyanate, ferric ion, and spectrophotometer.

The importance of an orderly geodatabase for scientific data analysis is rarely stressed, yet it considerably improves the performance and ease of data management. In ArcGIS, creating efficient file geodatabases has been observed to improve versatility, and re-usability of datasets, ease of migration, and editing and also enables customizable storage configuration, easy updates to spatial indices, and data compression (Childs 2009). Hence, in this study, a central geodatabase was created comprising the field data such as the groundwater levels, electrical conductivity, and total dissolved solids values, laboratory data comprising the analyzed cations and anions, and the secondary data obtained from a wide variety of sources using the ArcGIS 10.2 version (Fig. 2). The geodatabase stores the shapefiles containing the spatial attributes of the sampled boreholes and wells such as location, water level measurements, DEM values, and the in situ and laboratory-tested water quality information shown in Fig. 2. The geodatabase was used for the spatial and water quality analyses that are detailed in the remaining parts of this section. Subsequently, smaller sub-databases were created for each index for ease of carrying out the spatial analyses (Fig. 2).

Fig. 2
figure 2

Schematic representation for the file geodatabase creation workflow

Full size image

The geospatial analysis of the results of each analytical technique was explored using spatial interpolation techniques. The techniques were used for creating raster surface maps from the point data resulting from the analytical techniques. The two techniques used are the Inverse distance weighted (IDW) interpolation and kriging interpolation technique. The IDW technique is a deterministic form of interpolation where data from known scattered points are used to estimate the data for the unknown points within the data frame using a weighted average and local neighborhood (ESRI 2019). Kriging, a spatial interpolation method based on geostatistics, is an exact interpolator where interpolated values are the best linear unbiased predictors (Karami et al. 2018; CPH 2021). More detailed explanations of these techniques can be found in the literature (Mitas and Mitasova 1999; Karami et al. 2018; ESRI 2019). In this study, it was observed that the IDW method produced better results for the point data with a relatively smaller range of standard deviation values while the kriging method was more effective for the point data with relatively wide ranges of standard deviation value. The cartographic capacity of the GIS was fully employed in creating the final maps from the raster surfaces created thereby helping to create comprehensive, visual and spatial impressions of the results. The application of spatial interpolation techniques for groundwater characterization has also been adopted in previous studies (Ketata-Rokbani et al. 2011; Trabelsi et al. 2016).

GALDIT overlay index

The GALDIT overlay index, based on a numerical ranking procedure is an open-ended model specifically developed to assess the vulnerability of coastal aquifers to seawater intrusion. It was initially developed and improved in the early 2000s (Chachadi and Lobo-Ferreira 2001a, b, 2005; Lobo-Ferreira et al. 2005). The method focuses on the use of hydrological, hydrogeological, and geomorphological characteristics of an aquifer for assessing its intrinsic vulnerability to seawater intrusion. The method is a mapping approach that involves developing comparable maps of six measurable parameters identified as most critical in determining the vulnerability of a coastal aquifer to saline pollution. The ease of data collection and the simplicity of the approach have led to its wide application in many studies around the globe (Sophiya and Syed 2013; Bouderbala et al. 2016; Ezekiel et al. 2016; Chang et al. 2019; Amarni et al. 2020). The GALDIT acronym is derived from the first letters of the six main factors determining the impact of SWI on a coastal aquifer—“Groundwater occurrence, Aquifer hydraulic conductivity, Level of groundwater above MSL, Distance from shore, Impact of the existing status of SWI, and the Thickness of aquifer.” Further details on the conceptual and theoretical frameworks of the technique can be found in the literature (Chachadi and Lobo-Ferreira 2005; Lobo-Ferreira et al. 2005). Beyond the conceptual framework, several studies have also modified or optimized the technique to suit the specific study area or for better performance (Bordbar et al. 2019; Barzegar et al. 2021; Wei et al. 2021). In this study, the GALDIT factors were determined based on the theoretical frameworks (Chachadi and Lobo-Ferreira 2005; Lobo-Ferreira et al. 2005), the location-specific observations, and insights from previous related studies on the study area (Munga et al. 2006; Ezekiel et al. 2016). The factors and how they are determined for the study area are briefly described below.

Groundwater occurrence (G)

The groundwater occurrence describes the aquifer type—confined, unconfined, leaky-confined, or possessing geological boundaries. Confined aquifers are considered the most vulnerable due to the inherently high pressure in the water and relatively high cones of depression created during pumping. However, the characteristic lower pressures in unconfined aquifers also imply that the hydrodynamic resistance of the freshwater to SWI will be lower, hence making it more vulnerable (Sundaram et al. 2008). The parameter G was obtained from the geological reports and maps of the study area (Caswell 1954a, b, 2007). The maps were georeferenced, and the geology of the study area was clipped and processed from the georeferenced maps. Subsequently, the raster generated was stored in the sub-database for GALDIT indices (Figs. 2 and 3).

Fig. 3
figure 3

a–f GALDIT parameter maps for the coastal aquifer of Mombasa North coast, Kenya

Full size image

Aquifer hydraulic conductivity (A)

The hydraulic conductivity “k” of the aquifer measures the rate at which water moves by gravity through the saturated zone within the aquifer. High K values imply a greater flow rate per unit time, leading to wider cones of depression during abstraction and thereby, increasing the likelihood of SWI (Sophiya and Syed 2013). Parameter A for the study area was obtained from the literature (Munga et al. 2006; Ezekiel et al. 2016).

Level of groundwater above mean sea level (L)

This is one of the most significant factors influencing the SWI in a coastal aquifer. The height of the groundwater above MSL provides a countering freshwater hydraulic pressure against seawater. Hence, the lower the groundwater level, the lower the hydraulic pressure and the higher the risk of intrusion. This phenomenon is illustrated by the Ghyben-Herzberg equation (Verruijt 1968). Parameter L for this study was determined by setting the static water levels measured against the study area’s Digital Elevation Model (DEM) and determining the difference.

Distance from shore (D)

This is the distance inland in the direction perpendicular to the seawater body. Areas of higher proximities to the shore or creeks are more susceptible to SWI. This parameter was determined from the shapefile map of the study area using the NEAR tool of the spatial analyst on ArcGIS 10.2.

Impact of the existing status of SWI (I)

Several indices indicative of seawater intrusion are used for determining this parameter and are highlighted in Klassen et al. (2014), the most widely used being Simpson’s ratio or Revelle’s coefficient. The coefficient is a ratio of chloride ions to bicarbonate ions contained in a groundwater sample (Sudaryanto and Naily 2018). Hence, in this study, the “I” factor was determined for each sampling point by computing the ratios of chloride and bicarbonate ions.

Thickness of the aquifer (T)

This is the saturated thickness of the unconfined aquifer which also influences the extent of the impact of SWI. A larger thickness makes the aquifer more prone to SWI (Chang et al. 2019). As with parameter G, the thickness of the aquifer was obtained from the geological reports and maps of the study area (Caswell 1954a, b, 2007). The information shows that the unconfined aquifer is heterogeneous in nature comprising sandstone, limestone, and shales extending up to 100 m below the surface (Caswell 2007).

The overall GALDIT index vulnerability map was based on decision criteria where each GALDIT parameter was assigned weightages based on Chachadi and Lobo-Ferreira (2005). The ranges of values of the importance ratings were determined based on the uniqueness of the study area as shown in Table 1. The numerical rankings range from 0 to 10, where 2.5, 5.0, 7.5, and 10 represent very low, low, medium, and high vulnerabilities, respectively. The decision criteria are represented by Eq. 1 as introduced by Chachadi and Lobo-Ferreira (2005).

$$\text{GALDIT index}= {~}^{\sum\nolimits_{i=1}^{6}({W}_{i}{R}_{i})}\!\left/\!\!{~}_{\sum\nolimits_{i=1}^{6}{W}_{i}}\right.$$
(1)

where Wi and Ri are the weightages and the importance ratings of the ith GALDIT parameters, respectively. Based on Eq. 1, the final vulnerability map is scaled from 0 to 10 where < 2.5, 2.5 – 5.0, 5.0 – 7.5, and 7.5 – 10 represent very low, low, medium, and high vulnerabilities, respectively (Lobo-Ferreira et al. 2005).

Table 1 Summary of the GALDIT parameters, weightages, rates, and importance ratings
Full size table

Maps based on the importance ratings of these six parameters were created on ArcGIS 10.2. Subsequently, the six maps were overlayed according to the weightages, using the weighted sum tool in the spatial analyst toolbox to obtain the aggregated GALDIT vulnerability map.

Seawater intrusion groundwater quality index

One of the most widely used analytical methods for detecting and assessing the impact of seawater on coastal groundwater is the fresh/seawater mixing equation (Lee et al. 2015; Idowu et al. 2017; Alfarrah and Walraevens 2018). However, it has its limitations and a major one is that it only considers chloride ions as an indicator for SWI. SWI is a dynamic process that is not limited to the presence of chloride ions alone. Hence, a study by Tomaszkiewicz et al. (2014) attempted to address this limitation by incorporating other ions in the analytical process. This was achieved by improving on the fraction of seawater (fsea) equation also known as the seawater mixing ratio (Appelo and Postma 2005; Lee et al. 2015) and further developing an analytical framework with equations and for assessing the impact of SWI based on four cations—Ca2+, Mg2+, Na+, K+, and two anions—HCO−3, Cl− generally used to create piper plots (Tomaszkiewicz et al. 2014). This approach, known as SWI groundwater quality index (GQISWI), is expressed by Eqs. 2 to 6. The parameters were converted from mg/L to meq/L using the ASCE SI unit conversion (ASCE 2012).

$${f}_{\text{sea}}= \frac{{m}_{cl(\text{sample})}-{m}_{cl(\text{freshwater})}}{{m}_{cl(\text{seawater})}-{m}_{cl(\text{freshwater})}}$$
(2)

fsea is also referred to as the seawater mixing ratio (SMR)

$${GQI}_{{f}_{\text{sea}}}=\left(1-{f}_{\text{sea}} \right) \times 100$$
(3)
$${GQI}_{\text{piper (mix)}}=\left(\frac{{Ca}^{2+}+{Mg}^{2+}}{\text{Total Cations}}+\frac{{HCO}_3^-}{\text{Total Anions}}\right)\times50\;(in\;meq/l)$$
(4)
$${GQI}_{\text{piper (dom)}}=\left(\frac{{Na}^++K^+}{\text{Total Cations}}+\frac{{HCO}_3^-}{\text{Total Anions}}\right)\times50\;(in\;meq/l)$$
(5)
$${GQI}_{\text{SWI}}=\frac{{GQI}_{\text{piper (mix)}}+{GQI}_{{f}_{\text{sea}}}}{2}$$
(6)

A global analysis of different water types by Tomaszkiewicz et al. (2014) in relation to GQISWI shows the typical range of values for freshwater, mixed, saline, and seawater (Table 2).

Table 2 GQISWI ranges (Tomaszkiewicz et al. 2014)
Full size table

In this study, the groundwater is classified into different water types based on the computed GQISWI. Furthermore, the GQIpiper(mix) and GQIpiper(dom) were subjected to a hierarchical analytical framework (Tomaszkiewicz et al. 2014). The framework identifies the water type according to the piper diagram-based domains categorized as I, II, III, IV, V, and VI denoting Ca-HCO3, Na-Cl, mixed Ca-Na-HCO3, mixed Ca–Mg–Cl, Ca–Cl, and Na-HCO3 water types, respectively (Sarath Prasanth et al. 2012; Mtoni 2013). Domain I waters generally connote freshwater while domain II waters are impacted by salinity which may be due to SWI.

Total hardness

Total hardness (TH) is one of the predominant parameters of interest in assessments related to drinking water quality. It is the measure of the amount of calcium and magnesium ions as components of their respective salts dissolved in water. The effects of the total hardness of the water on human health, and phenomena such as taste, corrosion and scaling, desalination, reuse, and other considerations are well detailed in the WHO’s report on water hardness (WHO 2011). The total hardness of GW can also provide some insights into the nature of the water. The extensive work by Stuyfzand (1986, 1993) produced a classification scheme for water based on the total hardness calculated in meq/L using Eq. 7:

$$TH=5\times\left({Ca}^{2+}+{Mg}^{2+}\right)in\;meq/L$$
(7)

The total hardness values in meq/L obtained were converted into mmol/L. The water type classification according to the total hardness in mmol/L is presented in Table 3.

Table 3 Water-types based on the total hardness (Stuyfzand 1993)
Full size table

Water Quality Index

There is a need to understand the water quality on a spatial scale for effective management of groundwater. In many locations, groundwater is used conjunctively with other water sources for serving multiple needs like drinking, domestic, agricultural, and recreational use. Global guidelines, e.g., the WHO guidelines for drinking water and national standards and Kenya drinking water standards, provide quality thresholds not to be exceeded. WQI is a viable management tool for determining the suitability of a water type for different uses and when combined with geospatial techniques, WQI maps can be generated. WQI maps help to visualize locations of the most and least suitable areas for the criteria being observed e.g. drinking purposes based on its mineral contents (Bouderbala et al. 2016). The information that WQI maps provide can easily be understood by researchers and policymakers alike. This can help guide decision-making and contribute to creating a better understanding of the relationships between groundwater quality and other environmental parameters like groundwater depth and land use land cover (Rizwan and Gurdeep 2010). The concept of WQI is based on comparing the relevant water quality parameters in a location with regulatory standards/guidelines and aggregating all the water quality parameters considered to obtain single indices which represent the overall water quality status (Alastal et al. 2015). The single water quality indices were spatially interpolated to obtain the WQI maps. The maps express the water quality information in simple terms such as excellent, good, or poor water quality (Table 4).

Table 4 Classification of the WQI scores (Sahu and Sikdar 2008; Ketata-Rokbani et al. 2011)
Full size table

The procedure for computing the WQI involves three main steps (Rizwan and Gurdeep 2010; Ketata-Rokbani et al. 2011; Alastal et al. 2015). The first step involves assigning weights to each of the water quality parameters being considered based on their impacts, e.g., on health when drinking water standards are being considered.

In the second step, the relative weight (Wi) of each parameter is computed using Eq. 8:

$$Wi= \frac{wi}{\sum_{i=1}^{n}wi};$$
(8)

where (wi) is the weight of each parameter, (n) is the number of parameters, and (Wi) is the relative weight.

The third step entails calculating a quality rating scale (qi) for each parameter using Eq. 9:

$$qi= \frac{Ci}{Si} \times 100;$$
(9)

where “(qi) is the quality ranking, (Ci) is the concentration of each chemical parameter in each water sample in mg/l, and (Si) is the WHO standard for each chemical parameter in mg/L.” Finally, the WQI is determined for each chemical parameter using Eq. 10:

$$WQI= \sum\nolimits_{i=1}^{n}Wi\times qi;$$
(10)

where (qi) is the rating based on concentration if (ith) parameter and (n) is the number of parameters. Computed WQI values are usually categorized into five (Table 5): excellent, good, poor, very poor, and unsuitable for human consumption (Rizwan and Gurdeep 2010; Ketata-Rokbani et al. 2011; Alastal et al. 2015).

Table 5 The thresholds, weights, and relative weights of each parameter
Full size table

In this study, the WQI was assessed based on the World Health Organization (WHO) guidelines for drinking water (WHO 2017) and the Kenya Standards for potable water (KEBS 2015). The Kenya standard defines potable water as one “that is safe and suitable for human consumption” (KEBS 2015). The quality requirements set by the two regulatory bodies for the relevant parameters and their perceived weights based on the significance of their impacts on public health are presented in Table 5.

The choice of the parameters is informed by factors like the importance of the parameter as stipulated in the guidelines, the purpose of the index (human health and suitability for human consumption), and data availability. The assigned weights on each parameter connote the parameter’s significance to the overall quality of water on a scale of 1 to 5. The choice of the weights (wi) as expressed in Table 5 are largely based on the predominant weights previously used in earlier studies (Ketata-Rokbani et al. 2011; Alastal et al. 2015; Bouderbala et al. 2016; Bouderbala 2017; Chaurasia et al. 2018) and the implications of the parameters to human health as outlined in the WHO guidelines.

Results and discussion

GIS and spatial analysis have played an integral role in environmental assessment and their evolution over the years are well detailed by Goodchild and Haining (2003). The focus of this study is on the application of the GALDIT overlay index, GQISWI, total hardness, and WQI for an integrated assessment of the groundwater quality and vulnerability to seawater intrusion of a coastal aquifer. Besides the spatial analysis tools used for the GALDIT index–weighted overlays, spatial interpolation techniques were extensively used for creating raster surfaces from the point data resulting from the water quality analyses.

The statistical analysis of the data (Table 6) indicates a wide variation in the groundwater geochemistry of the coastal aquifer. Table 6 shows that the chemical parameters in the groundwater vary widely as indicated by the standard deviation values where the EC, and TDS values are 2410 µS/cm, and 1205 mg/L, respectively. Wide variations are also observed in Na, Cl, K, and SO4 values. The high standard deviation values imply that the quality and characteristics of the groundwater are heterogeneous and widely different water qualities and vulnerabilities can be experienced from one part of the aquifer to another.

Table 6 Statistical summary of the parameters and major chemical constituents in the coastal aquifer of Mombasa North coast
Full size table

These wide-ranging hydrochemical values are better interpreted using multiple techniques, which help to characterize the groundwater holistically. The rest of this section highlights and discusses the results of the GALDIT Index analysis, GQISWI, total hardness, and WQI in relation to the water quality and vulnerability to seawater intrusion.

The GALDIT index vulnerability map

The results of the GALDIT overlay index for the Mombasa North coast covering an area of 74.2 km2 show varying vulnerabilities to seawater intrusion. The parameters G and T which were obtained from the geological reports and maps of the study area (Caswell 1954a, b, 2007) indicate that the aquifer type is unconfined with varying thickness up to 100 m in some locations, and therefore, the G parameter was assigned an importance rating of 7.5 while the T parameter was assigned the maximum rating of 10. The broad importance rating of parameters G and T do not vary across the aquifer (Fig. 3a, f); however, they contribute to the weightages and overall vulnerabilities. Removing these parameters from the analysis will lead to errors where disproportionate weightages are assigned to other parameters. For instance, a recent study on the appraisal of the GALDIT index highlighted the sensitivity of the parameters to seawater intrusion, and the two parameters accounted for nearly 20% of the effective weightages (Seenipandi et al. 2019). Hence, the two parameters were included in the overlay analysis.

The hydraulic conductivity of the aquifer (parameter A) was observed to range from < 4 m to 12 m/day (Munga et al. 2006), and therefore, the ratings ranged from 2.5 to 5. It was observed that the static water levels and invariably the height of the groundwater above sea levels varied with seasons in the study area as seen from the dataset. The post-monsoon groundwater height above the mean-sea level ranged widely from − 0.84 to 33.08 m. These values were used to obtain the map for parameter L with importance ratings ranging from 2.5 to 10, depending on the location within the study area. For parameter D (distance to the shore), areas within 1000 m to the seawater body received the highest importance rating of 10 while areas greater than 2000 m were assigned the lowest ratings of 2.5. Finally, Simpson’s ratio used for obtaining the parameter I ranged from 0.65 to 3.58 with values < 1.0 being assigned the lowest rating of 2.5 and values > 2 ascribed the highest rating of 10. The maps for each parameter are highlighted in Fig. 3.

The aggregated GALDIT vulnerability map resulting from the weighted overlay of the GALDIT parameter maps is shown in Fig. 4.

Fig. 4
figure 4

GALDIT vulnerability map for the coastal aquifer of Mombasa North coast, Kenya

Full size image

The GALDIT overall index map was found to range from 3.83 to 8.33 which indicates that no part of the aquifer is under a “very low vulnerability,” i.e., (< 2.5). The computed area coverages for the vulnerability classes are shown in Table 7.

Table 7 Vulnerability categories and their percentage area coverages in the study area
Full size table

Table 7 shows that almost 60% of the aquifer’s area coverage is under moderate vulnerability while 29% and 12% are experiencing low and high vulnerabilities, respectively. A previous vulnerability study on the aquifer which was based on data obtained at pre-monsoon and the peak of the rainy season shows an equally high percentage of moderate vulnerability to seawater intrusion at 64% and 55% (Ezekiel et al. 2016). However, a closer look at the data trends shows a progressive increase in the low vulnerability area coverage from 13 to 20% observed at pre-monsoon and peak of the rainy season in previous studies to 29% in the current study (post-monsoon). This corroborates previous studies highlighting that rainfall may be responsible for an increase in groundwater recharge which in turn increases the pressure head of freshwater against seawater and hence reduce the impact of SWI (Idowu 2017). In other words, as the rainy season progresses, the aquifer tends to become less vulnerable to SWI impacts. More importantly, these findings all point to the fact that vulnerability to seawater intrusion is dynamic and varies with seasons.

Seawater mixing index

The GQISWI values for each groundwater sample represented in Table 8 show there is a wide variation in the water type in the aquifer from a minimum value of 28 (saline water) to a maximum value of 59 (mixed groundwater). A more detailed observation shows that 80% (12) of the samples fall within the range of the mixed water category while the remaining 20% (3 samples) were in the saline water category (Table 8). The surface raster map produced by spatially interpolating the GQISWI data further confirms that the groundwater is largely mixed, with the saline water mostly concentrated towards the southern parts of the study area close to the ocean (Fig. 5). Furthermore, the GQISWI map establishes that the aquifer’s groundwater is largely brackish especially in regions towards the northern part of the study area while saline water prevails mostly along the coastlines and towards the southern end of the study area. Hence, Fig. 5 further provides an idea of the spatial distribution of the GQISWI values which is a more holistic view of the groundwater quality related to seawater intrusion.

Table 8 Computed GQIpiper and domain category for each sampled location
Full size table
Fig. 5
figure 5

The GQISWI map of the coastal aquifer of Mombasa North coast, Kenya

Full size image

The statistical computation of the GQISWI map is 68% and 32% for the brackish and saline water areas, respectively. This finding compares with an earlier salinity study on the study area based on EC and TDS values, which shows that over 90% of the aquifer’s GW is either saline or brackish (Idowu et al. 2018). Two previous studies that focused on the direction of groundwater flow in the study area (Munga et al. 2006; Idowu 2017) report that the flow direction is mainly northwards towards the ocean. Higher water flow rates can decrease the concentration of salts (Hsu 2005). This implies that the constant flow of groundwater in the northward direction may lead to a lower build-up of salts in the groundwater along the flow paths. This might explain why the north and northeastern parts of the study area are characterized by brackish water, which possesses lower concentrations of salts than saline water predominantly concentrated on the southern part of the study area.

The computed GQIpiper(mix) and GQIpiper(Dom) according to Eqs. 4 and 5 are highlighted in Table 8. According to the hierarchical analytical framework for water quality classification by (Tomaszkiewicz et al. 2014), water samples whose values fall within the GQIpiper (mix) range of 0–25 and GQIpiper(Dom) range of 50–75 are classified as the Na-Cl water type belonging to domain II. It can be observed that for this study, the GQIpiper(mix) are all < 25 and the GQIpiper(Dom) values are all between 50 and 75 (Table 9). Hence, the groundwater is classified as a Na-Cl water type impacted by salinity.

Table 9 Computed GQISWI, Total hardness, and WQI values at each sampling location
Full size table

Table 9 presents the aggregation of the GQISWI, total hardness, and WQI computations as described in the methodology.

Total hardness

According to the WHO guidelines, there are currently no health-based threshold values for hardness in drinking water. However, extremely hard water may precipitate CaCO3 scales when boiled depending on the pH and alkalinity. On the other hand, “very soft” water may be more corrosive for water pipes due to its low buffering capacity (WHO 2017). The computed total hardness values according to Stuyfzand’s classification system range from 0.36 (very soft) to 2.60 (hard) with a mean value of 1.25 implying moderately hard groundwater (Table 9). It is further confirmed by the GIS-based total hardness map generated from the values (Fig. 6). The area computations for the hardness show that the percentages for very soft, soft, moderately hard, and hard water were 1%, 27%, 67%, and 5%, respectively. This implies that almost 70% of the groundwater in the study area can be considered moderately hard.

Fig. 6
figure 6

Total hardness map for the coastal aquifer of Mombasa North coast based on Stuyfzand’s classification system

Full size image

Based on Stuyfzand’s classification system, the natural occurrence of moderately hard water could range from fresh to brackish saltwater (Table 4) depending on the geochemical composition of the water. Previous groundwater quality studies in some portions of the study area show that the total hardness values in mg/L are quite high but are below both the WHO guidelines and Kenya standard values of 500 mg/L and 600 mg/L, respectively (Mwamburi 2013).

Water Quality Index

The WQI values obtained prove that the WHO guidelines are more stringent than the National standard as ranges of 53 to 252, and 48 to 190 were observed in each case, respectively (Table 9). The spatial maps generated from the WQI values (Fig. 7) provide a more holistic understanding of the WQI values. It can be observed that the WQI map based on WHO guidelines shows that more areas fall under the “poor water category.” The area computations for the WQI categories are presented in Table 10.

Fig. 7
figure 7

Water Quality Index maps for the coastal aquifer of Mombasa North coast based on WHO guidelines (a) and Kenya standard (b) for potable water

Full size image
Table 10 The area coverages for each WQI category according to the standards used
Full size table

About 60% and 84% of the aquifer’s coverage are considered good water while 38% and 14% are regarded as poor water for the WHO guidelines and Kenya standards respectively (Table 10). The comparatively high “good water quality” coverage (84%) and low “poor water” coverage (14%) observed for Kenya standards show that the WHO guidelines are more nuanced. Notwithstanding, these findings suggest that studies over multiple timeframes are required to establish a more enduring WQI status of the study area. The indices used for the WQI classification were mainly cations and anions related to SWI pollution, and the GQISWI analysis has established that the groundwater is the NaCl type; hence, areas with poor water quality are largely impacted by poor salinity due to SWI. The poor water regions on the western part of the study area coincide with the high altitude Ngu-tatu hills known for a high concentration of salts in the groundwater from a previous study (Idowu et al. 2017). However, further studies are needed to ascertain the cause of these high concentrations of salts, which appear to be due to factors beyond SWI. Additionally, the WQI results suggest that the thresholds stipulated in water quality standards are sometimes high enough to pass groundwater with reasonably high ion contents as good drinking water as observed in W1–W4, W7–W10, and W14 (Table 9). This observation is comparable with studies on other aquifers. For instance, a study on the Hajeb Layoun-Jelma basin in central Tunisia identifies the groundwater of the shallow aquifer as Na-Cl and Ca–Mg–Cl type but the water is largely suitable for drinking according to the WHO standards except in a few cases with high salinity levels (Aouiti et al. 2021).

Integrated characterization of the groundwater

Prior related studies generally interpreted the results of different groundwater characterization approaches separately and did not link the results of the multiple approaches together (Kura et al. 2014; Idowu et al. 2017; Alfaifi et al. 2019). By combining the GALDIT overlay index, with the GQISWI, the Stuyfzand TH classification scheme, and WQI some holistic observations on the groundwater characteristics could be made. The surface maps created for each classification scheme using spatial interpolation tools also helped elucidate more on the geochemistry. The results of the GALDIT overlay index which classifies the vulnerability of the aquifer into low (29%), moderate (59%), and high (12%) vulnerabilities (Table 7), are comparable with that of the GQISWI that categorizes the aquifer into 68% and 32% brackish and saline water, respectively (Fig. 5). Vulnerability to seawater intrusion is not necessarily the same as the seawater intrusion status of a coastal aquifer. Hence, the vulnerability percentages of the GQISWI variations do not exactly match in both figures (Figs. 4 and 5). However, it can be inferred that areas under moderate or high vulnerabilities according to GALDIT tend to have Na-Cl water type as shown in Table 9 where the GQISWI classifies all the samples as mixed groundwater of Na-Cl water type.

Throwing Stuyfzand’s TH-based classification into the discussion, the TH values range from a minimum value of 0.36 to a maximum value of 2.60 and an average of 1.25 (Table 9). These values indicate that the natural occurrence of the groundwater is the [F Fb B Bs], which stands for “Fresh,” “Fresh–brackish,” “Brackish,” and “Brackish–salt,” respectively (Table 3). This classification agrees with the earlier GQISWI classification that identifies the groundwater as brackish or saline (Table 9). It can also be observed that the same water type (Na-Cl water type based on GQISWI) may produce a wide range of water hardness from soft to hard. Although Stuyfzand’s TH-based classification gives a relatively broader description of the groundwater than the GQISWI classification system, it can act as a preliminary check into the nature of the groundwater. In other words, GQISWI classification can narrow down the observations from Stuyfzand’s TH-based classification system.

The first observation when the WQI classification is considered alongside the other approaches is that WQI values for the WHO and national standards can vary. For instance, the mean WQI value obtained for the WHO is 106 which indicates that the groundwater largely falls under the poor water category while the WQI value of 85 observed for the national standard is indicative of good water quality in this study. A more overarching observation is that WQI values of the groundwater have no direct association with the softness or hardness of the water. For instance, “very soft” water can be a “very poor” quality drinking water (Table 9 line 5) and “hard” water can be “good” drinking water. Also, the fact that the WQI maps do not have similar trends to the GALDIT vulnerability and GQISWI maps (Figs. 4, 5, and 7) suggests that water quality in coastal aquifers is not solely determined by seawater intrusion.

Policy implications

In redefining the roles of the government towards achieving the sustainable management of groundwater in sub-Saharan Africa, the following points have been highlighted:

  1. 1.

    Aquifer-level resource planning

  2. 2.

    Stakeholder engagement and participation

  3. 3.

    Administration of groundwater use

  4. 4.

    Ground waste and wastewater discharge control

  5. 5.

    Groundwater monitoring and information provision and,

  6. 6.

    Land surface zoning for groundwater conservation” (Tuinhof et al. 2011)

By linking the above roles with the findings of this study, the first observation is that the vulnerability map (Fig. 4) and other groundwater characterization maps (Figs. 5, 6, and 7) can be valuable for the land surface zoning for groundwater conservation in the study area (no. 6). The maps can be a basis for locating special control areas such as portions extremely prone to SWI where groundwater abstraction is prohibited to avoid further deterioration. Furthermore, the methodology for creating a geodata-base outlined in this study (Fig. 1) can act as a guide for creating a continuous and temporal groundwater monitoring regime for the study area and similar coastal aquifers (no. 5).

In summary, the methodology presented in this study where GALDIT overlay index and geospatial and analytical approaches were combined can act as a valuable template for the management and continuous monitoring of coastal aquifers. The results of this study contribute further insights into the status of groundwater in the coastal regions of Kenya and the impacts of SWI in the region. Additionally, the study demonstrates the importance and relevance of geospatial techniques in developing the knowledge base required for sustainable groundwater management decision-making. The findings of the study will inform policymakers and the relevant water resource management towards developing new solutions or improving existing solutions for managing groundwater and increasing access to potable water. By extension, the solutions can further contribute to the achievement of Goal 6 of the Sustainable Development Goals which aims to ensure clean water and sanitation for all.

Conclusion

In this study, the coastal aquifer of Mombasa North Coast was extensively studied using a combination of different approaches that complement each other for creating a detailed understanding of the aquifer. The approaches used include the GALDIT overlay index, seawater intrusion groundwater quality index GQISWI, total hardness computation based on Stuyfzand’s classification system, and the Water Quality Index (WQI) based on the WHO guidelines and the created a more visual understanding of the groundwater condition in the study area. The considerable standard deviation values of the groundwater hydrochemical parameters indicate a wide variation across the aquifer. The study suggests that areas under moderate or high vulnerabilities to seawater intrusion tend to have Na-Cl water type. It is also possible for a similar water type (Na-Cl water type based on GQISWI) to produce a wide range of water hardness from soft to hard. GQISWI classification can narrow down the observations from Stuyfzand’s TH-based classification system. Additionally, WQI values of the groundwater have no direct association with the softness or hardness of the water. Findings from the GALDIT index suggest that the coastal aquifer is largely experiencing a moderate impact of SWI intrusion (59%). The GQISWI assessment shows that 68% of the aquifer’s coverage is characterized by brackish water while the water type is of the NaCl type impacted by salinity (domain II). The GALDIT and GQISWI add credence to previous groundwater studies in the study area. Although the values indicate the presence of SWI in moderate levels, the total hardness and WQI analyses results do not suggest that the groundwater poses any major health risks based on its chemical compositions. However, it is recommended that further studies on temporal scales be done. The findings of this study provide a baseline for subsequent studies on the coastal aquifer of Mombasa North coast and the methodology introduced can act as a template for similar studies in other regions.