Development of an Integrated Irrigation Water Quality Index (IIWQ_Index) model

The groundwater quality largely depends on the dissolved solutes and trace metals in aquifers (Mostafa et al. 2017; Zaman et al. 2018). These substances come from weathering of rocks and the dissolution of minerals such as calcite, dolomite, gypsum, lime, silicate, etc., in the aquifer basement and soil minerals (DANR-UC 2021). These mineral-originated salts are spread to the water and remain on the soil or are absorbed by the plants. There are harvesting problems associated with the overall salt content, and specific management practices are therefore necessary for satisfactory crop growth and production. The suitability of water for irrigation uses is justified on the probable severity of hazard that can be predicted to develop during continuous use. The hazards that result differ both in type and quantity, and are affected by soil, local climate, and plant, along with the ability and knowledge of the irrigation activist (Shuhaimi-Othman et al. 2007; Islam & Mostafa 2021a; Rahim & Mostafa 2021). Thus, there is no set limit on water quality; rather, its suitability for use is determined by the conditions of utilization, which affect the accumulation of the water constituents and may restrict crop yield. The soil hazards most often encountered and used as a base to assess water quality are those connected to salinity, sodicity, soil structure, water permeability, metal and ion toxicity, and other miscellaneous problems.

Irrigation water quality depends on the physical and chemical soil properties, the nature of irrigation practice, firming management, and crop diversity. There is no single index model that could assess the water quality accurately. In addition, the potential of adverse effects of water components varies with crop variety and soil condition. There are many worldwide irrigation water quality index (IWQI) models developed to evaluate the irrigation water quality for better crop growth and production (Wilcox 1948; Doneen 1964; Ayers 1977; Ayers & Westcot 1985; CCME 2006; Simsek & Gunduz 2007; Meireles et al. 2010; Bauder et al. 2014; Bozdağ 2016; Arslan 2017; Hussain et al. 2018; Zaman et al. 2018; Singh et al. 2020). However, those models have limitations and are not able to accurately evaluate the irrigation water quality.

For instance, the Canadian Water Quality Index (by the Canadian Council of Ministers of the Environment, CCME) was used to assess irrigation, aquatic culture, and drinking water quality; however, specific hazard class or particular water quality parameters were not considered in this method (UNEP-GEMS/Water 2007). Simsek & Gunduz (2007) have proposed a geographical information system (GIS)-based irrigation water quality model with specific geochemical parameters and some toxic metals to evaluate the water quality for irrigation of the Simav Plain in Turkey. Another new technique, proposed by Meireles et al. (2010), uses only four water quality criteria. Ashraf et al. (2011) established an IWQI using GIS for calculating the sodium adsorption ratio (SAR), saturated sodium percentage (SSP), and residual sodium carbonate (RSC) with other parameters. Romanelli et al. (2012) assessed irrigation water quality (IWQ) by combining geological landscapes, water chemistry, and other indicators such as hydraulic conductivity, electrical conductivity (EC), SAR, sodium%, SSP, RSC, and aquifer width to assess water suitability for irrigation in Wet Pampa Plain, Argentina. Bozdağ (2016) used an analytic hierarchy process to quantify the irrigation water suitability in central Anatolia, Turkey. Besides, Singh et al. (2020) developed the entropy weighted irrigation water quality index in Uttarakhand, India. These diverse water quality strategies are only partly suitable for irrigating agriculture because of the inconsistency in cropland conditions and crop variety. Moreover, several limitations of these models were identified, including the limited number of parameters, hazard classes, and their rating, scoring, and weighting values. The present study proposed an indexing model that considered the maximum number of water quality parameters with a total of six hazard classes. Modified rating values of water parameters and the weight factor of each hazard class are used. The proposed indexing model explores the suitability of various types of water for every irrigation practice. The procedures presented in this paper have been improved to give more practical actions for assessing and managing water quality-related hazards of irrigated farming.

Several researchers have been working on agricultural water quality and suitability for irrigation purposes since 2000 (Eaton 1950; Wilcox 1955; Doneen 1964; Kelley 1963; Richard 1954; Hem 1970; UCCC 1974; Freeze & Cherry 1979; Todd 1980; Matthess 1982; Ayers & Westcot 1985; Raghunath 1987; Rhoades 1992; Gupta & Gupta 1998). However, there are few water quality indices developed after the 2000s (CCME 2001; Simsek & Gunduz 2007; Meireles et al. 2010; Maia & Rodrigues 2012; Hussain et al. 2018). The existing indexing models considered only some selected and limited numbers of parameters. Since then, no one method was considered a combination of using the water quality parameter value, recognized permissible range of parameters, and various ratings with scoring values of parameters and hazard classes. For example, Simsek & Gunduz (2007) used only the scoring of parameters and weight factor of hazard class, but did not consider the concentration of parameters. Another index model developed by Meireles et al. (2010) used a limited number of parameters, but the rating and weight factors were not included in their methods. These models showed different indexing values and categories for the same water samples, i.e., one model exhibited the water quality category as good, but the other model showed it as poor. Table 1 shows the diverse recognized index methods with the characteristics of their equations for assessing the irrigation water quality. This table shows that the evaluation techniques of water categorization are different between methods. Every method partially fulfilled the requirements of a well-fitted and complete indexing equation. The ambiguity of water quality indexing encourages the development of a diversified and improved model that considers maximum parameters and hazard classes for obtaining the best results for detecting irrigation water suitability.

Most of the index models considered the lowest rating value for the rejected category of the parameters to be 1 (Ayers & Westcot 1985; Simsek & Gunduz 2007); however, this value for a parameter can affect the total index value and hence the lowest score of the parameter for the rejected category should be zero. The acceptable and permissible ranges of water quality parameters for irrigation are retroactive and not uniform. According to recent research results, the study should consider the modified ranges of some parameters and convert them into a uniform pattern.

The contaminated surface water, including domestic and industrial wastewater, is used as irrigation water in many areas due to inadequate water resources that increase environmental hazards. The continuous use of this type of water in agricultural activities causes a substantial increase in the number of toxic metals in soils, and consequently an increase risk in food safety. Existing methods do not properly address the metal toxicity of irrigation water for calculating the indexing value. There are numerous ways that toxin can enter the foodstuffs through soil-crop-food transfer systems. Thus, the crops grown on the contaminated soils are the major routes of toxic hazards exposure to the human body. Several studies confirmed a significant portion of toxic substances uptake by crops, with consequences including the bioaccumulation in the human body and livestock (Karim et al. 2008; Chauhan & Chauhan 2014; Leblebici & Kar 2018; Chaoua et al. 2019; Islam & Mostafa 2021b). This proposed indexing model considered all the toxic hazards in water and soils and the uptake by crops yield.

The indexing method is a technique used to quantify the amount of change, if any, in the index. An index model is a popular tool for evaluating a physical state of a system. In a water quality index model, it can explore the qualitative state of a water body. The study considered mathematical models, which can predict the physical problems with a rational level of accuracy.

There are four steps considered for developing any water quality index model. These are the selection of parameters, their weightage, sub-index, and aggregation of sub-index. The selection of hazard class, scoring and weight of the class, and rating value of each parameter are considered in the proposed IIWQ_Index. A detail of these steps is discussed here.

The selection of parameters is a vital step in the constituents of a water quality index. Traditionally, the indices have a different number of parameters, fluctuating from 4 to 26 (Dojlido et al. 1994). The category of the system used for the choice of parameters can usually be divided into three classes: fixed, open, and mixed types. The maximum water quality index studied have used a fixed number of parameters (Liou et al. 2004; Almeida et al. 2012). For the fixed and mixed types, the selection of parameters aims to choose those that have the highest impact on water quality. In this study, we used the mixed type selection procedure of parameters. However, Abbasi & Abbasi (2011) stated that there was no way to obtain 100% precision by choosing the parameters.

The study considered six hazard classes: (1) salinity hazard, (2) sodicity problem, (3) permeability of the soil, (4) ions and trace metal toxicity to crop, (5) changing in soil structure, and (6) miscellaneous effects to plants (Table 2), where the hazard class, i.e., changing in soil structure, is incorporated with other hazard classes of existing models. Twenty-seven irrigation water quality parameters are considered in the above-mentioned classes. The fixed system allows the parameters to choose the hazard classes, i.e., the salinity hazard, sodicity problem, the permeability of the soil, and changes in soil structure, and the open system chooses the ions and trace metal toxicity and miscellaneous hazard classes. Some parameters are often considered for different categories of hazard classes because of their harmful impacts. For this reason, these are included as operating parameters to the calculation of sub-index values in these hazard classes. A brief description of hazard classes and parameters and their relative impacts on soil, yield quality, and environment are stated below.

Salinity hazard was measured through EC and/or TDS (both are used if correlation matrix, r<0.95) and was given the highest score to calculate the index value that affected plants and led to saline soil condition. Saline soils usually have a pH <8.5 and contain mainly Ca, Mg, and Na salt of Cl⁻, SO₄²⁻, CO₃²⁻, HCO₃, NO₃, and PO₄³. A salinity hazard exists if salt accumulates in the crop root zone at a concentration that causes a reduction in yield as roots of the plants are unable to uptake enough water to keep the plant hydrated in saline soil (Ayers & Westcot 1985).

The sodicity problems are usually defined separately due to the harmful effects of Na⁺ on the physical properties of soils and plants. It was given the second-highest score and is usually measured by Na%, SSP, and SAR (or adjusted sodium adsorption rate, SAR_adj) which affected soils, and led to sodic soil conditions (Lesch & Suarez 2009).

The infiltration rate was given the next priority after the sodicity hazards because it has a significant influence on the irrigation water quality index. The water quality infiltration rates depend on soil texture (the ratio of sand, silt, and clay in soil), organic/humic matter content, degree of compaction, and chemical make-up (USDA 2014).

This hazard class is very flexible in choosing water quality parameters. It can exclude or include any components related to toxicity in crops if required. The proposed model considered some components, including Na⁺, Cl⁻, B, K, Fe, Mn, As, Cu, and Zn that are toxic to plants if they present in higher concentrations in water and soil. These metals directly affect the soil environment, plant growth, and the quality of yields, which through bio-accumulation, are accumulated in the human body by the food chain (Saha et al. 2021).

Soil properties depend on the concentration of Na, Ca, Mg, organic matter, and microorganisms in the soil mixture. The study observed that once the soil is irrigated with high Na in waters, Na potentially removes the Ca and Mg in the soil by ion exchange, which deteriorates the soil structure.

A flexible hazard class named ‘miscellaneous’ included the rest of the water quality parameters: pH, Ca, Mg, NO₃⁻, SO₄²⁻, PO₄³⁻, CO₃²⁻, HCO₃⁻_, etc., which were considered less sensitive to crop and soil. This class included irrigation water quality parameters such as TH, RSBC, RSC, and MAR. A report showed that TH increased the soil's basicity and enhanced the micronutrient toxicity (Tan 1994).

The score of hazard classes and the rating value of parameters are important factors in the index calculation (Gazzaz et al. 2012; Sutadian et al. 2016). The study followed the well-recognized analytic hierarchy process to select the weight value of each hazard class. The proposed model is given a score from 1 to 6 for each hazard class based on their importance in the index model (Table 2). The salinity hazard is considered the most important issue in irrigation water quality evaluation, and its score is 6 (s = 6). The effects of the miscellaneous hazard class on plants are considered to be the minor factor, and the score of this category is given 1 (s = 1). The scores for the remaining four hazard classes are set to 5, 4, 3, and 2 for sodicity hazard, the water infiltration rate of the soil, toxic to crops, and changing soil structure, respectively. The proposed model sets modified scoring and rating values for each hazard class and parameter, based on the Food and Agriculture Organization guidelines for agricultural water quality and other literature (Ayers & Westcot 1985; Fipps 2003; Hoffman 2010; Hussain et al. 2010; Bauder et al. 2014; Zaman et al. 2018). These issues might show radical differences in topographical settings with different soil conditions and different crop patterns. The weight value of each class was measured by dividing by 21 (total score). For example, if the scoring of sodicity hazard is 5, then the weight value obtained is 0.238 as the total weight is 1. The rating value (r) of all parameters in each class is considered 3 to 0 (Table 2). The parameter value at rating 3 denoted the maximum value of the excellent range, but rating 0 signified the rejection category. The value of parameters at ratings 2 and 1 represented the ‘good’ and ‘poor’ ranges of water category for irrigation purposes, respectively.

This step aims to develop the water quality parameters into a scale where the actual values of the parameters are different units, and the ranges of parameters are varied (Abbasi & Abbasi 2011). In most of the water quality indices, the parameters can only aggregate when they are on the same unit. Therefore, rescaling or normalizing to form sub-indices is necessary. Three different methods are commonly employed to establish the sub-index functions of parameters. Here, establishing rating curves or sub-index functions are presented based on the permissible limits from the legislated standards, such as technical regulations, national water requirements, and World Health Organization /FAO standards or international directives. Rating factors for each parameter are calculated in Step 1, and then the sub-index value is measured in Step 2.

The rating coefficient (R_c) is the unitless and dimensionless factor. For r = 1, 2, and 3; R_c are 0.167, 0.333, and 0.5, respectively, but at r = 0, R_c may be excluded from the equation. In the case of TDS, 488.5, 450, and 2,000 mg/L are the values of V_i, V_min, and V_max (Tables 2 and 4); and R_c = 0.333. In case of any critical condition, such as heavy industrial discharge, lithological abnormality, abandoned mine, or a radioactive substance that may present in water to a minimum of ten times higher than the usual level, then the r-value should be −0.001 instead of 0 in Equation (1).

After obtaining the aggregated index value of water samples, it is essential to categorize the water quality status for irrigation suitability. The water samples are classified into five categories depending on the r value between 0 and 3 (Table 3). The scoring values of the hazard classes and rating values of the parameters are shown in Table 2. The medians of these values are used to set the upper and lower limits used in each category. The IIWQ_Index value <40 is rejected for irrigation water users (Table 3). Such waters could impair soil quality massively and result in production loss. If the index value is between 40 and <60, the suitability of the corresponding waters falls under the poor category. When the index value is between 60 and <70, the water quality falls within the moderate category, indicating the crops are moderately tolerable for irrigation purposes. When the score value is between 70 and <80, the water quality is considered to be in the good category. Finally, with a score of 80 and above, the water quality falls under the excellent category. The waters usually obtained higher index values when most of the hazard classes, including the salinity, sodicity, and permeability of the soil, fall within the permissible limits.

The proposed index model evaluates the irrigation water quality in such a way that it helps choose crop patterns and increase crop production. The existing different water quality indices showed different suitability categories for the same water, attracting the water users for a unique index model. This study used different models, i.e., the CCME WQI (2001), Simsek & Gunduz (2007), Meireles et al. (2010), and Maia & Rodrigues (2012) indices for a water parameter data set to evaluate the irrigation water quality suitability and found different categories for the same water. The calculated result of the CCME WQI revealed that about 65% of samples had an index value from 70 to 84 and were in the good category, and about 35% of samples were found in the fair category. The average value of this index was 71.52 with a standard deviation of ±4.56. The same samples were categorized as 97.5% in the excellent category, and the rest of the samples were in good classes using the Simsek & Gunduz index. The Meireles model classified that 32.5% of samples were in the no restriction and 67.5% were in the low restriction categories. Finally, 15.5% and 84.5% of samples were falling under the excellent and good quality categories using the Maia method. To avoid these dissimilarities in indexing results, it is important to construct a suitable indexing equation in which all possible water quality parameters, all hazard classes, perfect rating, and scoring factors should be included. The present study established an indexing equation that included the maximum number of parameters and hazard classes for achieving the best possible results.

Another important anomaly was observed in the limit value of selected water quality parameters in existing indexing methods. The minimum and maximum values of the parameters vary with permitted authorities and places. Different index models considered a different number of parameters and their limit values, and so the final index value is different. Ayers & Westcot (1985) proposed the lowest rating value of parameters was 1, followed by other recent index calculations for irrigation water suitability. However, the proposed model considered the lowest value of r as 0 for every parameter for the rejection category of water. If the r-value of all counted parameters is 0, then the final value of the IIWQ_Index becomes 0. The increased IIWQ_Index value indicates a better quality of water for irrigation (Table 3).

However, for any critical condition, if a highly hazardous substance has found a minimum of ten-folds higher than the usual concentration of groundwater, then the r-value should be considered −0.001 instead of 0 (rejection value) in Equation (1). According to Table 4, if the average arsenic (As) concentration in any groundwater samples was about 0.5 mg/L, then the calculated Q_i value would be −364. In this case, the S_i value of the toxicity class obtained was −3.52 instead of 13.83, and the water quality of this study area ultimately fell into the poor category.

There are four types of groundwater available for irrigation: calcite (Ca–HCO₃), calcite–dolomite (Ca–Mg–HCO₃), sodic (Na–Cl), and mixed type (Ca–Mg–Na–HCO₃–SO₄), and the proposed IIWQ_Index is equally suitable for those types of groundwater. Hence, the study considered two types of water – absolute calcite and sodic water to evaluate the suitability and fitness of this proposed method.

Forty sampling stations in the Kushtia District (the Ganges River basin area) situated in the middle-west part of Bangladesh (Figure 1(a)) were designated for this study as an example of calcite water. The concentration/value of analyzing geochemical is shown in Table 4. Other parameters of irrigation water quality such as Na%, SAR, SAR_adj, SSP, MAR, RSBC, RSC, and PI were calculated from the usual methods (Richards 1954; Doneen 1962; Todd 1980; Gupta 1983; Raghunath 1987; Saha et al. 2008). Studies illustrated that the groundwater of the study areas was highly mineralized, and almost 100% of samples were calcite-type (Uddin et al. 2011; Islam & Mostafa 2021c, 2021d). It was determined by Piper's and Chadha's diagram as well as other various statistical models. The parameter rating factor (Q_i), hazard sub-index (S_i), and IIWQ_Index were calculated using Equations (1)–(3) and reported in Table 4. The aggregated index value for this type of water was 75.77 and fell into the good category.

Twenty sampling sites in the Chittagong District (the coastal belt) situated in the southeast part of Bangladesh (Figure 1(b)) were categorized as 100% sodic water types. As a coastal part of the country, the groundwater samples were highly sodic types with high EC, Na, and Cl⁻, and low Ca²⁺ and Mg²⁺ loads (Islam & Majumder 2020; Serder et al. 2020). The analysis results of the water samples are shown in Table 5. The parameter rating factor (Q_i), hazard sub-index (S_i), and IIWQ_Index were calculated by the same procedure. The index value of this type of water was 36.51, which fell into the rejection category.

Table 5 showed that several water parameters exceeded the maximum acceptable limit. In this case, the r-value became 0 and the Q_i value reached 0. The excess concentration of Na⁺ (767 mg/L) and low concentration of Ca²⁺ and Mg²⁺ led to the higher value of Na%, SSP, and SAR, which made a lowering index value. This type of water is very harmful to both soil and crop growth, and hence, this water is completely unfit for irrigation purposes.

In addition to the two case studies of Bangladesh, eight water quality datasets of diverse geographical locations from seven countries were evaluated for the suitability of the proposed index IIWQ_Index model. The water types, EC, SAR, and the calculated IIWQ_Index values with water categories of the datasets are shown in Table 6. The sub-index values of salinity and sodicity hazard classes occupied the majority of the total IIQW_Index value. Generally, these two classes are dependent mainly on the EC and SAR of water. EC counts the total dissolved solutes of water, such as Ca, Mg, and Na salt of Cl⁻, SO₄²⁻, CO₃²⁻, HCO₃⁻, NO₃⁻, and PO₄³⁻. Heavy calcite and sodic water showed higher values of EC and TDS. A salinity hazard exists if salt accumulates in the crop root zone to a concentration that causes a loss in yield because the roots of the plants are unable to uptake enough water to keep the plant hydrated in saline soil. The reduced water uptake in crops hampered the growth rate of the crop (Ayers & Westcot 1985). SAR is an ideal index to evaluate the possibility of Na-alkali hazard because it measures the soil capacity to adsorb Na⁺ from irrigation water. Irrigation water with a higher SAR value can damage the soil structure by a cation-exchange reaction between Na⁺ in water and Ca²⁺ and Mg²⁺ in soil.

The IIQW_Index analysis results showed that very low or high concentrations of minerals in water samples were harmful to soil and plant health, which provided a lower index value. The concentration of the minerals should be within the permissible limit to obtain a suitable range of Qi and Si value as well the final index value. For example, in Ghana and Nigeria, the groundwater samples carry very low amounts of minerals, with less than 20 mg/L of Ca²⁺, Mg²⁺, and Na⁺, and they have very low EC and TDS values. For this reason, the index value was relatively lower (Ghana: 72.11 and Nigeria: 63.06), although these waters were more non-sodic than in other countries as shown in Table 6. The low concentrations of all components present in water showed abnormal values of irrigation water quality parameters such as SAR, Na%, SSP, and others. This type of water did not present a suitable index value. Hence, the lower IIWQ_Index values for these above-mentioned places supported the proposed model. The calculated results showed that the absolute calcite or dolomite type water samples did not give the highest index value. The IIWQ_Index values of different waters in Table 6 followed the order: sodic < calcite < calcite-dolomite < calcite-dolomite-sodic. In Bangladesh (1), all water samples were calcite-type (Ca–HCO₃), and the calculated index value indicated the water was in the good category. In this case, high Ca²⁺ and HCO₃⁻ concentration and the excess hardness of water deducted some points from the excellent category's index value (>80). The sodic-type samples of the coastal belt areas of Bangladesh (2) showed a very low index value and were categorically unfit for irrigation uses. In China, Pakistan, and Morocco, the water samples are mixed type, i.e., dolomite-calcite-sodic water with low SAR and medium values of EC and TDS. The concentrations of Ca²⁺, Mg²⁺, and Na⁺ in these waters were at a medium level that indicating the balanced constituents. These waters were of the excellent category (IIWQ_Index>80) for irrigation uses. Due to the relatively high sodicity, SAR, and EC, the samples of India (1) were fallen in the moderate category with an index value of 65.86. However, the mixed-type water with high EC in India (2) made it good quality water for irrigation. In addition, the study results revealed that the computed index values were apparently reversely proportional to the EC and SAR (Figure 2 and Table 6). Several studies supported the present findings (Rhoades 1992; Fipps 2003; Hoffman 2010). The calculated index values of the water of different locations support the acceptability of the proposed IIWQ_Index model.

A recent study showed that climate change has greatly influenced both irrigation water quality and quantity (Serder et al. 2020). These impacts directly affected soil fertility and crop production as well. At extreme weather conditions, changes in the rainfall pattern, heavy drought, and variations of physicochemical parameters value of water collectively influence the rate of productivity. This proposed indexing model may measure the variations of irrigation water quality through climate change.

The IWQI is the tool that can evaluate water quality more accurately. Several established irrigation water quality indices are used, but most of the indices have some limitations. The present study critically reviewed the methods and identified the causes and types of limitation. It proposed a new IIWQ_Index using the precise sub-index and aggregated equations that included the maximum number and types of water parameters to evaluate irrigation water conveniently. This model evaluated two types of water to identify the suitability category for irrigation purposes. The results showed the index values were 75.77 and 36.51, and their water category fell into the good and rejected categories for the calcite (Ca–HCO₃) and sodic (Na–Cl) type of water, respectively. The proposed IIQW_Index model was justified using diversified water quality datasets of eight different geographical locations that showed satisfactory results. The index model has some flexibility which makes it more suitable. This indexing method is easy to compute and deliver practical information on the suitability of irrigation water, this helping to decrease crop and soil damage from using poor-quality groundwater. The proposed index model is practical, precise, and applicable to all types of water in any topographical location and provides a simple analysis tool, even for a non-technical decision maker and farm manager.

All relevant data are included in the paper or its Supplementary Information.