Ecological Indicators: Sciencedirect [PDF]

Zitiervorschau

Ecological Indicators 113 (2020) 106229

Contents lists available at ScienceDirect

Ecological Indicators journal homepage: www.elsevier.com/locate/ecolind

Simultaneous comparison of modified-integrated water quality and entropy weighted indices: Implication for safe drinking water in the coastal region of Bangladesh

T

Abu Reza Md. Towfiqul Islama, , Abdullah Al Mamuna, Md. Mostafizur Rahmanb, , Anwar Zahidc,d ⁎

⁎

a

Department of Disaster Management, Begum Rokeya University, Rangpur 5400, Bangladesh Department of Environmental Sciences, Jahangirnagar University, Dhaka 1342, Bangladesh Department of Geology, University of Dhaka, Dhaka 1000, Bangladesh d Bangladesh Water Development Board, Dhaka, Bangladesh b c

ARTICLE INFO

ABSTRACT

Keywords: Water quality index Entropy theory Principal component analysis Coastal area Spatial autocorrelation index Fuzzy GIS

Drinking-water sources are subjected to vulnerability due to chemical contamination and health risk in the southern coastal region of Bangladesh. Different types of water quality indices are being used to express the water quality, though most of them have both advantages and limitations. In this study, we modified and proposed an integrated water quality index by principal component analysis (PCA), and compared with the entropy theory. Spatial autocorrelation index and fuzzy GIS analysis are applied to delineate the suitability of groundwater in the study area based on randomly selected 377 groundwater samples from various depths (15–300 m). This study determined 20 water quality parameters. To avoid the bias of parameter selection, PCA was performed to reduce 12 parameters loads (TDS, EC, Ca2+, Mg2+, Na+, K+, Cl−, HCO3−, NO3−, pH, PO₄2−, and F−). Unlike traditional indices, both desired and tolerable limits were considered in the modified-integrated water quality index (MIWQI). The results of entropy analysis depict NO3−, Cl−, and Na+ as the most contributing parameters influencing the groundwater quality and much emphasis should be paid on these parameters to stop further groundwater pollution. MIWQI and Entropy water quality index (EWQI) results classify 21.22% and 42.43% of water samples as good quality for drinking, respectively. The rest 79% and 58% are classified as medium to extremely poor-quality water in the study area. A significant spatial pattern was identified for both water quality indices which were positively correlated (Moran’s I > 0). The MIWQI classification system exhibited commendable outcomes compared to EWQI due to comprehensiveness. Fuzzy GIS map revealed the spatial heterogeneity of groundwater parameters showing weak spatial dependency.

1. Introduction Groundwater is the most valuable natural resource that provides support for the existence of all life forms in the universe, exploitation with huge xenobiotics load threatens this resource aside from the depletion of groundwater quantity (Macdonald et al., 2016). In many countries, groundwater is the major source of potable water. Quality degradation along with quantity depletion puts pressure on the groundwater resources. The regular quality surveillance program is thus necessary to become updated with the dynamic hydrogeochemistry of groundwater and to manage this resource sustainably.

⁎

Some factors usually govern the characteristics of groundwater such as the status of rainfall and other depositions and discharges, holding time, and reactions between rocks and waters (Wagh et al., 2017; Islam et al., 2017a). Both natural processes and anthropogenic activities are involved in groundwater quality aspects. In Bangladesh, about 98% of the drinking water is being supplied from the groundwater sources and unfortunately, about 6.5 million to 24.4 million people exposed to a varying degree of risk from high arsenic, salinity, and groundwater-storage depletion (Shamsudduha et al., 2019). In Bangladesh, there are some factors to play role in water quality and quantity directly or indirectly including prevailing climatic

Corresponding authors. E-mail addresses: [email protected] (A.R.Md. T. Islam), [email protected] (Md. M. Rahman).

https://doi.org/10.1016/j.ecolind.2020.106229 Received 6 November 2019; Received in revised form 18 January 2020; Accepted 16 February 2020 1470-160X/ © 2020 Elsevier Ltd. All rights reserved.

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

attributes, nature of river flow, lowering of the groundwater table, size of populations, agricultural practices, and anthropogenic activities (Biswas et al., 2014; Bodrud-Doza et al. 2016; Islam et al. 2017b). Recent literature warns about the presence of arsenic, brackish saline water and salinity intrusion in the groundwater system in the coastal districts of Bangladesh. For instance, arsenic (> 50 μg/L) and the Barkish saline water have been found in a shallow aquifer and saline water in the deep aquifer in that area (BGS DPHE, 2001; Ahmed et al. 2004; Halim et al. 2010). Therefore, regular monitoring of groundwater quality in the coastal area is critically important to avoid potential risks to human health. Thus, finding an appropriate tool such as a comprehensive drinking water quality index has attracted increasing attention in the recent decade. Hence, water quality indices (WQIs) can be used to easy access to groundwater quality by checking key parameters (Ishaku 2011; Rubio-Arias et al. 2012; Bhuiyan et al. 2016 Sun et al. 2016). A lot of efforts have been made by scientists all over the world to find out a comprehensive WQI system for universal use (Horton, 1965; Abbasi and Abbasi, 2012; Sutadian et al., 2016; Hou et al., 2016; Tripathi and Singal, 2019; Mukate et al., 2019). To avoid complexity, biasness and arbitrary weight selection problems to handle large data matrix, this study showcases a statistical way to select the meaningful contributing water quality parameters by PCA for the 1st step of MIWQI as a comprehensive and unbiased water quality index, based on physicochemical parameters following the ECR (1997) drinking water quality guidelines of Bangladesh. This index is also avoiding the weighted problems likes other weighted indices. The MIWQI will be considering threshold values in terms of both excessiveness and deficiency as both of the situations might have some deleterious effects on human health. Finally, this MIWQI will present a take-home message for the user about the actual condition of the water to drink or not to drink.. The methods to assess the suitability of water quality includes weighted techniques, including set pair tools (Li et al. 2011; Feng et al., 2014), multivariate tools (Wu et al., 2014), matter-element technique (Li et al., 2016), blind number (Yan and Zou, 2014), and analytic hierarchy procedure (Hosseinimarandi et al., 2014). However, these methods have some critical drawbacks, such as it does not properly estimate the weight of the parameters as well as only considering the upper thresholds of parameters of interest. To solve these problems, the entropy-weighted water quality index (EWQI) can be useful that compute the groundwater quality correctly (Li et al., 2010a, b). In this study, newly developed MIWQI will be applied to measure the suitability of the groundwater to drink and compare with the well-known EWQI to examine the sensitivity of the MIWQI in the entire coastal area of Bangladesh for the first time so far. In addition, spatial distribution by inverse distance weighting (IDW) and Fuzzy logic analysis have been examined as a tool to monitor groundwater quality (Burrough and McDonell, 1998; Sarath-Prasanth et al., 2012, Bonham-Carter, 1996, Kamrani et al., 2016). In spatial distribution, the IDW technique is considered as one of the best interpolation methods and we applied it in this study (Islam et al., 2017a; Habib et al., 2020). Therefore, the simultaneous application of the water quality index classifications should be facilitated the scientific justification and decision-making on a wide range of water quality parameters and covering a huge geographical area (Mukate et al., 2019; Habib et al., 2020). Some of the very recent integrated approach already presented a reasonable results of water quality evaluation (Li et al., 2010a, b; Wu et al., 2011; Fagbote et al., 2014; Amiri et al., 2014; Su et al., 2017; Gorgij et al., 2017; Tripathi and Singal, 2019; Mukate et al., 2019). However, in Bangladesh, such an approach is still missing although most people are depending on groundwater as the sole source of drinking water. Therefore, in this study, so far, the simultaneous comparison of the MIWQI and the EWQI was applied first time to appraise drinking water suitability using multivariate techniques, spatial autocorrelating index and fuzzy logic tools in Bangladesh.

2. Materials and methods 2.1. Study area specification The coastal region covers approximately 710 km length and comprises of 19 districts having 140 upazilas of Bangladesh which has been chosen for this study (Uddin et al., 2019). The elevation of the most coastal upazilas is about 1–5 m the above mean sea level. These 19 districts are strongly influenced by numerous geogenic and anthropogenic activities. The total coastal zone is nearly 47,201 km2 (WARPO, 2006) which locates within the sub-tropical regions between 21 23″ N and 89 93″ E (Fig. 1). The coastal districts have been defined as susceptible due to the strong effects on water quantity and quality from various natural and anthropogenic events including the overpopulation growth, sea-level rise, and high saline water intrusion (Iyalomhe et al., 2015). In the subtropical humid climate setting, pre-monsoon encompasses from March to May followed by rainy monsoon season from June to September and dry winter extends from November to February. Agricultural activities are the major source of income and most of the land is utilized for agricultural purposes. The coastal area belongs to the high density of the population with 20,000 populations per km2. The average temperature of the hottest month (May) is 30 °C, and the average annual temperature of the coldest month (January) is 18 °C (Bahar and Salim Reza, 2010). Approximately 75% of precipitation occurs during the rainy season from June to September. The annual average evapotranspiration (ETp) of 1000 mm is about 2.5 times slower than the annual average precipitation (Sarker et al., 2018). Humidity differs from 62% to 92% in the winter and rainy seasons respectively. In general, public health concerned in these 19 coastal districts is associated with wide degradation of groundwater quality via drinking water. 2.2. Hydrogeologic settings Tectonically, the coastal region is under the Bengal Foredeep Basin of late Holocene age which is influenced by little mild or no folding (Adhikari et al., 2006). Geology of coastal Bangladesh is strongly affected by active neo-tectonism such as sea-level fluctuation and tremendous tidal activity. The lithology of the coastal region consists of coarse to fine-grained sandstone and silty clay with peaty soil which overlaying by calcareous and non-calcareous soil layers. The lithostratigraphy of the region is demarcated into 7 sequences of sedimentation. Each sequence originated with coarse-grained sand that ended with fine-grained silt and clay sediments serving as a part of groundwater occurrence (Adhikari et al., 2006). This region occupies with Tertiary and Quaternary sediment deposits (Umitsu 1993). Hydrogeological point of view, the coastal region of Bangladesh occupies the Holocene coastal plain deposit which is mainly covered by the tidal deltaic sediment and marshy peat overlying by thick silty clay deposits. The depth of the lower shallow aquifer formation varied between 10 and 50 m that has inconstant saline water contents mostly < 1500 mg/L. The high salinities in groundwater mainly depend on salt fishponds, channels, and numerous rivers as good water sources for local inhabitants (Khan et al., 2011). Apart from various sources, rainwater harvesting gives an alternate water source during the rainy season. Furthermore, rainwater is infiltrated excessively in the shallow aquifer formation via managed aquifer recharge for the drinking water pathway during the dry season (Sultana et al., 2015). The unconsolidated floodplain sediments mainly sand, clay, silty clay, and sandy clay of the recent Quaternary age are the major drivers for the aquifer system forming in the coastal area (UNDP, 1982). A hydrogeological cross-section at the local, as well as regional level from north to south direction, provides a valuable clue about the aquifer features of Bangladesh (Fig. S1). Based on the features of the aquifer formation, the aquifers in the coastal region are demarcated into 3 2

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Fig. 1. Location map showing sampling sites of the southern coastal region.

types (UNDP, 1982). First, the upper shallow aquifer formation is mostly a few centimeters to 60 m thickness in the northwest part which is unconfined, where people mostly have withdrawn groundwater via excavated, and hand tube wells. Next, the middle shallow aquifer belongs to a few 10 m to more than 100 m thickness (DPHE, 1999) which contains pockets of saltwater invasion from coastal to estuarine flooded conditions. Third, the deeper aquifer formation occurs more than 200 m thickness with an inconstant feature, which is demoralized mostly in the southern coastal belt. Based on the pumping test characteristics, the transmissivity rate of the upper aquifer varies from 600 to 888 m2/day, and hydraulic conductivity varies between 7.83 and 19.4 m/day, which is distinctive for sandy aquifer formation (BGS DPHE, 2001). The storage coefficient value varied from 1.07 × 10−4 to 0.0165 (Sarker et al., 2018). The aquifer recharges mostly happen from precipitation and complex river and channel network systems in this region (Bahar and Salim Reza, 2010). Moreover, the lowest recharge happens in the northwestern portion, whereas the highest recharge occurs in the central area to the elevated evapotranspiration which declined groundwater flow in the northwestern coastal region. The piezometric table of groundwater mostly drops during the dry winter season and the overexploited water is replenished during the rainy season. Regional groundwater flow happens from north to south direction during the dry seasons suggests groundwater recharge is mostly from the outer side of the northern part, where upper clay deposit is lacking, while the water is discharged into the south of the Bay of Bengal (BoB). The groundwater flow may assist the saltwater invasion into the aquifers of the study region. Although salt or brackish origin of groundwater found in the entire coastal region, few isolated pockets of groundwater occurring in several sites where active flushing of groundwater takes place during the rainy season. The water chemistry in the coastal area is mostly governed by the parent rock and water quality is mostly influenced by the groundwater-rock exchanges and the different minerals have come into

interaction with the water-bearing formation. Based on available geological information, the deltaic and fluvial aquifers of the coastal region are mainly unconfined to semi-confined. 2.3. Sampling design and analytical techniques This study was monitored only the groundwater samples during the dry winter season. The groundwater samples were fitted with tube wells which are utilized for ingestion and irrigation uses without any prior treatment. A total of 377 samples were collected randomly in the dry season. The geographic locations of each tube well were estimated by a GPS (Explorist 200, Megellan). the local well owners and the Department of Public Health Engineering office have provided in detail info about tube well depths. Based on well depth ranges and availability, samples have been obtained from three types of tube wells: (1) deep tube wells (146–300-m depth), (2) transitional tube wells (71–145-m depth) and (1) shallow tube wells (15–70-m depth). The sampling sites are displayed in Fig. 1 and all the samples were acquired from several campaigns during 2015–2017. Before the collecting samples, the samples were removed from standing water by pumping the groundwater in each tube well as a minimum of 20 min. The sampling tube wells were pumping until stable pH and EC (electrical conductivity) were attained. We use the prewashed high-density polypropylene bottles following the guideline procedures for collecting all the groundwater samples (APHA-AWWA-WEF, 2005). It should be mentioned that two sets of replicated samples were obtained in each site. The HDPP sampling bottles were preserved in a cooler box at 4 °C and sent quickly to the laboratory for further analyses. The physicochemical variables were determined following the standard technique which is suggested by the American Public Health Association (APHA, 2005). The values of pH, EC, and TDS were estimated in the field survey, using portable pH/EC/TDS meters (Hanna HI 9811-5). The concentrations of main anions (Cl−, NO3−, HCO3− PO4, 3

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

CO3− and SO₄−) in groundwater samples were estimated by an ion chromatograph (761 Compact IC, Metrohm). The elemental concentrations (i.e. Na, K, Ca, Mg, F, I, Br, and Mn) in groundwater samples were determined by inductively coupled plasma mass spectrometry (Thermo Scientific X-Series2 ICP-MS). The precision of the analytical procedure was further estimated by using the ion charge balance error (ICBE) as described in milliequivalent per liter (meq/L). In the study, the ICBE varied between 2.54 and 9.52% with an average value of 6.03% which was within the permissible level of ± 10% (Domenico and Schwartz, 1990).

not associated, it means that they are presenting the inherent information of dimensionality in the multivariate dataset. Nevertheless, these outcomes cannot be attained if the inherent parameters are extremely associated with each other (OCED, 2008). To address this issue, three steps of the PCA technique are performed in this case: first, calculating the correlation coefficient if the associations identified between each parameter is high, the dataset is utmost appropriate for executing the PCA. Next, identifying the number of PCs to be regarded based on the percentage (%) variance defined by them. Third, rotating factors to enhance their understanding e.g., by the maximizing loading on each factor by each parameter (OCED, 2008). However, after executing the PCA, the parameters influencing maximum (greater than 0.35; either positive or negative) to the first 5 PCs have been selected which elucidated > 62.55% of the total variance. These include 14 parameters e.g., TDS, EC, Ca2+, Mg2+, Na+, K+, Cl−, HCO3−, pH, PO42−, salinity, Eh, I-, and Br-. Those parameters have been taken into account that was minimum association based on the correlation coefficient analysis. After the successful selection and reduction of 20 parameters by performing PCA, we found 12 parameters (e.g., TDS, EC, Ca2+, Mg2+, Na+, K+, Cl−, HCO3−, NO3−, pH, PO₄2−, and F−) that are the minimum association. Finally, 12 parameters are chosen for MIWQI further analyzing to illustrate the suitability of drinking water using the newly developed MIWQI based on their required and tolerable limits defined by Environmental Conservation Rules (ECR, 1997), Bangladesh.

2.4. Computation of modified-integrated water quality index (MIWQI) Considering the modified-integrated water quality index (MIWQI) computation, a total of 20 parameters were chosen for the current study, e.g., temperature, pH, EC, Eh, TDS, potassium (K), calcium (Ca), magnesium (Mg), sodium (Na), phosphate (PO4), nitrate (NO3), sulphate (SO4), chlorine (Cl), bromine (Br), iodine (I), salinity, carbonate (CO3), bicarbonate (HCO3), fluorine (F), and manganese (Mn). After choosing these 20 parameters, we transformed the values of these parameters by computing their Z-score values (Normalization) by the normality test. The main limitation of z-values transformation is that the sample size requires a large number, however, the sample size of the study is 377 which was collected from 19 coastal districts of Bangladesh during the dry winter season. The normalization is of paramount significance for the environmental dataset as the parameters might have various units and thus, it makes no logic to cumulative two values with various units (Dobbie and Dail 2013). After developing a dataset of zscore transformation values of these variables for all sampling sites, the reduction and selection of final parameters are required for the development of MIWQI. We developed the MIWQI in the following five steps:

2.4.2. Step 2: Computation of range values The values of tolerable concentration (TC) and required concentration (RC) are demonstrated based on their risks to human health by the (ECR, 1997) for the corresponding parameters. The values of TC, RC and range of corresponding parameters are outlined in Table 1. The range was computed by using the RC and TC illustrated by the (ECR, 1997). Statistically, it can be written by the following Eq. (2)

2.4.1. Step 1: Selection and reduction of parameters via principal component analysis Normalized z-scores provide robust insight into the parameters with an average of 0 and a standard deviation of 1. However, there is a systematic problem to determine the index value after selecting 20 parameters. To solve this problem, the reduction of the parameter is mandatory for making this index scientifically viable. Principal component analysis (PCA) is a widely employed tool for reducing the parameters in some water quality indices studies (OCED, 2008; Karkra et al., 2016; Hou et al., 2016; Tripathi and Singal 2019). The PCA tool brings some potential advantages. For instance, it performs the work without loss of inherent much information. Another advantage is that when the inherent parameters are associated, the higher orders PCs are capable to capture the more variability in the inherent dataset than any other parameters. The PCs were extracted after substituting each inherent parameter by a normalized value of that parameter as a unit variance. MATLAB software (Version 2018a) was used to perform all the mathematical analyses. PCA is a pillar of modern data tools to reduce the dimensionality of the dataset by elucidating the total variance (Primpas et al., 2010; Karkra et al. 2016). It provides precise information that is separated in many dimensions into a reduced number of dimensions that unassociated. By executing PCA, the parameter numbers have reduced from 20 to 12 e.g., TDS, EC, Ca2+, Mg2+, Na+, K+, Cl−, HCO3−, NO3−, pH, PO₄2−, and F− in this study. This is defined by assuming a multivariate dataset comprising k parameters x1, x2, …, Xk. determined on n entities. The PCs are performed on the correlation coefficient; thus, it is assumed that without loss of inherent information that each parameter has a sample average zero and variance of one. Any PCs for the dataset show a linear relationship of the parameters which is written in the following Eq. (1)

Y = a1 x1 + a2 x2 +

+ak xk

Range = TC

(2)

RC

where, TC = Tolerable Concentration RC = Required Concentration 2.4.3. Step 3: Calculation of modified tolerable concentration (MTC) Additionally, the range of each parameter has been recalculated in the following Eq. (3)

MTC = TC

(3)

(15%Range )

Again, the range was demarcated as the difference between the TC and 15% deficit of the range of the corresponding parameter (Table 1). It should be noted that the 15% deficit adopted to conscious the circumstances before reaching the threshold level of groundwater Table 1 Drinking water quality standards with calculated range and modified tolerable concentration values (ECR, 1997). Parameters

RC

TC

Range

MTC (15% deficit to original)

TDS EC Ca2+ Mg2+ Na+ K+ Cl− HCO3− NO3− pH PO₄2− F−

– 50 – 30 – – 150 – – 6.5 – –

1000 1500 75 35 200 12 600 500 10 8.5 6 1

1000 1450 75 5 200 12 450 500 10 2 6 1

850 1282.5 63.75 34.25 170 10.2 532.5 425 8.5 8.2 5.1 0.85

RC = Required Concentration, TC = Tolerable Concentration, MTC = Modified Tolerable Concentration. All the values are expressed in mg/l, EC in (μc/cm) and pH on scale.

(1)

where the ai (i = 1 to k) are constants (Jolliffe, 1973). As the PCs are 4

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

contamination since when the aquifer is polluted, it is difficult to retrieve them to their original condition (Table 1). The deficit % have a flexibility that can be transformed as per user requirement.

where m (i = 1, 2, 3, 4, ……., m) denotes the total number of groundwater samples; n (i = 1, 2, 3, 4, ……., n) indicates the number of physiochemical parameters of each sample. Second step, the standardized value “yij” can be estimated and then standard assessment matrix “Y” can be acquired following Eqs. (9) and (10), respectively:

2.4.4. Step 4: Calculation of sub index (SI) The threshold limits below required concentration (RC) as well as above the tolerable concentration (TC) are unsuitable for drinking purposes. The values that are less than minimum RC and above the MTC will influence the groundwater quality. Nevertheless, the values in between RC and MTC can be assumed as excellent for drinking uses. The observed value of ith parameter is above RC and below the MTC the quality rating as following the Eq. (4)

Yij

Y=

For the observed values of ith parameter below RC are quality rating as Eq. (5)

SI2 =

RC

Pá ¢ RC

P á ¢ MTC MTC

(6)

ej =

j

j=1

SIij

(7)

X=

x m1 xm2

xmn

m

(10)

Yij

i=1

1 lnm

m i=1

1 n j=1

Cj Sj

(11)

PijlnPij

(12)

ej (1

ej )

(13)

× 100

EWQI =

(14)

n j=1

j qj

(15)

The groundwater quality was divided into five ranks classification (rank-I, rank-II, rank-III, rank-IV, and rank-V), and the more details information are outlined in Table 2.

EWQI is one of the robust techniques to offer precise and inclusive information about the overall groundwater quality for ingestion purposes (Islam et al., 2017a; Wang et al., 2019; Adimalla et al., 2019; Habib et al., 2020). The following five steps are needed to compute the EWQI. The initial step is to determine the assessment matrix “X” which is calculated by the following Eq. (8)

x1n x2n

ymn

where Cj indicates the content of the parameter j (mg/L) and Sj denotes the drinking water quality standards (ECR, 1997) of the parameter j (mg/L). Finally, the EWQI is calculated by using the following Eq. (15)

2.5. Entropy water quality index (EWQI)

x12 x22

ym1 ym2

=

qj =

where, SIij = Sub-index value of ith sample and jth groundwater quality parameter

x11 x21

y1n y2n

The fourth step is to calculate the quantitative rating scale “qj” of the “j” parameter by the following Eq. (14)

2.4.5. Step 5: Calculation of final MIWQI The final MIWQI is computed as the sum of all sub-indices (SI) of each parameter acquired from Eq. (4) to Eq. (6) in step 4. The values of the final MIWQI are divided as per the Table 2 n

y11 y12 y21 y22

Pij = Yij /

Where, RC = Required Concentration SI = sub-index Pi = water quality of ith parameter MTC = Modified Tolerable Concentration Eventually, the difference between Pi and RC or MTC is demarcated by the corresponding RC or MTC to normalize the value to recognize the declining or increasing content of each parameter regarding its RC or MTC. Thus, that values can be included to the ultimate MIWQI.

MIWQIi =

(9)

where, xij is the primary matrix; (xij)min and (xij)max are the lowest and highest values of the physiochemical parameters of the groundwater samples, respectively. The next step is to computed the information entropy “ej” and entropy weight “wj” by the following Eqs. (11)–(13)

(5)

If Pi is above than the MTC, then the quality rating as Eq. (6)

SI3 =

(Xij ) min

The standard-grade matrix is defined in Eq. (10)

(4)

SI1 = 0

Xij

(Xij ) max (Xij ) min

2.6. Spatial distribution and fuzzy GIS analysis Different spatial interpolation models namely kriging, cokriging, inverse distance weighted model and so on were employed for measuring the spatial variations of the water sample. Among these interpolation model, the inverse distance weighted (IDW) model was utilized to show the spatial variations of MIWQI and EWQI in this study because of its simplicity and measure precision in comparison with other interpolation models like kriging (Islam et al., 2017a, b, c). The spatial variation of the groundwater sample was carried out by using the ArcGIS software (version 10.5). The main justification for using the

(8)

Table 2 Category criterions of MIWQI and EWQI for this study. MIWQI value

Category

EWQI value

Category

Possible usage

Rank

5

Excellent Good Fair Poor Unfit for Drinking

< 50 50–100 100–150 150–200 > 200

Excellent Good Moderate Poor Extremely Poor

Excellent for Drinking Good for Drinking Domestic, Irrigation, Industrial Not Suitable for Drinking Unacceptable

1 2 3 4 5

5

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

this situation in the coastal aquifers of Bangladesh. F− (range 0.01–16.11 mg/L; mean 0.823 ± 1.536 mg/L), and PO₄2− (range 0.07–1130 mg/L; mean 19.301 ± 116.066 mg/L) were found in the study area. The TDS (mean 3052.641 ± 4185.56 mg/L) value surpassed limits of 1000 mg/L for drinking water implies the polluted groundwater water in entire study area. Moreover, the Na+ (mean 780.562 ± 1106.163 mg/L) concentration overrides the Cl− concentration (mean 2045.937 ± 3183.464 mg/L) demonstrated the seawater intrusion into the water tables. Our findings are being corroborated with others because the area is already being affected by varying degrees of salinity due to the location near the Bay of Bengal and the high salinity in the surface water of that region (Shammi et al., 2017). Oinam et al. (2012) reported elevated Na+ concentration in groundwater attributed to rock dissolution, cation exchange along with salinity intrusion. The K+ ion (mean 21.634 ± 30.033 mg/L) might be of natural origin (Garg et al., 2009). However, the Ca2+ (mean 92.556 ± 113.941 mg/L) and Mg2+ (mean 100.519 ± 143.618 mg/ L) were found very similar in levels in the study area but both exceed the tolerable limit. The tolerable limit of HCO3− is 500 mg/L in drinking water and the study area has the mean concentration of 249.543 ± 145.769 mg/L) indicating the suitable condition for freshwater (Wu et al. 2017). NO3− concentration (mean 12.401 ± 27.672 mg/L) in groundwater occurs from the sea as well as from anthropogenic activities. However, it is recommended that the skewness must be within standard range ± 2 to be normal and more than that indicates extreme (Islam et al. 2017b). In the data set, Mg2+, NO3−, PO₄2− and F− showed higher skewness thus can be regarded as pivotal contributing factors in the study area.

IDW model is that it calculates the spatially interpolated values very quickly and precisely (Islam et al., 2017c). However, the fuzzy logic membership model is also utilized to describe the input dataset 0.0–1.0 scale based on the probability of a member of a comprehensive dataset (Bonham-Carter, 1996). The input values are transferred into several functions and operators which is demarcated into input values to a 0 to 1 probability scale (Ahmed et al., 2019). Geospatial maps of fuzzy membership values of the selected physicochemical parameters are prepared using the IDW model through Arc GIS 10.5 software. The advantage of using fuzzy logic membership is that it is a robust tool to clearly delineate the spatial variability of a particular site. Spatial autocorrelation namely Moran’s I index was utilized to appraise the spatial pattern of water quality indices. This technique contains global spatial autocorrelation, which defines the total spatial relationship for the study sites, and local spatial autocorrelation, which detects the degree of spatial autocorrelation in a particular site. The Moran’s I index is a parametric technique which exhibits autocorrelation value ranged within −1 and +1 and produces p-value and z-score to evaluate the autocorrelation level (Liu and Mao, 2020). A positive Moran’s I value suggests the data clustered spatially whereas a negative Moran’s I value implies the data dispersed randomly (Islam et al., 2017a). Further, the Moran’s I index was verified using 999 transformations at the p < 0.05 confidence level. The more detailed about calculation procedure of Moran’s I index can be found elsewhere (Islam et al., 2017a). All statistical analyses were done using SPSS (version 23.0) and MATLAB (version 2018a) softwares. Kolmogorov-Smirnov (K-S) test was used to identify the normality and homogeneity of the groundwater dataset. One-way ANOVA test was done to check the differences in the variance at a p < 0.05 significance level.

3.2. Fuzzy membership spatial variations of major water quality parameters in the study area

3. Results and discussions

To separate the groundwater into different classes the fuzzy logic membership approach was used along with GIS (Fig. 2). The maps were prepared using the logic values of the following parameters; TDS, EC, Ca2+, Mg2+, Na+, K+, Cl−, HCO3−, NO3−, pH, PO₄2−, and F− of groundwater samples that are selected for MIWQI and EWQI for appraising drinking water suitability. The maps show high TDS value in the southern parts of Barguna, Bagerhat and Khulna districts. The southwestern and southcentral parts exhibit to Medium TDS levels and the rest of the area has lower concentrations (Fig. 2a). High-level Mg2+ is noticed in the southern part of Barguna districts in central and southwestern are cover medium concentration of Mg2+ (Fig. 2b). EC and Na+ map demonstrates a low distribution level in the Northeastern to the South and Southeastern and medium level in the

3.1. General characteristics of groundwater in the study area Although this study analyzed 20 different parameters, in this section, the most significant parameters are discussed specifically. Physicochemical characteristics of water quality parameters including TDS, EC, Ca2+, Mg2+, Na+, K+, Cl−, HCO3−, NO3−, pH, PO₄2− and F− were analyzed. The details results are given in Table 3. Most of the water quality parameters found to crossed the limit value of drinking water as of (ECR, 1997) Bangladesh. However, some of the parameters such as HCO3−, pH and F− were within the standard limit. Subsequently, the EC, TDS and Cl‾ showed elevated concentration (Table 3). The intrusion of salinity into the groundwater table might be the pivotal cause of Table 3 Descriptive statistics of groundwater samples in the study area (n = 377). Parameters

TDS EC Ca2+ Mg2+ Na+ K+ Cl− HCO3− NO3− pH PO₄2− F−

Min

15.39 27 0.28 0.35 4.08 0 9 24.4 0.019 3.6 0.07 0.01

Max

27,775 50,500 659.05 1261.9 8848.4 246.75 19,133 1049.2 236 11 1130 16.11

Mean

3052.641 5701.253 92.556 100.519 780.562 21.634 2045.937 249.543 12.401 7.444 19.301 0.823

SD

4185.56 7675.118 113.941 143.618 1106.163 30.033 3183.464 145.769 27.672 0.9208 116.066 1.536

CV (%)

137.112 134.621 123.104 142.876 141.713 138.823 155.599 58.414 223.148 12.368 601.345 186.703

Skewness

2.561 2.493 2.119 3.5003 2.865 2.322 2.705 1.274 4.843 0.1507 7.984 7.262

Kurtosis

11.138 10.639 7.785 21.491 14.391 11.223 12.059 5.7101 31.051 5.461 67.708 61.957

K-S Test

0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

RC

0 50 0 30 0 0 150 0 0 6.5 0 0

TC

1000 1500 75 35 200 12 600 500 10 8.5 6 1

Entropy weight calculation for EWQI Informatin entropy(ej)

Entropy weight (wj)

Limit values

0.883 0.886 0.9004 0.888 0.879 0.881 0.852 0.966 0.809 0.994 0.893 0.953

0.109 0.107 0.094 0.105 0.114 0.112 0.139 0.031 0.1802 0.004 0.081 0.058

1000 1500 75 35 200 12 600 500 10 6.5 6 1

RC = Required Concentration, TC = Tolerable Concentration, Limit values (ECR, 1997 Bangladesh standard). All the values are expressed in mg/l, EC in (μs/cm) and pH on scale. 6

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

northwestern to central. The south part including Bagerhat and Barguna reported as high-level concentration and in case Na+ the Bhola districts are also included in high concentration (Fig. 2c and d). Moreover, Ca2+ map exhibits a low level covering most of the coastal

region (Fig. 2e). Central parts of Barisal and Bhola districts represent high-level membership of Ca2+ values and medium concentration are in Northwestern and central part of the coastal region. However, K+ Map identified that the southern and northeastern parts including

Fig. 2. Fuzzy membership spatial maps of TDS, EC, Ca2+, Mg2+, Na+ and K+, Cl−, HCO3−, NO3−, pH, Phosphate and fluoride concentrations. 7

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Fig. 2. (continued)

Pirojpur and Bagerhat districts whereas medium concentration of Cl− extends from southwestern to south-central of the study area and rest of area in low concentration (Fig. 2g). pH concentration is most of the study region falls in medium concentration and high concentration falls in the northwestern part of

Barguna, Bagerhat, Pirojpur and Chandpur are in high concentration. (Fig. 2f). The northwestern part of Jessore, Norail, Khulna, Gopalganj and southeastern part of Feni, Noakhali, Chittagong and Cox’s Bazar falls in low concentration of K+ (Fig. 2f). The results also show that high Cl‾ concentration is observed in southern parts of Barguna, 8

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Jessore, Norail and Gopalganj districts of the study area. (Fig. 2h). Phosphate map represents interestingly that only the northern part of shariatpur and Chandpur are identifying in medium and high concentrations while the rest of the coastal region are falls in low concentration of Phosphate concentration (Fig. 2j). Medium concentration of HCO3− is determined that most of the study area are falls under this category (Fig. 2i), the Noakhali, Chandpur and Pirojpur districts it represents in a high concentration of HCO3−. Fig. 2k shows that NO3− covers almost all of region as in low concentration. In the northern part of Barisal covers only high concentration of NO3− because of leakage of municipal sewage and impact of agricultural return flow through the percolation of agrochemicals and chemical fertilizer (Amiri et al. 2014). Fluoride high concentration is observed in southeastern part of Chittagong and Cox’s bazar districts in the study area, while the medium concentration is in southwestern and northeastern parts including Satkhira, Khulana, Bagerhat and Chandpur, Shariatpur and Lakshimipur districts. The rest of fluoride-concentration are falls in the low level of occurrences (Fig. 2l).

systematic PCA using 2D and 3D biplots along with correlation matrix confirmed 12 water quality parameters out of 20 to be used in preparing the MIWQI of the study area (Fig. 4 and Fig. S2). As seen from Table 5, it is observed that 12 physicochemical parameter pairs of water samples have a significant positive correlation such as TDS vs EC (r = 0.99), TDS vs Mg2+ (r = 0.67), TDS vs Na+ (r = 0.52), TDS vs Cl‾ (r = 0.97), EC vs Mg2+ (r = 0.67), EC vs Cl‾ (r = 0.97), EC vs Na+ (r = 0.54), Na+ vs Cl‾ (r = 0.52), Mg2++ vs Na + (r = 0.53), and Ca2+ vs Mg2+, (r = 0.58) at the 99% confidence level respectively. A positive significant correlation indicates a similar source, which can be either natural or anthropogenic source and mobility (Haloi and Sarma, 2012). For example, Ca2+ and Mg2+ are significantly positively correlated and both parameters have been included in the shortlist as both are of paramount importance, similar is the case of HCO3− and NO3−. The final shortlisted parameters can further be reduced in number as parameters like Ca2+ vs Mg2+, HCO3− vs NO3− are significantly positively correlated. However, they are vital for representing the water quality which cannot be overlooked. In addition, some pairs have also displayed an insignificant negative correlation. This indicates that the source of these parameters is independent of each other (Islam et al. 2017a). Hence, 12 parameters were finalized for the development of the MIWQI.

3.3. Principle components analysis (PCA) for MIWQI parameters selection It is one of the most important tasks in the development of MIWQI to select the most relevant contributing factors out of a large list of water quality parameters. It is noted that the MIWQI has been developed from 377 groundwater samples covering all the 19 coastal districts of Bangladesh. The Kaiser–Meyer–Olkin (KMO) and Bartlett’s tests of sphericity was conducted to the data set before preparing for PCA analysis. The results of KMO in the present data set confirmed the suitability for the PCA as KMO = 0.815 closer to 1 and higher than 0.05. Moreover, Bartlett’s test of Sphericity (level of significance < 0.001) examines that there is a significant relationship among the water quality parameters. Finally, PCA was run on the normalized data through the varimax rotation using the SPSS program (Table 4). The z-scores plot was found to be a straight line (Fig. 3a) and the PCA has been done on 20 parameters. To avoid biases projected data were classified into the first 5 PCs depending on the factor loadings (Table 4). The scree plot only exhibits ten PCs that elucidated more than 84.96% of the total variance (Fig. 3b). However, the first five PCs explained > 62.54% of the total variability, so the first 5 PCs have been reasonably taken to reduce dimensions further. Table 4 shows the first 5 PCs as they accounted for 62.54% of total variance having each Eigen values > 1. Finally, by performing the

3.4. MIWQI for the groundwater in the study area The MIWQI is developed based on physicochemical parameters through PCA and range calculation of the 12 selected parameters defined by the (ECR, 1997) Bangladesh standard. The results of the MIWQI is presented in Table 6. However, the rank was obtained according to the values presented in Table 2. Where, both MIWQI and EWQI categories were synthesized together to designed the ‘Rank‘. The Rank 1 depicted as ‘Excellent for drinking‘ followed by Rank 2 as ‘Good for drinking‘; Rank 3 as ‘Domestic, Industrial and Irrigation‘; Rank 4 as ‘Not suitable for drinking‘ and Rank 5 as ‘ Unacceptable‘ (Table 2). From the results, it illustrates that 5.039% of samples are excellent in quality for drinking, 16.180% are good and 11.405% samples are fair for drinking from the study area. However, in the study area, there is no water treatment plant is in operation thus the fair suitable category water is also subjected to drink without treatment. MIWQI value 3–5 depicts the involvement of anthropogenic activities to limit the water quality and are not suitable for drinking (Mukate et al., 2019). In the coastal region, 9.549% samples scored index value 3–5, might be

Table 4 Rotated factor loadings and Eigen values for water quality data. Rotated factor loadings for water quality data

TDS EC Ca2+ Mg2+ Na+ K+ Cl‾ HCO3− NO3− pH SO₄2− CO32− PO₄2− Salinity Temperature Mn Eh F− I Br

Eigen values of water quality data set

PC1

PC2

PC3

PC4

PC5

0.3905 0.3921 0.2208 0.3343 0.2753 0.1707 0.39 −0.007 0.1157 −0.107 0.1438 0.1165 0.1165 0.3896 0.066 0.1232 0.0968 0.0762 0.0997 0.0821

−0.1346 −0.1413 0.3722 0.1449 0.0101 −0.1107 −0.1053 −0.2296 0.2473 −0.1526 −0.2692 0.0432 −0.1333 −0.154 0.2152 0.1559 0.2061 −0.3403 0.3732 0.3968

0.0606 0.0569 −0.027 0.0383 0.0691 −0.236 0.0489 0.2476 −0.091 0.5847 −0.053 0.1384 −0.042 0.0531 0.219 −0.041 −0.546 −0.064 0.2937 0.227

0.1284 0.1058 0.11 −0.045 −0.356 −0.476 0.1267 −0.357 −0.184 −0.008 0.3412 −0.198 −0.439 0.1044 0.1614 0.1463 0.0083 0.1269 −0.032 −0.055

0.0233 0.0225 0.11 0.1764 0.0795 −0.127 0.0595 −0.327 −0.186 −0.003 −0.329 0.3269 0.0771 0.0184 0.1215 0.0485 −0.223 −0.408 −0.397 −0.419

9

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Eigen Values

Variance (%)

Cumulative Variance (%)

5.784 2.185 1.555 1.521 1.465 1.064 0.968 0.916 0.806 0.730 0.576 0.509 0.473 0.462 0.387 0.305 0.258 0.033 0.004 0.002

28.921 10.924 7.775 7.603 7.325 5.321 4.838 4.579 4.030 3.648 2.878 2.544 2.366 2.310 1.935 1.525 1.288 0.163 0.019 0.008

28.921 39.845 47.620 55.223 62.548 67.869 72.707 77.286 81.316 84.964 87.841 90.385 92.751 95.061 96.996 98.522 99.809 99.973 99.992 100

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Fig. 3. Z score variability of data and Scree Plot of variance of PCs. (a). Z score variability, (b) Scree Plot of variance.

Fig. 4. 2D biplot of two PCs in this study. Table 5 Correlation analysis between the twelve shortlisted parameters in the studied coastal region.

TDS EC Ca2+ Mg2+ Na+ K+ Cl− HCO3− NO3− pH PO₄2− F−

TDS

EC

Ca2+

Mg2+

Na+

K+

Cl−

HCO3−

NO3−

pH

PO₄2−

F−

1 0.99** 0.34** 0.67** 0.52** 0.27** 0.97** 0.01 0.15** −0.159** 0.18** 0.20**

1 0.34** 0.67** 0.54** 0.29** 0.97** 0.02 0.15** −0.155** 0.21** 0.21**

1 0.58** 0.32** 0.04 0.37** −0.280** 0.28** −0.199** 0.06 −0.08

1 0.53** 0.27** 0.70** −0.153** 0.25** −0.177** 0.25** −0.0

1 0.44** 0.52** 0.14** 0.15** −0.09 0.30** 0.01

1 0.27** 0.15** 0.26** −0.167** 0.32** 0.13**

1 −0.03 0.15** −0.172** 0.18** 0.18**

1 0.01 0.16** 0.11* 0.20**

1 −0.121* −0.00 −0.02

1 0.01 0.03

1 0.07

1

*. Correlation is significant at the 0.05 level. **. Correlation is significant at the 0.01 level. 10

Ecological Indicators 113 (2020) 106229

attributed to the high concentration of Na+, Cl−, TDS, and EC. The ions such as Mg2+, HCO3−, and PO₄2− from saline water sources originated in the sea. However, most of the groundwater samples (57.824%) have MIWQI values more than 5 indicate unsuitable for drinking might (Table 6) corroborated with elevated concentration of Na+, Cl−, TDS, EC, Mg2+, HCO3− and PO₄2−. The ‘Rank 5‘ of MIWQI encompasses 12 coastal districts out of 19 except Jessore, Khulna, Narail, Gopalganj, Chandpur, Feni and Chittagong district (Fig. 5).

9.02% 7.95% 40.60%

100% N = 377

Average Poor Extremely Poor 100–150 150–200 > 200

34 30 153

78 Good 50–100

20.68%

Sample numbers 1, 3, 4, 11, 12, 67, 88, 121–126, 128–137, 140–142, 149–151, 153, 154, 158, 160, 162–164, 167–174, 184–191, 194, 198–201, 203–206, 216, 230–233, 236, 242, 253, 254, 260, 264, 266, 274, 293, 296, 298, 317, 318, 323, 328, 331 2, 5, 10, 18–22, 29, 39, 41, 68, 84, 85, 89, 90, 101, 112–114, 116–119, 127, 138, 144, 155–157, 161, 175, 179, 182, 192, 195, 196, 202, 212–215, 217, 219–221, 223, 229, 237, 238, 244, 245, 247, 262, 265, 269, 270, 271,273, 275, 282–285, 291, 292, 295, 299, 303, 304, 319, 322, 327, 329, 359, 360, 375, 376 6, 30, 40, 56, 60, 95, 115, 120, 143, 152, 178, 207–211, 234, 235, 243, 250, 257, 258, 268, 272, 297, 300, 301, 363, 365–369, 377 57, 92, 96, 97, 103–106, 145, 180, 222, 224, 239, 246, 249, 255, 256, 259, 267, 281, 302, 321, 332, 338, 361, 364, 370, 372–374 7–9, 13–17, 23–28, 31–38, 42–55, 58, 59, 61–66, 69–83, 86, 87, 91, 93, 94, 98–100, 102, 107–111, 139, 146–148, 159, 165,166,176,177, 181, 183, 193, 197, 218, 225–228, 240, 241, 248, 251, 252, 261, 263, 276–280, 286–290, 294, 305–316, 320,324–326, 330, 333–337, 339–358, 362, 371 % of samples 21.75% No. of samples 82

100% N = 377

Entropy Water Quality Index EWQI values Category < 50 Excellent

9.55% 57.82% Poor Unfit for drinking 3–5 >5

36 218

Fair 2–3

43

11.41%

1, 2, 131, 134, 135, 142, 168, 199, 200–202, 205, 206, 220, 266, 275, 283, 328, 329 3, 4, 18–20, 29, 68, 117, 121, 123–125, 128, 132, 138, 140, 141, 149–151, 169–175, 182, 184–187, 190, 191, 194, 198, 203, 204, 213, 214, 219, 229, 230, 232, 237, 238, 244, 245, 247, 264, 269, 274, 284, 285, 291, 292, 296, 298, 303, 331, 359 21, 22, 39, 85, 88, 90, 101, 116, 119, 122, 129, 130, 133, 136, 137, 144, 153–156, 158, 160, 162, 164, 188, 189, 192, 212, 215–217, 221, 231, 233, 236, 253, 254, 260, 270, 273, 293, 317, 318 5, 10–12, 30, 67, 84, 89, 113, 114, 118, 126, 127, 157, 161, 163, 179, 207, 208, 211, 223, 234, 242, 243, 262, 265, 271, 272, 295, 299, 304, 319, 323, 360, 375, 376 6–9, 13–17, 23–28, 31–38, 40–66, 69–83, 86, 87, 91–100, 102–112, 115, 120, 139, 143, 145–148, 152, 159, 165–167, 176–178, 180, 181, 183, 193, 195–197, 209. 210, 218, 222, 224–228, 235, 239–241, 246. 248–252, 255–259, 261, 263, 267, 268, 276–282, 286–290, 294, 297, 300–302, 305–316, 320–322, 324–327, 330, 332–358, 361–374, 377 5.04% 16.18% Excellent Good Na+ > K+ > TDS > EC > Mg2+ > Ca2+ > PO₄2− > F− > HCO3− > pH. It is also identified that HCO3− and pH have a low impact on groundwater quality of the coastal area. Islam et al. (2017a) have reported that NO3 has the highest influential factor affecting the groundwater quality in the Surma basin, Bangladesh which is similar to our study. Our study shows that NO3−, Cl− and Na+ are the three influential parameters affecting the groundwater in the studied coastal region. However, more emphasis should be needed to NO3−, Cl− and Na+ parameters which pollute the groundwater, because these parameters have higher entropy weights and stress large impacts on water quality in the southern coastal region. For quality ranking with EWQI 21.75% of sample is Excellent, 20.68% are fall in a good category while 9.02% are a moderate category. Poor and extremely poor categories are 7.95% and 40.60% respectively (Table 6). 3.6. Comparison between the MIWQI and the EWQI To appraise drinking water quality in the coastal region of Bangladesh this two-indices defined differently by their quality rating values. In the case of MIWQI showed only 19 groundwater samples (5.04%) are categorized as excellent for drinking water (Rank 1), whereas the entropy water quality index represents that, 82 groundwater samples (21.75%) are categorized as excellent (Rank 1) (Table 6). Table 6, shows ‘good category‘ (Rank 2) in both MIWQI and EWQI which comprising 16.18% and 20.68% with 61 and 78 groundwater samples respectively. ‘Fair category‘ (Rank 3) in MIWQI covers 11.41% while EWQI covers 9.02% in moderate category (Rank 3). A total of 36 groundwater samples that falls in ‘poor category‘ (Rank 4: not suitable for drinking) and covers 9.55% of the total samples water quality as poor in MIWQI. However, EWQI covers 7.95% of sample as poor categories (Rank 4: not suitable for drinking) and including 30 groundwater samples. In MIWQI and EWQI ‘unfit for drinking‘ quality ranking is most frequent that covers 57.82% as (Rank 5: unacceptable) and 40.60% of the sample respectively (Table 6). The comparative analysis of MIWQI and EWQI with other studies that results in MIWQI are consistent with previous observations of Mukate et al. (2019) where results showed that water quality categories are classified as the same condition (Supplementary Table S1). The findings of EWQIs are also consistent with previous studies of Amiri et al. (2014) and Islam et al. (2017a). 11

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Fig. 5. . Spatial distribution map of MIWQI and EWQI obtained by groundwater samples. (a) MIWQI, (b). EWQI).

12

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Table 7 Spatial autocorrelation analysis between MIWQI and EWQI models in the study. MIWQI Features Moran's I Z-scores P values Distribution

Rank 1 0.491 0.716 0.473 Random

Rank 2 −0.315 −0.693 0.487 Random

Rank 3 0.818 1.892 0.058 Clustered

Rank 4 −0.085 −0.088 0.929 Random

Rank 5 0.013 0.120 0.904 Random

MIWQI model 0.179 1.832 0.066 Clustered

EWQI Features Moran's I Z-scores P values Distribution

Rank 1 −0.015 −0.008 0.993 Random

Rank 2 0.529 1.258 0.208 Random

Rank 3 −0.037 −0.011 0.990 Random

Rank 4 −1.461 −1.505 0.132 Random

Rank 5 0.085 0.466 0.640 Random

EWQI model 0.204 2.076 0.037 Clustered

central part of the Jessore, Khulna, Bagerhat, Narail, Gopalganj, Barisal, Patuakhali, Lakshimipur, Noakhli, Feni, Chittagong and Cox’s bazar districts. MIWQI values (rank 2 and rank 3) are observed in the northwestern part of Jessore, Satkhira and central parts including Bagerhat, Jhalakati and the southeastern part of Chittagong, Cox’s bazar and Bhola districts. By contrast, in the case of EWQI (Rank 2) is distributed in the southwestern part of Satkhira, Jessore, where (Rank 3) in the south and southcentral parts of, Barguna, Pirojpur, Barisal, Bhola, and the southeastern part of Cox’s bazar, Chittagong, Feni and northern parts of Chandpur, Gopalganj districts of the sampling sites. Poor water quality (rank 4) of MIWQI is observed in the south, southcentral and southeastern parts of Patuakhali, Gopalganj, Barisal, Noakhli and Chittagong districts while in EWQI it is identified in the southern and northeastern parts of Barguna, Pirojpur, Shariatpur, Chandpur and Lakshimipur districts of the sampling sites (Fig. 5a and b). These high MIWQI values might be attributed to the combined impacts of pollution from the release of solute ions, excessive withdrawn of groundwater, escape of industrial effluents from agrochemical fields (Islam et al. 2017b). Thus, groundwater should be checked before ingestion for drinking purposes by residents, especially in Patuakhali, Gopalganj, Barisal, Noakhli and Chittagong districts. The advantages of MIWQI over others‘ are many folds, first, this index is free from any kind of biasness in allocating weight. Second, selection of parameters is completely done using statistical and mathematical tools. Then, both threshold lower and upper limit are considered as contributing factors to water quality judgement. Further, deficit (%) selection is flexible can be choose based on specific criteria. Finally, convenient to construct with excellent reliability and ranks are sharp to define potability. Despite the immense advantages of the MIWQI over the existing water quality indices there are some possibilities of backdrop for the index such as the possibilities of uncertainties, homogeneity of the aquifer’s depth should be maintained, and the sampling periods should be kept short. However, further researches should be focused on land-use change affecting the groundwater system in the coastal region.

3.7. Spatial patterns of water quality indices Spatial autocorrelation was carried out on the water quality indicators (rank) for the purpose of appraising their spatial differences (Table 7). A high significant positive spatial autocorrelation (p < 0.05) was observed for rank 3 of the MIWQI in the study area, indicating that geographically neighbor sites have similar levels of rank 3 water quality. Similarly, A statistically significant difference was observed for the MIWQI (p < 0.1), which were also correlated positively (Moran’s I = 0.18, Z value = 1.83), demonstrating the substantial spatial aggregation of water quality indicators in the coastal region of Bangladesh (Table 7). By contrast, the Moran’s I values showed a weak autocorrelation at a significance level of p > 0.05 in all other ranks of MIWQI, which indicate a localization in water quality rank. In case of EWQI, non-significant positive Moran’s I values (p > 0.10) were detected for all ranks of EWQI in the study sites, which may be ascribed to the spatial heterogeneity in contaminant loads and localize water quality development plans (Table 7). On the other hand, a high significant spatial autocorrelation (p < 0.05) was identified for the EWQI, which were correlated positively (Moran’s I = 0.20, Z value = 2.07), revealing the considerable spatial integration in the study sites. The difference in spatial autocorrelation between rank and overall water quality indices may be attributed to groundwater flow regulation and effect of monsoon rainfall (Islam et al., 2017a; Liu and Mao, 2020). 3.8. The spatiality of MIWQI and EWQI in the study area Inverse distance weighted interpolation (IDW) method is used to generate spatial variation maps of groundwater samples (n = 377) for two groundwater quality indices (MIWQI and EWQI) in ArcGIS environment (Fig. 5) to visualize the spatial spread of water quality over the region. This spatial map is helpful to identify good, fair/moderate and vulnerable water quality sites in the study area for better management of water resources. Spatial distribution maps of MIWQI and EWQI demonstrate an increasing trend from southwest to northeast directions in the study area (Fig. 5). NWQI category (rank 5: unacceptable) is found more frequently in the south and southeastern parts of Satkhira, Bagerhat, Jhalakati , Pirojpur, Patuakhli, Barguna, Bhola, Naokhali, Chittagong, cox’s Bazar and northeastern parts of Gopalganj, Barisal, Chandpur, Sariatpur and Lakshimipur districts of the sampling locations, while High EWQI values (rank 5: unacceptable) are distributed in the southeastern and central parts of Satkhira, Pirojpur, Jhalakati, Patuakhli, Barguna, Bhola, Lakshimipur and Barisal districts, of Bangladesh, showing amalgamate point factors that affect groundwater quality by salinity intrusion from the Bay of Bengal. (Fig. 5a and b). Rank 1 of MIWQI is observed in northwestern and southwestern parts of Narail. Gopalganj, Chandpur, Chittagong and Feni, where EWQI Value (rank 1) is observed in northwestern, southwestern and

3.9. Implication for safe groundwater management In the context of water quality issues for safe drinking and domestic uses, the role of water quality indices is immense by reducing the effort to know actual quality scenario. There are a good number of water quality indices (WQIs) but they are not free from the drawbacks. The selection of groundwater quality parameters and assigning weights are the main problems. This research provided a clear picture of overall groundwater quality on a regional scale using MIWQI compared to EWQI and recognized the key parameters influencing groundwater quality in the coastal region. Particularly, in Bangladesh people in the coastal districts have been facing drinking water problems due to different types of pollution problems along with the water shortage. Inhabitants in the coastal areas as well as other parts of the country 13

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Fig. 6. A conceptual framework for the implication of MIWQI to determine the suitability of groundwater as drinking water.

clearly demonstrates their attitude towards avoidance of contaminated water for domestics uses. Moreover, the national targets to achieve the sustainable developments goals put a strong stewardship to water security for the whole nation including the vulnerable areas like coastal districts in Bangladesh. Furthermore, multiple risk factors such as salinization, trace metals, ionic imbalances, and scarcity altogether put extra pressure to the policy makers in formulating a suitable policy toolkit for ensuring safe water services. This research has the opportunity to contribute a lot in the water quality determination for potability. Here, both the upper permissible limit and lower threshold limit were applied to ensure the suitable and safe water quality judgement. Here, a set of 20 water quality parameters were analyzed and obviously it s quite unsuitable for judging with a short time. Therefore, this study successfully reduced the parameters to only 12 using statistical and mathematical techniques. And finally, 5 ranks (1: excellent to 5: unsuitable for drinking) were allocated systematically to decide the potability which become rather convenient to use. The findings will help further evaluation of groundwater quality in the southern coastal area as well as regional management activities (Fig. 6). First, MIWQI plays a pivotal role in drinking water quality appraisal, because it is the first initiative in developing an inclusive, unbiased water quality index namely modified-integrated water quality index (MIWQI) using the PCA and comparative study this index with EWQI to apprise drinking water suitability in Bangladesh. Furthermore, NO3−, Cl− and Na+ are the high influencing chemical parameters, thus degrading groundwater quality, and special focus should be paid on these parameters to stop further contamination of groundwater. As an integrated single value, MIWQI is vital for water managers, who require precise information about groundwater quality rather than different appraisal outcomes by various techniques. Second, this study also assessing the groundwater quality by Fuzzy membership GIS analysis which provides a scientific basis to represent the water quality status for the study region. To delineate the drinking water quality in the coastal region, regional management activities should draw much attention to the area the rank 5 unsuitable for drinking water purposes. So, this research provides an apprising of drinking water suitability that can be useful for testing the potable water quality considering the potential human health risk.

20.68% are as good category (rank 2); 11.41% and 9,02% in fair/ moderate (rank 3) respectively. Poor and unfit for drinking define 9.55% and 57.82% by MIWQI whereas 7.95% and 40.60% of sample is assessing as poor and extremely poor by EWQI which are not fit for drinking purposes. The spatial pattern of water quality indices was explored using spatial autocorrelation. Spatially, both water quality indices were statistically correlated positively (Moran’s I > 0) in the coastal region. The spatial distribution maps demonstrate that south, southwestern and southeastern parts of Satkhira, Bagerhat, Patuakhali, Barguna, Bhola, Chittagong, Cox’s Bazar and central region of Pirojpur, Jhalakati, Barisal districts are falls in extremely poor categories (rank 5). On the other hand, the northern and northwestern parts of Jessore, Narail, Khulna, Gopalganj, and northeastern part of Shariatpur, Chandpur, Feni, and some parts of Chittagong and Cox’s bazar of Bangladesh, showing less degraded groundwater quality. However, fuzzy membership analysis indicates that the southern part of the study region is highly polluted except in some parameters where medium concentrations of parameters cover the central portion of the coastal region. Thus, MIWQI appraises drinking water suitability properly as per the guidelines of Bangladesh. This technique can also be useful for any other guideline parameter. This research can be a suitable reference-scale for the local and regional water stakeholders as well as this index can be applied in other parts of the world. Author contributions A.R.M.T.I., designed, planned, conceptualized, drafted the original manuscript, and A.A.M., was involved in statistical analysis, interpretation; M.M.R., and A.Z., contributed instrumental setup, data analysis, validation; M.M.R., contributed to editing the manuscript, literature review, proofreading; A.A.M., and A.R.M. T.I., were involved in software, mapping, and proofreading during the manuscript drafting stage. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

4. Conclusion MIWQI is solely focused on the ionic contents presented in groundwater and assessing the groundwater quality properly. However, heavy weighted parameters that influencing MIWQI and solve this problem EWQI is performed over MIWQI to appraise drinking water perfectly. The comparison study between MIWQI and EWQI provides a comprehensive output about this groundwater quality with spatial variations that helps to easily identified groundwater quality for this region. The dominance major physicochemical parameters in the groundwater are in the following order, NO3− > Cl‾ > Na+ > K+ > TDS > EC > Mg2+ > Ca2+ > PO₄2‾ > F‾ > HCO3− > pH. In this study, MIWQI and EWQI reveals that 5.04% and 21.75% of groundwater samples (n = 377 and 19 coastal districts) are categorized as having an excellent for drinking (rank 1), 16.18% and

Acknowledgements The present study has been done in accordance with and supported by the following project headed by Dr. Anwar Zahid and entitled “Establishment of monitoring network and mathematical model study to assess salinity intrusion in groundwater in the coastal area of Bangladesh due to climate change” implemented under the Bangladesh Water Development Board and financed by the Bangladesh Climate Change Trust Fund, Ministry of Forest, Environment and Climate change, Government of Bangladesh. The authors would like to acknowledge the anonymous reviewers for improving quality of the manuscript. 14

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al.

Appendix A. Supplementary data

chemosphere.2019.125183. Halim, M.A., Majumder, R.K., Nessa, S.A., Oda, K., Hiroshiro, Y., Jinno, Km., 2010. Arsenic in shallow aquifer in the eastern region of Bangladesh: insights from principal component analysis of groundwater compositions. Environ Monit Assess 61:453–472. https://doi. Org /10.1007/s10661-009-0760-9. Horton, R.K., 1965. An index number system for rating water quality. J. Water Pollut. Control Fed. 37, 300–306. Hosseinimarandi, H., Mahdavi, M., Ahmadi, H., Motamedvaziri, B., Adelpur, A., 2014. Assessment of groundwater quality monitoring network using cluster analysis, ShibKuh Plain, Shur Watershed, Iran. J. Water Resour. Prot 06 (6), 618–624. https://doi. org/10.4236/jwarp. 2014.66060. Hou, D.W., Zeng, S.Z., Liu, J., Yan, M.T., Weng, S.P., Huang, Z.J., 2016. Characterization of prokaryotic and eukaryotic microbial community in pacific white shrimp ponds. J. Aquacul. Res. Dev. 7, 463–472. https://doi.org/10.4172/2155-9546.1000463. Ishaku, J.M., 2011. Assessment of groundwater quality index for Jimeta- Yola area, northeastern Nigeria. J. Geol. Min. Res. 3 (9), 219–231. Islam, A.R.M.T., Ahmed, N., Bodrud-Doza, M., Chu, R., 2017a. Characterizing groundwater quality ranks for drinking purposes in Sylhet district, Bangladesh, using entropy method, spatial autocorrelation index, and geostatistics. J. Environ. Sci. Pollut. Res. https://doi.org/10.1007/s11356-017-0254-1. Islam, A.R.M.T., Shen, S., Bodrud-Doza, M., Rahman, M.A., Das, S., 2017b. Assessment of trace elements of groundwater and their spatial distribution in Rangpur district, Bangladesh. Arab. J. Geosci. 10 (4), 95. https://doi.org/10.1007/s12517-0172886-3. Islam, A.R.M.T., Shen, S., Haque, M.A., Bodrud-Doza, M., Maw, K.W., Habib, M.A., 2017c. Assessing groundwater quality and its sustainability in Joypurhat district of Bangladesh using GIS and multivariate statistical approaches. Environ. Dev. Sustain 10.1007/ s10668-017-9971-3. Iyalomhe, F., Rizzi, J., Pasini, S., Torresan, S., Critto, A., Marcomini, A., 2015. Regional Risk Assessment for climate change impacts on coastal aquifers. Sci. Tot. Environ. 537, 100–114. Jolliffe, I.T., 1973. Discarding variables in principal component analysis II: real data. Appl. Statist. 22 (1), 21–31. Kamrani, S., Rezaei, M., Amiri, V., Saberinasr, A., 2016. Investigating them efficiency of information entropy and fuzzy theories to classification of groundwater samples for drinking purposes: Lenjanat Plain, Central Iran. Environ. Earth Sci. 75, 1370. https:// doi.org/10.1007/s12665-016-6185-1. Karkra, R., Kumar, P., Bansod, B.K.S., Bagchi, S., 2016a. Classification of heavy metal ions present in multi-frequency multi-electrode potable water data using evolutionary algorithm. Appl. Water Sci. https://doi.org/10.1007/s13201-016-0514-0. Khan, A. E., Ireson, A., Kovats, S., Mojumder, S. K., Khusru, A., Rahman, A., Vineis, P., 2011. Drinking Water Salinity and Maternal Health in Coastal Bangladesh: Implications of Climate Change. Environ Health Perspect 119:1328–1332 (2011). http://dx.doi.org/10.1289/ehp.1002804. Li, P., Li, X., Meng, X., Li, M., Zhang, Y., 2016. Appraising groundwater quality and health risks from contamination in a semiarid region of northwest China. Expo Health 3, 1–19. https://doi.org/10.1007/s12403- 016-0205-y. Li, P., Qian, H., Wu, J., 2010a. Groundwater quality assessment based on improved water quality index in Pengyang County, Ningxia, Northwest China. J. Chem. 7 (S1), S209–S216. Li, P., Qian, H., Wu, J., 2011. Application of set pair analysis method based on entropy weight in groundwater quality assessment—a case study in Dongsheng City, Northwest China. J. Chem. 8 (2), 851–858. https://doi.org/10.1155/2011/879683. Li, P., Wu, J., Qian, H., 2010b. Groundwater quality assessment based on entropy weighted osculating value method. Int. J. Environ. Sci. 27 (3), 31–34. https://doi. org/10.6088/ijes.00104020018. Liu, Y., Mao, D., 2020. Integrated assessment of water quality characteristics and ecological compensation in the Xiangjiang River, south-central China. Ecol. Indicator 110, 105922. Macdonald, A.M., Bonsor, H.C., Ahmed, K.M., et al., 2016. Groundwater quality and depletion in the Indo-Gangetic basin mapped from in situ observation. Nat. Geosci. https://doi.org/10.1038/NGEO2791. Mukate, S., Wagh, V., Panaskar, D., Jacobs, J. A., and Sawant,.A., 2019. Development of new integrated water quality index (IWQI) model to evaluate the drinking suitability of water. https://doi.org/10.1016/j.ecolind.2019.01.034. OCED, 2008. Handbook on Constructing Composite Indicators: Methodology and User Guide. doi:10.1787/9789264043466-en. Oinam, J.D., Ramanathan, A.L., Singh, G., 2012. Geochemical andstatistical evaluation of groundwater in Imphal and Thoubal district of Manipur, India. J. Asian Earth Sci. 48, 136–149. https://doi.org/10.1016/j.jseaes.2011.11.017. Primpas, I., Tsirtsis, G., Karydis, M., Kokkoris, G.D., 2010. Principal component analysis: Development of a multivariate index for assessing eutrophication according to the European water framework directive. Ecol. Indicators 10 (2010), 178–183. https:// doi.org/10.1016/j.ecolind.2009.04.007. Rubio-Arias, H., Contreras-Caraveo, M., Manuel-Quintana, R.A., Saucedo-Teran, R., Pinales-Munguia, A., 2012. An overall water quality index (WQI) for a man-made aquatic reservoir in Mexico. Int. J. Environ. Res. Public Health 9, 1687–1698. Sarath-Prasanth, S.V., Magesh, N.S., Jitheshlal, K.V., Chandrasekar, N., Gangadhar, K., 2012. Evaluation of groundwater quality and its suitability for drinking and agricultural use in the coastal stretch of Alappuzha District, Kerala, India. Appl. Water Sci. 2, 165–175. https://doi.org/10.1007/s13201-012-0042-5. Sarker, M.M.R., Van Camp, M., Islam, M., et al., 2018. Hydrochemistry in coastal aquifer of southwest Bangladesh: origin of salinity. Environ. Earth Sci. 77, 39. https://doi. org/10.1007/s12665-017-7196-2. Shammi, M., Rahman, M. M., Islam, M. A., Bodrud-Doza, M., Zahid, A., Akter, Y., Quaiyum, S., Kurasak, M., 2017. Spatio-temporal assessment and trend analysis of

Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ecolind.2020.106229. References Abbasi, T., Abbasi, S.A., 2012. Water Quality Indices. Elsevier. Adhikari, D.K., Roy, M.K., Datta, D.K., Roy, P.J., Roy, D.K., Mailk, A.R., Alam, A.K.M.B., 2006. Urban geology: a case study of Khulna City Corporation. Bangladesh. J Life Earth Sci 1 (2), 17–29. Adimalla, N., Qian, H., Li, P., 2019. Entropy water quality index and probabilistic health risk assessment from geochemistry of groundwaters in hard rock terrain of Nanganur County, South India. Geochemistry. https://doi.org/10.1016/j.chemer.2019.125544. Ahmed, K.M., Bhattacharya, P., Hasan, M.A., Akhter, S.H., Alam, S.M., Bhuyian, M.H., Sracek, O., 2004. Arsenic enrichment in groundwater of the alluvial aquifers in Bangladesh: an overview. Appl. Geochem. 19 (2), 181–200. Ahmed, N., Bodrud-Doza, M., Islam, A.R.M.T., Hossain, S., Moniruzzaman, M., Dev, N., Bhuiyan, M.A.Q., 2019. Appraising spatial variations of As, Fe, Mn and NO3 contaminations associated healthrisks of drinking water from Surma basin, Bangladesh. Chemosphere 218, 726–740. Amiri, V., Rezaei, M., Sohrabi, N., 2014. Groundwater quality assessment using entropy weighted water quality index (EWQI) in Lenjanat. Iran. Environ. Earth Sci. 2014 (72), 3479–3490. https://doi.org/10.1007/s12665-014-3255-0. APHA-AWWA-WEF, 2005. Standard Methods for the Examination of Water and Wastewater. 21th Edition. New York, Total Solids Suspended, Method 2540 D, 2-55 a 2-59. Bahar, M.M., Salim Reza, M.S., 2010. Hydrochemical characteristics and quality assessment of shallow groundwater in a coastal area of Southwest Bangladesh. Environ. Earth Sci. 2010 (61), 1065–1073. https://doi.org/10.1007/s12665-009-0427-4. BGS DPHE-, 2001. Arsenic Contamination of Groundwater in Bangladesh, British Geological Survey (BGS) and Department of Public Health Engineering (DPHE) Govt. of Bangladesh; rapid investigation phase, Final Report. Bhuiyan, M.A.H., Bodrud-Doza, M., Islam, A.R.M.T., Rakib, M.A., Rahman, M.S., Ramanathan, A.L., 2016. Assessment of groundwater quality of Lakshimpur district of Bangladesh using water quality indices, geostatistical methods, and multivariate analysis. Environ. Earth Sci. 75 (12), 1020. https://doi.org/10.1007/s12665-0165823-y. Biswas, R.K., Roy, D.K., Islam, A.R.M.T., Rahman, M.M., Ali, M.M., 2014. Assessment of drinking water related to arsenic and salinity hazard in Patuakhali district, Bangladesh. Int. J. Adv. Geosci. 2 (2), 82–85. https://doi.org/10.14419/ijag. v2i2. 3011. Bodrud-Doza, M., Islam, A.R.M.T., Ahmed, F., Samiran, D., Narottam, S., Rahman, M.S., 2016. Characterization of groundwater quality using water evaluation indices, multivariate statistics and geostatistics in central Bangladesh. Water Sci. 30, 19–40. https://doi.org/10.1016/j.wsj. Bonham-Carter, G.F., 1996. Geographic Information Systems for Geoscientists: Modelling with GIS. Computer Methods in the Geo-sciences. vol. 13. Elsevier Sci Pub, Pergamon, pp. 98. Burrough, P.A., McDonell, R.A., 1998. Principles of Geographical Information Systems. Oxford University Press, Oxford, pp. 333. Dobbie, M.J., Dail, D., 2013. Robustness and sensitivity of weighting and aggregation in constructing composite indices. Ecol. Indic. 29, 270–277. https://doi.org/10.1016/j. ecolind.2012.12.025. Domenico, P.A., Schwartz, F.W., 1990. Physical and Chemical Hydrogeology. John Wiley and sons, New York, pp. 410–420. DPHE, 1999. Main report and volumes S1–S5, report on Phase I, Groundwater studies for arsenic contamination in Bangladesh, Dhaka, Bangladesh. ECR (The Environment Conservation Rules) Bangladesh 1997. Drinking water quality standard. Fagbote, E.O., Olanipekun, E.O., Uyi, H.S., 2014. Water quality index of the ground water of bitumen deposit impacted farm settlements using entropy weighted method. Int. J. Environ. Sci. Technol. 11 (1), 127–138. https://doi.org/10.1007/s13762-0120149-0. Feng, L.H., Sang, G.S., Hong, W.H., 2014. Statistical prediction of changes in water resources trends based on set pair analysis. Water Resoure. Manage. 28, 1703–1711. https://doi.org/10.1007/s11269-014-0581-7. Garg, R.K., Rao, R.J., Saksena, D.N., 2009. Water quality and conservation management of Ramsagar reservoir, Datia, Madhya Pradesh. J. Environ. Biol. 30 (5), 909–916. Gorgij, A.D., Kisi, O., Moghaddam, A.A., Taghipour, A., 2017. Groundwater quality ranking for drinking purposes, using the entropy method and the spatial autocorrelation index. J. Environ. Earth Sci. 76, 269. https://doi.org/10.1007/s12665017-6589-6. Guey-Shin, Sh., Bai-You, Ch., Chi-Ting, Ch., Pei-Hsuan, Y., Tsun-Kuo, Ch., 2011. Applying factor analysis combined with kriging and information entropy theory for mapping and evaluating the stability of groundwater quality variation in Taiwan. Int. J. Environ. Res. Public Health 8, 1084–1109. Haloi, N., Sarma, H.P., 2012. Heavy metal contaminations in the groundwater of Brahmaputra floodplain: an assessment ofwater quality in Barpeta District, Assam (India). Environ. Monit. Assess. 184 (10), 6229–6237. Habib, M.A., Islam, A.R.M.T., Bodrud-Doza, M., Mukta, F.A., Khan, R., Siddique, M.A.B., Phoungthong, K., Techato, K., 2020. Simultaneous appraisals of pathway and probable health risk associated with trace metals contamination in groundwater from Barapukuria coal basin, Bangladesh. Chemosphere 242. https://doi.org/10.1016/j.

15

Ecological Indicators 113 (2020) 106229

A.R.Md. T. Islam, et al. surface water salinity in the coastal region of Bangladesh. Environ Sci Pollut Res. DOI 10.1007/s11356-017-8976-7. Shamsudduha, M., Zahid, A., Burgess, W.G., 2019. Security of deep groundwater against arsenic contamination in the Bengal Aquifer System: a numerical modeling study in southeast Bangladesh. Sustainable Water Resour. Manage. 2019 (5), 1073–1087. https://doi.org/10.1007/s40899-018-0275-z. Su, H., Kang, W., Xu, Y., Wang, J., 2017. Assessing groundwater quality and health risks of nitrogen pollution in the Shenfu mining area of Shaanxi Province, Northwest China. Expo Health. https://doi.org/10.1007/s12403-017-0247-9. Sultana, S., Ahmed, K.M., Mahtab-Ul-Alam, S.M., Hasan, M., Tuinhof, A., Ghosh, S.K., Rahman, M.S., Ravenscroft, P., Zheng, Y., 2015. Low cost aquifer storage and recovery: implications for improving drinking water access for rural communities in coastal Bangladesh. J. Hydrol. Eng. 20 (3) B5014007-1-12. Sun, W., Xia, C., Xu, M., Guo, J., Sun, G., 2016. Application of modified water quality indices as indicators to assess the spatial and temporal trends of water quality in the Dongjiang River. Ecol. Indic. 66, 306–312. https://doi.org/10.1016/j.ecolind.2016. 01.054. Sutadian, A.D., Muttil, N., Yilmaz, A.G., Perera, B.J.C., 2016. Development of river water quality indices—a review. Environ. Monit. Assess. 188 (1), 58. Tripathi, M., and Singal, S, M., 2018. Use of Principal Component Analysis for parameter selection for development of a novel Water Quality Index: A case study of river Ganga India. https://doi.org/10.1016/j.ecolind.2018.09.025. Uddin, M.N., Islam, A.K.M.S., Bala, S.K., Islam, G.M.T., Adhikary, S., Saha, D., Haque, S., Fahad, M.G.R., Akter, R., 2019. Mapping of climate vulnerability of the coastal region of Bangladesh using principal component analysis. Appl. Geogr. 102, 47–57.

Umitsu, M., 1993. Late Quaternary sedimentary environments and landforms in the Ganges Delta. Sediment. Geol. 83, 177–186. https://doi.org/10.1016/0037-0738(93) 90011-S. UNDP, 1982. Groundwater Survey: The Hydrogeological Conditions of Bangladesh. UNDP Technical Report DP/UN/BGD-74-009/1, 113p. Wagh, V.M., Panaskar, D.B., Muley, A.A., Mukate., S.V., 2017. Groundwater suitability evaluation by CCME WQI model for Kadava River Basin, Nashik, Maharashtra, India. Model Earth Syst Environ. https://doi.org/10.1007/s40808-017-0316-x. Wang, D., Wu, J., Wang, Y., Ji, Y., 2019. Finding High-Quality Groundwater Resources to Reduce the Hydatidosis Incidence in the Shiqu County of Sichuan Province, China: Analysis, Assessment, and Management. Exposure and Health. WARPO, 2006. Coastal Development Strategy. Ministry of Water Resources, Government of the People’s Republic of Bangladesh, Dhaka. Wu, J., Li, P., Qian, H., Duan, Z., Zhang, X., 2014. Using correlation and multivariate statistical analysis to identify hydrogeochemical processes affecting the major ion chemistry of waters: case study in Laoheba phosphorite mine in Sichuan, China. Arab. J. Geosci. 7 (10), 3973–3982. https://doi.org/10.1007/s12517-013-1057-4. Wu, J., Xue, C., Tian, R., Wang, S., 2017. Lake water quality assessment: a case study of Shahu Lake in the semiarid loess area of northwest China. Environ. Earth Sci. 76, 232. https://doi.org/10.1007/s12665-017-6516-x. Wu, J.H., Li, P., Qian, H., 2011. Groundwater quality in Jingyuan County, a semi-humid area in Northwest China. E J. Chem. 8, 787–793. Yan, H., Zou, Z., 2014. Water quality evaluation based on entropy coefficient and blind number theory measure model. 1986-1874. J. Networks 9 (7). https://doi.org/10. 4304/jnw.9.7.1868-1874.

16