Water pollution during festival periods is a major problem in all festival cities across the world. Reliable prediction of water pollution is essential in festival cities for sewer and wastewater management in order to ensure public health and a clean environment. This article aims to model the biological oxygen demand (BOD5), and total suspended solids (TSS) parameters in wastewater in the sewer networks of Karbala city center during festival and rainy days using structural equation modeling and multiple linear regression analysis methods. For this purpose, 34 years (1980–2014) of rainfall, temperature and sewer flow data during festival periods in the study area were collected, processed, and employed. The results show that the TSS concentration increases by 26–46 mg/l while BOD5 concentration rises by 9–19 mg/l for an increase of rainfall by 1 mm during festival periods. It was also found that BOD5 concentration rises by 4–17 mg/l for each increase of 10,000 population.

INTRODUCTION

Modeling pollution loading in an urban storm water is important for non-point source (NPS) pollution control and water treatment (Vandewalle et al. 2012; Llopart et al. 2014). This is particularly important in festival cities, where the population can suddenly surge several fold during a festival period (Obaid et al. 2014; Hussein et al. 2015). The total number of population during the peak festival periods is increasing in almost all festival cities across the world. This is due to the growth of population, enhanced ability to move, and more economical capability (Shinde 2012). As the world population will continue to increase (UN 2005), and the economy will continue to develop, it can be expected that the number of visitors in festival cities during a festival period will continue to grow (Vijayanand 2012). Androgenic activities during a festival period by a huge number of the population cause a huge release of a pollutant to a sewer system, which subsequently causes acute impact on receiving waters (Gasperi et al. 2010).

Rainfall in many festival or pilgrimage cities, particularly those located in semi-arid regions is very erratic, sporadic and sparse. Therefore, sudden rainfall after a long dry spell often causes a high level of pollution in the storm water in such cities. This alteration includes increased runoff volumes and peaks, decreased time of concentration, decreased base flow recharge (Burns et al. 2005), and increases of the NPS pollution in the runoff (Ying & Sansalone 2010). Urban NPS pollutants include suspended and dissolved solids, nutrients, oxygen-demanding organisms, bacteria, pesticides, metals, oil, and grease. The transport of pollutants in the runoff from the land areas into water bodies is a natural process. However, inner city NPS pollution is intensified by activities associated with domestication. The most prevalent of these activities include increased impervious surfaces (e.g., roads, parking lots, sidewalks, roofs, and compacted areas), increased application of fertilizers and pesticides on municipal lawns and gardens, erosion from land disturbance due to the construction activities. Furthermore, increasing use of vehicles may cause pollutant inputs into the air and subsequent atmospheric deposition transferred to nearby aquatic systems by the runoff (Baird et al. 1996). The contamination in rainwater is discharged untreated into water courses from the natural land within the watershed in the urban areas. The magnitude of wet weather flows affects contaminant loads and their subsequent acute impact on receiving waters (Gasperi et al. 2010). Several studies have proven that the quality of water degradation in the receiving water bodies is caused by the storm-water runoff (Gasperi et al. 2010; Fortier & Mailhot 2014). The runoff in the municipal areas also plays key roles in the water-quality deterioration in the adjacent hydraulic components (USEPA 2008; Abdellatif et al. 2014). An accurate estimate of non-polluting loads, runoff from the city and forecast quality of the receiving water are important (Lee et al. 2010).

A festival is a day or time of religious or other celebration, which motivates people to visit holy places or festivals. A large proportion of such journeys is motivated by religious needs or entertainment as well as tourist purposes (Rinschede 1992; Gladstone 2005; Sharpley & Sundaram 2005). This causes an increase of population in pilgrimage or festival cities. Different anthropogenic activities by a large floating population, such as burning wood for cooking and disposal of waste, particularly food waste, increase pollution in sewer water. The festival cities are mostly historical cities. The sewer systems in most of these historical cities are very old and often unable to cope during high sewer flow due to sudden increase of population. Consequently, sewer overflow is very common in such cities, which significantly impacts public health and the urban environment. One of the prerequisites to manage sewer overflow is accurate forecasting of the sewer flow and sewage quality. A number of studies revealed that only reliable sewer flow or quality forecasting could help authorities to take necessary actions in advance to protect from the negative impacts of sewer pollution (Vazquez et al. 2010; Aziz et al. 2011; Mouri et al. 2012; Abyaneh 2014a, 2014b; Maruéjouls et al. 2014). Therefore, development of a sewer pollution simulation model is essential for appropriate management of the sewer system in festival cities, particularly during festival periods.

However, sewer pollution in festival cities does not only depend on population, rather it depends on many other factors, such as rainfall, sewer leakage, and type of human activities (Huong & Pathirana 2013). For example, rainfall during the festival period often changes the pollution scenario. Although studies from different parts of the world indicate both increase and decrease of pollution during rainfall, there is no doubt that rainfall has significant impact on sewer pollution, particularly during the festival period. The objective of the present study is to develop a sewer quality simulation model for Karbala city of Iraq. Karbala city is considered as one of the holiest cities in the world. Several million people gather in Karbala city during peak pilgrim or festival days. This causes an increase of population in the city by several fold. The deterioration of sewer water quality during the festival period is a great concern for city authorities. Reliable prediction of sewer quality is essential for wastewater management in the city.

Numerous physically based and data-driven models have been developed and applied to water quality modeling (Ferreira et al. 2005; Pathiratne et al. 2007; Kock et al. 2009; Vazquez et al. 2010; Aziz et al. 2011; Zeferino et al. 2012; Abyaneh 2014a, 2014b; Llopart et al. 2014). The models have been justified for different cities, according to availability and quality of data, geography of the city, and climate of the region. The novelty of the present study was to develop a model to predict the concentrations of biological oxygen demand (BOD5) and total suspended solids (TSS) in a sewer system in Karbala city center using structural equation modeling (SEM). It is expected that the study will be interesting to evaluate the performance of maintenance for sewerage quality as well as to assess the impacts of population and rainfall on urban wastewater quality.

METHODS AND MATERIALS

There are many problems in science and engineering that have been successfully solved using techniques such as statistical analysis. Statistical methods have been widely used for solving problems that have a linear relationship between the input and output variables. SEM is an advanced technique enabling researchers to assess a complex model that has many relationships, performs confirmatory factor analysis, and incorporates both unobserved and observed variables (Hair et al. 2006). It brings together the characteristics of both factor analysis and multiple regressions, which helps the researcher to simultaneously examine both direct and indirect effects of independent and dependent variables (Bagozzi & Phillips 1982; Gefen & Straub 2000; Hair et al. 2006). Furthermore, SEM allows researchers to measure the contribution of each item in explaining the variance, which is not possible in regression analysis; additionally, SEM can measure the relationship between the construct of interests at the second-order level (Hair et al. 2006).

SEM contains two branches of statistical techniques: covariance-based modeling, e.g., analysis of moment structures (AMOS) (Byrne 2013), and variance-based modeling, e.g., partial least square (PLS) regression (Gefen & Straub 2000). Among the two techniques, PLS is gaining more popularity and getting researchers' interest due to its ability to work efficiently with a small sample size and a complex model, make fewer assumptions about the distribution of data, and handle both reflective and formative measurement models (MMs), as well as single item construct (Henseler et al. 2009; Urbach & Ahlemann 2010; Hair et al. 2012; Muserere et al. 2015). PLS can maximize the explained variance of the endogenous latent constructs (dependent variables) and minimize the unexplained variances. It does not assume normality of data distribution and can develop a model with few observations (Asyraf & Afthanorhan 2013). PLS models are able to explain the variance of independent variables and correlation with response variables more efficiently (Haadland & Thomas 1988). It has been reported that the value of factor loadings/outer loadings and average variance extracted (AVE) in PLS-SEM is better than in AMOS (Urbach & Ahlemann 2010; Hair et al. 2012).

SEM has been used by different authors in recent years to analyse wastewater quality data as well as to validate the research model and test the proposed research hypotheses (Ji et al. 2013; Agustsson et al. 2014; Ai et al. 2015). Hurlimann et al. (2008) used SEM to explain and predict components of community satisfaction with recycled water use through the dual water supply system in Australia and reported that the SEM-based model can be used for the management of recycled water projects. Pantsar et al. (1999) used principal component analysis and SEM to study the factors affecting the composition of sewage of domestic origin in Australia and found that lifestyle of residents, day of the week and sampling time or weather have the highest influence on the pollutant levels. Ji et al. (2013) used SEM to assist in the establishment of a lake nutrient standard for drinking water sources in China and reported that the most predictive indicator for designated use-attainment is total phosphorus and chlorophyll. Agustsson et al. (2014) used SEM to analyse simultaneously the chemical oxygen demand (COD) and turbidity (TUR) concentrations in water in Switzerland and established the relation between the measured and the reference concentrations of COD and TUR in the water samples. Ai et al. (2015) used SEM to elucidate the linkages between some specific water contaminants and the watershed characteristic variables in China and reported that urban land use, largest patch index and the hypsometric integral were the dominant factors affecting specific water contaminant. The above-mentioned studies indicate the efficacy of SEM in wastewater quality modeling.

A PLS path model consists of two elements: (1) a structural model (SM) that represents the constructs and the relationships between the constructs; and (2) the MMs of the constructs that display the relationships between the constructs and the indicator variables. There are two types of constructs in an SEM: the exogenous latent variables and the endogenous latent variables. The higher-order construct allows reduction of complexity; therefore it needs to assess formative constructs and indicators (Edwards 2001). An important condition for a multidimensional construct is to identify the relationship with its underlying dimensions based on theoretical evidence and empirical considerations. The validation of a reflective MM can be established by testing its internal consistency, indicator reliability, convergent validity, and discriminant validity (Lewis et al. 2005).

The internal consistency refers to the interrelatedness of parameters. Traditionally, a measurement item's internal consistency is evaluated using Cronbach's alpha (CA) (IDRE 2015),

Cronbach' 
formula
where N = number of indicators assigned to the factor; σ2i = variance of indicator i: 
formula
where ρ is internal consistency reliability (composite reliability); λi = loadings of indicator i of a latent variable; ɛi = measurement error of indicator i; j = flow index across all reflective MM.

Constructs with high CA values mean that the items within the construct have the same range and meaning. CA is most appropriately used when the items measure different substantive areas within a single construct (Manerikar & Manerikar 2015). As the water quality data were collected from different manholes of Karbala city, CA can be considered as more appropriate to evaluate internal consistency.

Indicators' reliability evaluates the extent to which a variable or a set of variables is consistent with what it intends to measure (Urbach & Ahlemann 2010). The reliability construct is independent of and calculated separately from other constructs. Conversely, convergent validity involves the degree to which individual items reflect a construct converging in comparison to items measuring different constructs (Urbach & Ahlemann 2010). Using PLS, convergent validity can be evaluated using the value of the AVE. 
formula
where λ2i = squared loadings of indicator i of a latent variable, var(εi) = squared measurement error of indicator i.

Discriminant validity is used to differentiate measures of a construct from one another. In PLS, two measures of discriminant validity are commonly used-cross loading (Chin 1998) and Fornell–Larcker's criterion (Fornell & Larcker 1981).

Validating SM can help the researcher to evaluate systematically whether the hypotheses expressed by SM are supported by the data (Urbach & Ahlemann 2010). SM can only be analysed after the MM has been validated successfully. In PLS, SM can be evaluated using coefficient of determination (R2) and path coefficients.

RESULTS AND DISCUSSION

Sewage quality depends on many factors including rainfall, leakage in sewerage system, and type of human activities. In a pilgrimage or festival city, number of population and their activities also define sewage quality. More population indicates more waste and therefore the possibility of more pollution in sewer water. Rainfall-generated runoff carries anthropogenic waste to the storm-water system, from where the polluted waste leaks to the sewerage system in Karbala. Average daily sewer discharge and maximum daily sewer discharge are defined by the floating population or rainfall in a day and at a particular period of a day, respectively, and thus indicate the possibility of sewer pollution. Anthropogenic activities of a large floating population in Karbala city center cause sewer pollution, and therefore the straight distance between the city center and sewer manholes can be used to indicate pollution amount at a manhole. Maximum daily discharge at the pump station defines amount of polluted sewer water taken out from the sewerage system, and therefore indicates sewage quality. Considering all the factors in the context of Karbala city, five parameters, namely, population, rainfall, average daily sewer discharge, maximum daily sewer discharge at the pump station, and the straight distance between the city center and sewer manholes were considered for the development of the model. Figure 1 illustrates the reflective–reflective type I hierarchical component model used in the present study (Ringle et al. 2013).

Figure 1

Hypothetical model of BOD5 and TSS concentrations.

Figure 1

Hypothetical model of BOD5 and TSS concentrations.

The model is concerned with the variance shared by directly measured observed variables and the latent unobserved variables that are theorized to explain these observed variables. A standard equation for MM is below: 
formula
where yi = manifest variable (e.g., item of rainfall, population, Qmax., distance, Qav.; see Table 1); = loading of first order of loading value (LV); = first-order LV (e.g., BOD5 or TSS); and = measurement error.

Therefore, an MM has satisfactory internal consistency reliability when the composite reliability (CR) of each construct exceeds the threshold value of 0.7. Tables 1 and 2 show that the CR of each construct for this study ranges from 0.931 to 1 and this is above the recommended threshold value of 0.7 for both BOD5 and TSS concentrations. Thus, the results indicate that the items used to represent the constructs have satisfactory internal consistency reliability.

Table 1

Evaluation of convergent validity of the BOD5 concentrations

Constructs Items Factor loading AVE CR 
BOD5 B0.886 0.658 0.931 
B0.857   
B0.800   
B0.854   
B0.811   
B0.718   
B0.741 1.000 1.000 
The straight distance of sewer manholes D 1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
Population dynamic P 0.955 0.958 0.996 
P0.986   
P0.995   
P0.969   
P0.994   
P0.955   
P0.994   
P0.984   
P0.969   
P0.985 1.000 1.000 
Max. sewer discharge of pump station Qmax.3 1.000   
Qmax.4 1.000   
Qmax.5 1.000   
Qmax.6 1.000   
Qmax.7 1.000   
Qmax.8 1.000   
Qmax.9 1.000   
Av. sewer discharge in the monitoring points Qav. 0.924 0.767 0.967 
Qav.2 0.843   
Qav.3 0.958   
Qav.4 0.754   
Qav.5 0.922   
Qav.6 0.900   
Qav.7 0.760   
Qav.8 0.927   
Qav.9 0.868   
Rainfall R 1.000 1.000 1.000 
Constructs Items Factor loading AVE CR 
BOD5 B0.886 0.658 0.931 
B0.857   
B0.800   
B0.854   
B0.811   
B0.718   
B0.741 1.000 1.000 
The straight distance of sewer manholes D 1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
Population dynamic P 0.955 0.958 0.996 
P0.986   
P0.995   
P0.969   
P0.994   
P0.955   
P0.994   
P0.984   
P0.969   
P0.985 1.000 1.000 
Max. sewer discharge of pump station Qmax.3 1.000   
Qmax.4 1.000   
Qmax.5 1.000   
Qmax.6 1.000   
Qmax.7 1.000   
Qmax.8 1.000   
Qmax.9 1.000   
Av. sewer discharge in the monitoring points Qav. 0.924 0.767 0.967 
Qav.2 0.843   
Qav.3 0.958   
Qav.4 0.754   
Qav.5 0.922   
Qav.6 0.900   
Qav.7 0.760   
Qav.8 0.927   
Qav.9 0.868   
Rainfall R 1.000 1.000 1.000 

Indicator reliability of MM was measured by examining the item loadings. MM is said to have satisfactory indicator reliability when each item's loading is at least 0.5 and is significant at least at the level of 0.05 (Hulland 1999). Based on the analysis, all items in the MM exhibited loadings exceeding 0.5, ranging from a lower bound of 0.718 to an upper bound of 1. All items are significant at the level of 0.001. Factor loadings for straight distance of sewer manholes from city center and maximum sewer discharge at pump station were found to be 1 for both BOD5 and TSS Tables 1 and 2), which indicates that these two factors strongly affect the BOD5 and TSS concentrations in Karbala city. Based on the results, it can be remarked that all items used for this study have demonstrated satisfactory indicator reliability.

Table 2

Evaluation of convergent validity of the TSS concentrations

Constructs Items Factor loading AVE CR 
TSS T 0.911 0.742 0.958 
T0.920   
T0.754   
T0.890   
T0.916   
T0.855   
T0.766   
T0.860   
The straight distance of the manholes D 1.000 1.000 1.000 
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
Population dynamic P 0.953 0.958 0.996 
P0.986   
P0.995   
P0.970   
P0.993   
P0.953   
P0.993   
P0.985   
P0.970   
P0.985 1.000 1.000 
Max. sewer discharge of pump station Qmax.3 1.000   
Qmax.4 1.000   
Qmax.5 1.000   
Qmax.6 1.000   
Qmax.7 1.000   
Qmax.8 1.000   
Qmax.9 1.000   
Av. sewer discharge in the monitoring points Qav. 0.916 0.75 0.964 
Qav.2 0.839   
Qav.3 0.945   
Qav.4 0.787   
Qav.5 0.896   
Qav.6 0.873   
Qav.7 0.778   
Qav.8 0.904   
Qav.9 0.841   
Rainfall R 1.000 1.000 1.000 
Constructs Items Factor loading AVE CR 
TSS T 0.911 0.742 0.958 
T0.920   
T0.754   
T0.890   
T0.916   
T0.855   
T0.766   
T0.860   
The straight distance of the manholes D 1.000 1.000 1.000 
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
D1.000   
Population dynamic P 0.953 0.958 0.996 
P0.986   
P0.995   
P0.970   
P0.993   
P0.953   
P0.993   
P0.985   
P0.970   
P0.985 1.000 1.000 
Max. sewer discharge of pump station Qmax.3 1.000   
Qmax.4 1.000   
Qmax.5 1.000   
Qmax.6 1.000   
Qmax.7 1.000   
Qmax.8 1.000   
Qmax.9 1.000   
Av. sewer discharge in the monitoring points Qav. 0.916 0.75 0.964 
Qav.2 0.839   
Qav.3 0.945   
Qav.4 0.787   
Qav.5 0.896   
Qav.6 0.873   
Qav.7 0.778   
Qav.8 0.904   
Qav.9 0.841   
Rainfall R 1.000 1.000 1.000 

The MM's convergent validity was assessed by examining its AVE value. Convergent validity is adequate when constructs have an AVE value of at least 0.5 or more. Tables 1 and 2 show that all constructs have AVE ranging from 0.658 to 1, more than the recommended threshold value of 0.5. This result shows an adequate convergent validity of the MM.

The MM's discriminant validity was assessed by using two measures: (1) Fornell & Larcker's (1981) criterion and (2) cross loading for BOD5 and TSS concentrations. MM has discriminant validity when (1) the square root of the AVE exceeds the correlations between the measure and all other measures and (2) the indicators' loadings are higher against their respective construct compared to other constructs. Table 3 shows that all measurement items loaded higher against their respective intended latent variable compared to other variables.

Table 3

Discriminant validity evaluation of the BOD5 and TSS concentrations (highest value in each row is marked as bold)

 B D P Qmax. Qav. R 
BOD5 
 BOD5 0.955      
 Distance 0.656 1.000     
 Population 0.744 0.758 0.998    
 Max. sewer discharge of pump station 0.610 0.564 0.793 1.000   
 Av. sewer discharge in the monitoring point 0.443 0.726 0.656 0.548 0.981  
 Rainfall depth 0.744 0.153 0.351 0.184 0.052 1.000 
TSS 
 Distance 1.000      
 Population 0.754 0.979     
 Av. sewer discharge in the monitoring point 0.735 0.685 0.866    
 Max. sewer discharge of pump station 0.564 0.793 0.558 1.000   
 Rainfall depth 0.153 0.352 0.101 0.184 1.000  
 TSS 0.504 0.625 0.401 0.481 0.867 0.861 
 B D P Qmax. Qav. R 
BOD5 
 BOD5 0.955      
 Distance 0.656 1.000     
 Population 0.744 0.758 0.998    
 Max. sewer discharge of pump station 0.610 0.564 0.793 1.000   
 Av. sewer discharge in the monitoring point 0.443 0.726 0.656 0.548 0.981  
 Rainfall depth 0.744 0.153 0.351 0.184 0.052 1.000 
TSS 
 Distance 1.000      
 Population 0.754 0.979     
 Av. sewer discharge in the monitoring point 0.735 0.685 0.866    
 Max. sewer discharge of pump station 0.564 0.793 0.558 1.000   
 Rainfall depth 0.153 0.352 0.101 0.184 1.000  
 TSS 0.504 0.625 0.401 0.481 0.867 0.861 

The second assessment of discriminant validity was to examine the indicators' loadings with respect to all construct correlations. Tables 4 and 5 show the output of cross loading between constructs and indicators. The loading clearly separates each latent variable as authorized in the conceptual model. Thus, the cross loading output confirmed that the second assessments of the MM's discriminant validity were satisfied. This study, therefore, concludes that the MM has established its discriminant validity.

Table 4

Loadings and cross loadings of the BOD5 concentrations (highest value in each row is marked as bold)

 Biochemical oxygen demand The straight distance Population dynamic Max. sewer discharge Av. sewer discharge Rainfall depth 
B0.886 0.473 0.688 0.590 0.301 0.834 
B0.857 0.486 0.638 0.520 0.307 0.589 
B0.800 0.546 0.466 0.494 0.393 0.549 
B0.854 0.611 0.679 0.396 0.464 0.559 
B0.811 0.587 0.704 0.616 0.482 0.534 
B0.718 0.426 0.515 0.471 0.405 0.503 
B0.741 0.600 0.493 0.348 0.147 0.638 
D 0.656 1.000 0.758 0.564 0.726 0.153 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.656 1.000 0.758 0.564 0.726 0.153 
D0.656 1.000 0.758 0.564 0.726 0.153 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.657 1.000 0.759 0.566 0.727 0.153 
D0.656 1.000 0.758 0.564 0.726 0.153 
P 0.753 0.917 0.955 0.752 0.732 0.280 
P0.710 0.663 0.986 0.722 0.593 0.387 
P0.754 0.802 0.995 0.823 0.683 0.326 
P0.688 0.574 0.969 0.806 0.553 0.376 
P0.756 0.822 0.994 0.799 0.692 0.326 
P0.753 0.917 0.955 0.752 0.732 0.280 
P0.756 0.822 0.994 0.799 0.692 0.326 
P0.706 0.635 0.984 0.754 0.582 0.385 
P0.688 0.574 0.969 0.806 0.553 0.376 
P0.707 0.642 0.985 0.747 0.585 0.385 
Qmax.3 0.611 0.566 0.794 1.000 0.549 0.184 
Qmax.4 0.611 0.565 0.794 1.000 0.549 0.184 
Qmax.5 0.610 0.563 0.793 1.000 0.547 0.183 
Qmax.6 0.612 0.568 0.795 1.000 0.550 0.184 
Qmax.7 0.607 0.558 0.790 1.000 0.544 0.183 
Qmax.8 0.610 0.565 0.793 1.000 0.548 0.184 
Qmax.9 0.611 0.565 0.794 1.000 0.549 0.184 
Qav. 0.508 0.843 0.698 0.586 0.924 0.034 
Qav.2 0.377 0.793 0.506 0.324 0.843 0.014 
Qav.3 0.352 0.734 0.611 0.477 0.958 −0.089 
Qav.4 0.458 0.424 0.641 0.441 0.754 0.314 
Qav.5 0.232 0.571 0.456 0.460 0.922 −0.147 
Qav.6 0.169 0.515 0.409 0.436 0.900 −0.176 
Qav.7 0.489 0.548 0.604 0.538 0.760 0.265 
Qav.8 0.257 0.544 0.451 0.465 0.927 −0.131 
Qav.9 0.179 0.456 0.371 0.408 0.868 −0.171 
R 0.744 0.153 0.351 0.184 0.052 1.000 
 Biochemical oxygen demand The straight distance Population dynamic Max. sewer discharge Av. sewer discharge Rainfall depth 
B0.886 0.473 0.688 0.590 0.301 0.834 
B0.857 0.486 0.638 0.520 0.307 0.589 
B0.800 0.546 0.466 0.494 0.393 0.549 
B0.854 0.611 0.679 0.396 0.464 0.559 
B0.811 0.587 0.704 0.616 0.482 0.534 
B0.718 0.426 0.515 0.471 0.405 0.503 
B0.741 0.600 0.493 0.348 0.147 0.638 
D 0.656 1.000 0.758 0.564 0.726 0.153 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.656 1.000 0.758 0.564 0.726 0.153 
D0.656 1.000 0.758 0.564 0.726 0.153 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.656 1.000 0.757 0.564 0.726 0.152 
D0.657 1.000 0.759 0.566 0.727 0.153 
D0.656 1.000 0.758 0.564 0.726 0.153 
P 0.753 0.917 0.955 0.752 0.732 0.280 
P0.710 0.663 0.986 0.722 0.593 0.387 
P0.754 0.802 0.995 0.823 0.683 0.326 
P0.688 0.574 0.969 0.806 0.553 0.376 
P0.756 0.822 0.994 0.799 0.692 0.326 
P0.753 0.917 0.955 0.752 0.732 0.280 
P0.756 0.822 0.994 0.799 0.692 0.326 
P0.706 0.635 0.984 0.754 0.582 0.385 
P0.688 0.574 0.969 0.806 0.553 0.376 
P0.707 0.642 0.985 0.747 0.585 0.385 
Qmax.3 0.611 0.566 0.794 1.000 0.549 0.184 
Qmax.4 0.611 0.565 0.794 1.000 0.549 0.184 
Qmax.5 0.610 0.563 0.793 1.000 0.547 0.183 
Qmax.6 0.612 0.568 0.795 1.000 0.550 0.184 
Qmax.7 0.607 0.558 0.790 1.000 0.544 0.183 
Qmax.8 0.610 0.565 0.793 1.000 0.548 0.184 
Qmax.9 0.611 0.565 0.794 1.000 0.549 0.184 
Qav. 0.508 0.843 0.698 0.586 0.924 0.034 
Qav.2 0.377 0.793 0.506 0.324 0.843 0.014 
Qav.3 0.352 0.734 0.611 0.477 0.958 −0.089 
Qav.4 0.458 0.424 0.641 0.441 0.754 0.314 
Qav.5 0.232 0.571 0.456 0.460 0.922 −0.147 
Qav.6 0.169 0.515 0.409 0.436 0.900 −0.176 
Qav.7 0.489 0.548 0.604 0.538 0.760 0.265 
Qav.8 0.257 0.544 0.451 0.465 0.927 −0.131 
Qav.9 0.179 0.456 0.371 0.408 0.868 −0.171 
R 0.744 0.153 0.351 0.184 0.052 1.000 
Table 5

Loadings and cross loadings of the TSS concentrations (highest value in each row is marked as bold)

 The straight distance Population dynamic Max. sewer discharge Av. sewer discharge Rainfall depth TSS 
D 1.000 0.754 0.564 0.735 0.153 0.505 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.153 0.504 
D1.000 0.754 0.564 0.735 0.153 0.504 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.153 0.504 
D1.000 0.756 0.566 0.735 0.153 0.505 
D1.000 0.754 0.564 0.735 0.153 0.504 
P 0.917 0.953 0.752 0.754 0.280 0.610 
P0.663 0.986 0.722 0.629 0.387 0.612 
P0.802 0.995 0.823 0.710 0.326 0.625 
P0.574 0.970 0.806 0.587 0.376 0.593 
P0.822 0.993 0.799 0.719 0.326 0.626 
P0.917 0.953 0.752 0.754 0.280 0.610 
P0.822 0.993 0.799 0.719 0.326 0.626 
P0.635 0.985 0.754 0.617 0.385 0.608 
P0.574 0.970 0.806 0.587 0.376 0.593 
P0.642 0.985 0.747 0.620 0.385 0.609 
Qmax.3 0.566 0.794 1.000 0.559 0.184 0.482 
Qmax.4 0.565 0.794 1.000 0.559 0.184 0.482 
Qmax.5 0.563 0.793 1.000 0.558 0.183 0.481 
Qmax.6 0.568 0.795 1.000 0.560 0.184 0.483 
Qmax.7 0.558 0.790 1.000 0.554 0.183 0.479 
Qmax.8 0.565 0.793 1.000 0.559 0.184 0.482 
Qmax.9 0.565 0.793 1.000 0.559 0.184 0.482 
Qav. 0.843 0.696 0.586 0.916 0.034 0.398 
Qav.2 0.793 0.503 0.324 0.839 0.014 0.309 
Qav.3 0.734 0.609 0.477 0.945 −0.089 0.246 
Qav.4 0.424 0.642 0.441 0.787 0.314 0.465 
Qav.5 0.571 0.454 0.460 0.896 −0.147 0.128 
Qav.6 0.515 0.407 0.436 0.873 −0.176 0.087 
Qav.7 0.548 0.603 0.538 0.778 0.265 0.446 
Qav.8 0.544 0.449 0.465 0.904 −0.131 0.141 
Qav.9 0.456 0.369 0.408 0.841 −0.171 0.064 
Rain 0.153 0.352 0.184 0.101 1.000 0.867 
T 0.642 0.633 0.434 0.395 0.717 0.911 
T0.338 0.544 0.319 0.501 0.833 0.920 
T0.237 0.212 0.028 0.090 0.830 0.754 
T0.434 0.643 0.442 0.473 0.797 0.890 
T0.565 0.564 0.412 0.335 0.761 0.916 
T0.482 0.762 0.452 0.546 0.739 0.855 
T0.286 0.412 0.287 0.467 0.627 0.766 
T0.383 0.342 0.230 0.406 0.706 0.860 
 The straight distance Population dynamic Max. sewer discharge Av. sewer discharge Rainfall depth TSS 
D 1.000 0.754 0.564 0.735 0.153 0.505 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.153 0.504 
D1.000 0.754 0.564 0.735 0.153 0.504 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.152 0.504 
D1.000 0.754 0.564 0.735 0.153 0.504 
D1.000 0.756 0.566 0.735 0.153 0.505 
D1.000 0.754 0.564 0.735 0.153 0.504 
P 0.917 0.953 0.752 0.754 0.280 0.610 
P0.663 0.986 0.722 0.629 0.387 0.612 
P0.802 0.995 0.823 0.710 0.326 0.625 
P0.574 0.970 0.806 0.587 0.376 0.593 
P0.822 0.993 0.799 0.719 0.326 0.626 
P0.917 0.953 0.752 0.754 0.280 0.610 
P0.822 0.993 0.799 0.719 0.326 0.626 
P0.635 0.985 0.754 0.617 0.385 0.608 
P0.574 0.970 0.806 0.587 0.376 0.593 
P0.642 0.985 0.747 0.620 0.385 0.609 
Qmax.3 0.566 0.794 1.000 0.559 0.184 0.482 
Qmax.4 0.565 0.794 1.000 0.559 0.184 0.482 
Qmax.5 0.563 0.793 1.000 0.558 0.183 0.481 
Qmax.6 0.568 0.795 1.000 0.560 0.184 0.483 
Qmax.7 0.558 0.790 1.000 0.554 0.183 0.479 
Qmax.8 0.565 0.793 1.000 0.559 0.184 0.482 
Qmax.9 0.565 0.793 1.000 0.559 0.184 0.482 
Qav. 0.843 0.696 0.586 0.916 0.034 0.398 
Qav.2 0.793 0.503 0.324 0.839 0.014 0.309 
Qav.3 0.734 0.609 0.477 0.945 −0.089 0.246 
Qav.4 0.424 0.642 0.441 0.787 0.314 0.465 
Qav.5 0.571 0.454 0.460 0.896 −0.147 0.128 
Qav.6 0.515 0.407 0.436 0.873 −0.176 0.087 
Qav.7 0.548 0.603 0.538 0.778 0.265 0.446 
Qav.8 0.544 0.449 0.465 0.904 −0.131 0.141 
Qav.9 0.456 0.369 0.408 0.841 −0.171 0.064 
Rain 0.153 0.352 0.184 0.101 1.000 0.867 
T 0.642 0.633 0.434 0.395 0.717 0.911 
T0.338 0.544 0.319 0.501 0.833 0.920 
T0.237 0.212 0.028 0.090 0.830 0.754 
T0.434 0.643 0.442 0.473 0.797 0.890 
T0.565 0.564 0.412 0.335 0.761 0.916 
T0.482 0.762 0.452 0.546 0.739 0.855 
T0.286 0.412 0.287 0.467 0.627 0.766 
T0.383 0.342 0.230 0.406 0.706 0.860 

Structural model

The evaluation process of the inner model incorporates several steps as recommended by researchers (Götz et al. 2010). A model with the existence of lower-order factors and significant inter-correlations among the factors implies the existence of at least one second-order factor. The result of SM for BOD5 and TSS concentrations is presented in Figure 2. The standard exogenous model was replaced by the following endogenous model: 
formula
where ηj = first-order factor; Г = loading of second-order LV; ξk = second-order LV (e.g., system reliability); ζj = error of first-order factors.
Figure 2

The structural model of BOD5 and TSS concentrations.

Figure 2

The structural model of BOD5 and TSS concentrations.

Table 6 shows that the results of R2 are more than 0.26 for BOD5 concentrations. The value of R2 is higher than the minimum recommended value of 0.10 (Falk & Miller 1992; Bohle et al. 2011). The effect size (ƒ2), which refers to the contribution of exogenous constructs in explaining the variance in the endogenous construct is more than 0.35 for distance, population, Qmax and rainfall. As the effect sizes of 0.02, 0.15, and 0.35 are termed small, medium, and large, respectively (Cohen 1988), it can be remarked that distance, population, Qmax and rainfall have substantial effect. The model's predictive ability was decided based on Stone–Geisser's Q2 (Geisser 1974). A Q2 value greater than zero indicates that the model has predictive ability (Fornell & Cha 1994).

Table 6

The predictive relevance and effect size results of the BOD5 and TSS concentrations

 R2
 
Q2
 
ƒ2
 
 BOD5 TSS BOD5 TSS BOD5 TSS 
Distance 0.430 0.254 0.183 0.148 0.754 0.341 
Population 0.554 0.390 0.412 0.266 1.242 0.640 
Qmax. 0.372 0.161 0.333 0.051 0.593 0.302 
Qav. sewer flow daily 0.196 0.232 0.009 0.189 0.244 0.191 
Rainfall 0.554 0.752 0.561 0.738 1.242 3.025 
 R2
 
Q2
 
ƒ2
 
 BOD5 TSS BOD5 TSS BOD5 TSS 
Distance 0.430 0.254 0.183 0.148 0.754 0.341 
Population 0.554 0.390 0.412 0.266 1.242 0.640 
Qmax. 0.372 0.161 0.333 0.051 0.593 0.302 
Qav. sewer flow daily 0.196 0.232 0.009 0.189 0.244 0.191 
Rainfall 0.554 0.752 0.561 0.738 1.242 3.025 

Similarly, for TSS concentrations (Table 6), the results of R2 are more than 0.13, higher than the minimum recommend value of 0.10. The effect size (ƒ2) for population is more than 0.35, meaning that population has a substantial effect, whereas f2 values are more than 0.15 for distance to city center, Qav, Qmax and rainfall, meaning that those factors have a moderate effect.

Since the model used was reflective and flexible between the first and second order, and the direction of causality was from construct to measuring, it was necessary to check multi-collinearity. The collinearity statistics by the contractual model are given in Table 7. The table shows that variance inflation factor (VIF) for distance from city center, Qmax, and the Qav population has achieved the requirement for reflective MM. However, VIF for rainfall was less than 3.3 for BOD5 concentrations and also for TSS concentrations.

Table 7

The collinearity statistics of contractual model of the BOD5 and TSS concentrations

 Tolerance VIF 
 BOD5 TSS BOD5 TSS 
Distance 0.324 0.234 2.675 2.675 
Population 0.190 0.461 1.690 1.690 
Qmax.PS 0.352 0.334 2.089 2.089 
Qav. sewer flow daily 0.427 0.150 2.524 2.524 
Rainfall 0.793 0.112 1.058 1.058 
 Tolerance VIF 
 BOD5 TSS BOD5 TSS 
Distance 0.324 0.234 2.675 2.675 
Population 0.190 0.461 1.690 1.690 
Qmax.PS 0.352 0.334 2.089 2.089 
Qav. sewer flow daily 0.427 0.150 2.524 2.524 
Rainfall 0.793 0.112 1.058 1.058 

Table 8 shows the loading for each item and its t-statistic values on their respective constructs for BOD5 and TSS concentrations. Based on the results, all items used for this study have demonstrated satisfactory indicator reliability, but t-value of BOD5 -> Qav and TSS -> Qav is less than 1.28 (Hair et al. 2006). Also the path coefficients of BOD5 -> Qav, TSS -> Qmax are less than 0.5. The study has proposed that higher discharges caused the instability of concentrations of BOD5 and TSS; therefore, it has been noticed that the concentrations of BOD5 and TSS decreases when the discharge of wastewater is high.

Table 8

Summary of the structural model of the BOD5 and TSS concentrations (path coefficient and t-value less than prescribed values are indicated as bold)

Hypotheses Path Path coefficient Standard error t-value P-value Decision 
H1 BOD5 → distance 0.656 0.200 3.283 0.001 Supported 
H2 BOD5 → population 0.744 0.193 3.864 0.0001 Supported 
H3 BOD5Qmax. 0.610 0.206 2.962 0.002 Supported 
H4 BOD5Qav. sewer flow daily 0.443 0.479 0.924 0.178 Non-supported 
H5 BOD5 → rainfall 0.744 0.111 6.698 0.000 Supported 
H1 TSS → distance 0.504 0.206 2.451 0.007 Supported 
H2 TSS → population 0.625 0.194 3.228 0.001 Supported 
H3 TSS → Qav. sewer flow daily 0.401 0.473 0.847 0.198 Non-supported 
H4 TSS → Qmax.PS 0.481 0.211 2.286 0.011 Supported 
H5 TSS → rainfall 0.867 0.087 9.928 0.0001 Supported 
Hypotheses Path Path coefficient Standard error t-value P-value Decision 
H1 BOD5 → distance 0.656 0.200 3.283 0.001 Supported 
H2 BOD5 → population 0.744 0.193 3.864 0.0001 Supported 
H3 BOD5Qmax. 0.610 0.206 2.962 0.002 Supported 
H4 BOD5Qav. sewer flow daily 0.443 0.479 0.924 0.178 Non-supported 
H5 BOD5 → rainfall 0.744 0.111 6.698 0.000 Supported 
H1 TSS → distance 0.504 0.206 2.451 0.007 Supported 
H2 TSS → population 0.625 0.194 3.228 0.001 Supported 
H3 TSS → Qav. sewer flow daily 0.401 0.473 0.847 0.198 Non-supported 
H4 TSS → Qmax.PS 0.481 0.211 2.286 0.011 Supported 
H5 TSS → rainfall 0.867 0.087 9.928 0.0001 Supported 

Using the factors having values higher than the standard value, two multiples linear regression models, one for BOD5 and the other for TSS concentrations, during two festival periods, namely, Muharram and Safar were developed, as shown in Figure 3.

Figure 3

Observed vs. predicted BOD5 (left) and TSS (right) concentrations during the Muharram period (only 5 days) and Safar periods (only 15 days).

Figure 3

Observed vs. predicted BOD5 (left) and TSS (right) concentrations during the Muharram period (only 5 days) and Safar periods (only 15 days).

The normality of model parameters was evaluated by estimating skewness (the symmetry of the distribution) and kurtosis values (peakedness or flatness of distribution) (Pallant 2007). The results of skewness were less than 2.58 for distance, population, Qmax, Qav, TSS, BOD5, and rainfall (Hair et al. 2006). Histograms and P–P plots of regression standardized residual for the BOD5 and TSS concentrations during Muharram and Safar periods are shown in Figure 4. The figures indicate normality in model residuals.

Figure 4

Histogram and normal P–P plot diagrams of the residuals for the BOD5 and TSS concentrations during Muharram and Safar periods.

Figure 4

Histogram and normal P–P plot diagrams of the residuals for the BOD5 and TSS concentrations during Muharram and Safar periods.

Multiple linear regression analysis produced equations relating BOD5, TSS, distance, Qmax, Qav, rainfall, and population in the study area. To quantify the influence of rainfall and population on BOD5 and TSS concentrations, data of both festival periods were merged and analyzed together. The obtained equations are given in Table 9.

Table 9

The regression equations showing relation of population and rainfall with maximum and average BOD5 and TSS concentrations in Karbala city

Concentrations of BOD5 and TSS (mg/l) Equations 
Maximum BOD5 BOD5 = 17.26 R + 0.0023 P + 0.0008 Qmax. + 0.0293 D + 0.0006 Qav. + 293 
Average BOD5 BOD5 = 12.935 R + 0.001 P + 0.002 D + 0.0001 Qmax. − (4.11×10−4) Qav. + 208 
Maximum TSS TSS = 31.13 R + 0.0016 P + 0.0028 Qmax. + 0.0624 D − 0.0008 Qmax. + 324 
Average TSS TSS = 24.68 R + 0.0005 P + 0.0017 Qmax. + 0.0193 D − 0.0008 Qmax. + 207.5 
Concentrations of BOD5 and TSS (mg/l) Equations 
Maximum BOD5 BOD5 = 17.26 R + 0.0023 P + 0.0008 Qmax. + 0.0293 D + 0.0006 Qav. + 293 
Average BOD5 BOD5 = 12.935 R + 0.001 P + 0.002 D + 0.0001 Qmax. − (4.11×10−4) Qav. + 208 
Maximum TSS TSS = 31.13 R + 0.0016 P + 0.0028 Qmax. + 0.0624 D − 0.0008 Qmax. + 324 
Average TSS TSS = 24.68 R + 0.0005 P + 0.0017 Qmax. + 0.0193 D − 0.0008 Qmax. + 207.5 

The equations given in Table 9 were used to assess the sensitivity of TSS and BOD5 to rainfall and population. The results revealed that the TSS concentration increases by 26–46 mg/l while BOD5 concentration rises by 9–19 mg/l for each increase of 1 mm rainfall. Conversely, the BOD5 concentration was found to rise by 4–17 mg/l for each increase of 10,000 population. However, the TSS concentration was not found to be very much sensitive to population.

Referring to the above results in Table 10, the factors that have the greatest impact on BOD5 and TSS concentrations are rainfall, maximum daily discharge at sewer pump station, the straight distance between the city center and the manholes, population and average daily discharge in the sewer networks, respectively.

Table 10

Standardized coefficient of BOD5 and TSS concentrations

Parameters Standardized coefficient of BOD5 Standardized coefficient of TSS 
Rainfall 0.3865 0.491 
Qmax. of pump station 0.1542 0.287 
Population 0.0545 0.107 
Distance 0.1112 0.116 
Maximum daily sewer discharge −0.0091 −0.068 
Parameters Standardized coefficient of BOD5 Standardized coefficient of TSS 
Rainfall 0.3865 0.491 
Qmax. of pump station 0.1542 0.287 
Population 0.0545 0.107 
Distance 0.1112 0.116 
Maximum daily sewer discharge −0.0091 −0.068 

Verification model

The path coefficient of different parameters shown in Table 8 matches very well with the standardized coefficient values presented in Table 10. Rainfall, population, daily maximum sewer discharge, maximum discharge at pump station, and distance of manhole from city center were found to have significant influence on both BOD5 and TSS. Among those parameters, rainfall, population, maximum discharge at pump station, and distance of manhole from city center have positive influence on both BOD5 and TSS. As those input parameters increase, BOD5 and TSS in the sewer also increase. Conversely, maximum daily sewer discharge has negative influence on both BOD5 and TSS. Among these parameters, rainfall (β = 0.3865, α < 0.01) has the highest influence and maximum daily sewer discharge (β = −0091, α < 0.01) has the lowest influence on BOD5. A similar result was found for TSS. Table 8 shows that the path between daily average sewer discharge and BOD5 or TSS is not significant. The results of multiple linear regression analysis (MLRA) also support the findings. Influence of daily average sewer discharge on BOD5 or TSS was found non-significant in MLRA.

Assessment of parameter sensitivity

The equations in Table 9 show that BOD5 and TSS concentrations in the study area increase sharply with the increase in rainfall. However, increase in population is not found to affect TSS concentrations very much. The BOD5 and TSS concentrations in sewage with 95% confidence interval are given in Table 11.

Table 11

Sewage quality values for BOD5 and TSS concentrations with 95% confidence interval

Parameters Mean 95% confidence interval 
BOD5 (mg/l) 320 172–467a 
TSS (mg/l) 425.5 201–650a 
Parameters Mean 95% confidence interval 
BOD5 (mg/l) 320 172–467a 
TSS (mg/l) 425.5 201–650a 

ap < 0.01.

The results showed that for a similar environment, the TSS loads in the sampling areas increase more compared to BOD5. The mean concentration of BOD5 and TSS were 320 mg/l and 467 mg/l, respectively. The concentration change of BOD5 was found in the range from 222 to 426 mg/l and for TSS between 184 and 316 mg/l for the confidence interval of 95% (Table 11). Typical municipal wastewater characterization adopted by Rossman (2010) is shown in Table 12. Comparing the results given in Table 11 with the prescribed values given in Table 12, it can be remarked that the BOD5 in sewer water is higher than the medium value, and the TSS higher than the highest limit prescribed during festival periods (Table 12).

Table 12

Typical municipal wastewater characterization adopted by Rossman (2010) 

Parameter Low Medium High 
BOD 230 350 560 
TSS 250 400 600 
Parameter Low Medium High 
BOD 230 350 560 
TSS 250 400 600 

Muserere et al. (2014) characterized raw sewage in order to evaluate the performance of the primary treatment system at Firle sewage treatment works in Harare, Zimbabwe. They reported that raw sewage of the Firle works is generally in the low to medium range for most parameters, which means that the sewage is relatively easy to treat by biological processes. In the case of Karbala city, the raw sewage is found to be in the medium to high range and over the higher limit for BOD5 and TSS, respectively during festival periods. This indicates that quality of sewer water drastically deteriorates during the festival period due to the huge floating population.

CONCLUSIONS

The analysis of confirmatory factor and the inter-relationship among variables are more reliable to construct a reliable model. Therefore, models were developed in the present study using those processes. The results show that both the BOD5 and TSS concentration in sewer have positive correlations with rainfall, population, maximum sewer discharge, discharge at pump station, and distance of the sampling point from city center. The results showed that the concentrations of BOD5 and TSS increase significantly during rainfall. The modeling provided useful results for comparing non-point pollutant loadings originating from various areas of Karbala city center. It was observed that the TSS concentration increases by 26–46 mg/l while BOD5 concentration rises by 9–19 mg/l for each 1 mm increase of rainfall. Conversely, the BOD5 concentration in sewage increases by 4–17 mg/l for each increase of 10,000 population. It can be expected that the finding of the study will help different stakeholders such as the municipal authority, sewer network designers, sanitary managers, and planning authorities to understand the impacts of a floating population and rainfall on sewage quality in order to adopt necessary mitigation measures.

ACKNOWLEDGEMENTS

We gratefully acknowledge the financial support of the Ministry of Municipalities and Public Works, and the Ministry of Higher Education in Iraq. We are also grateful to the Ministry of Education-Malaysia and Universiti Teknologi of Malaysia (UTM) for supporting this research through FRGS research grant no. 4F541.

REFERENCES

REFERENCES
Abdellatif
M.
Atherton
W.
Alkhaddar
R. M.
Osman
Y. Z.
2014
Quantitative assessment of sewer overflow performance with climate change in northwest England
.
Hydrological Sciences Journal
60
(
4
),
636
650
.
Agustsson
J.
Akermann
O.
Barry
D. A.
Rossi
L.
2014
Non-contact assessment of COD and turbidity concentrations in water using diffuse reflectance UV-Vis spectroscopy
.
Environmental Science: Processes and Impacts
16
(
8
),
1897
1902
.
Asyraf
M.
Afthanorhan
B. W.
2013
A comparison of partial least square structural equation modeling (PLS-SEM) and covariance based structural equation modeling (CB-SEM) for confirmatory factor analysis
.
International Journal of Engineering Science and Innovative Technology
2
(
5
),
198
205
.
Aziz
M. A.
Imteaz
M. A.
Choudhury
T. A.
Phillips
D. I.
2011
Artificial neural networks for the prediction of the trapping efficiency of a new sewer overflow screening device
.
19th International Congress on Modelling and Simulation
,
Perth
.
Bagozzi
R. P.
Phillips
L. W.
1982
Representing and testing organizational theories: a holistic construal
.
Administrative Science Quarterly
27
(
3
),
459
489
.
Baird
C.
Jennings
M.
Ockerman
D.
Dybala
T.
1996
Characterization of nonpoint sources and loadings to the Corpus Christi Bay National Estuary Program study area
.
Final report (No. PB--96-196845/XAB)
.
Natural Resources Conservation Service
,
Washington, DC,
USA
.
Bohle
P.
Willaby
F.
Quinlan
M.
McNamara
M.
2011
Flexible work in call centres: working hours, work-life conflict and health
.
Applied Ergonomics
42
,
219
224
.
Burns
D.
Vitvar
T.
McDonnell
J.
Hassett
J.
Duncan
J.
Kendall
C.
2005
Effects of suburban development on runoff generation in the Croton River basin, New York, USA
.
Journal of Hydrology
311
(
1
),
266
281
.
Byrne
B. M.
2013
Structural Equation Modeling with AMOS: Basic Concepts, Applications, and Programming
.
Routledge, London
.
Chin
W. W.
1998
Issues and opinions on structural equation modeling
.
MIS Quarterly
22
(
1
),
7
26
.
Cohen
J.
1988
Statistical Power Analysis for the Behavioral Sciences
,
2nd edn
.
Academic Press, London
.
Falk
R. F.
Miller
N. B.
1992
A Primer for Soft Modeling
.
University of Akron Press, Akron, OH, USA
.
Ferreira
A. P.
de Lourdes
C.
da Cunha
N.
2005
Anthropic pollution in aquatic environment: development of a caffeine indicator
.
International Journal of Environmental Health Research
15
(
4
),
303
311
.
Fornell
C.
Cha
J.
1994
Partial least squares
. In:
Advanced Methods of Marketing Research
(
Bagozzi
R. P.
, ed.).
Blackwell
,
Cambridge, MA, USA
, pp.
52
78
.
Fortier
C.
Mailhot
A.
2014
Climate change impact on combined sewer overflows
.
Journal of Water Resources Planning and Management
141
(
5
),
art. 04014073
.
Gefen
D.
Straub
D. W.
2000
The relative importance of perceived ease of use in IS adoption: a study of e-commerce adoption
.
Journal of the Association for Information Systems
1
(
1
),
art. 8
,
30 pp.
Gladstone
D. L.
2005
From Pilgrimage to Package Tour: Travel and Tourism in the Third World
.
Routledge
,
New York
.
Götz
O.
Liehr-Gobbers
K.
Krafft
M.
2010
Evaluation of structural equation models using the partial least squares (PLS) approach
. In:
Handbook of Partial Least Squares
(V. E. Vinzi, W. W. Chin, J. Henseler & H. Wang, eds)
,
Springer, Berlin
, pp.
691
711
.
Hair
J. F.
Black
W. C.
Babin
B. J.
Anderson
R. E.
Tatham
R. L.
2006
Multivariate Data Analysis
(Vol.
6
).
Pearson Prentice Hall
,
Upper Saddle River, NJ, USA
.
Hair
J. F.
Sarstedt
M.
Ringle
C. M.
Mena
J. A.
2012
An assessment of the use of partial least squares structural equation modeling in marketing research
.
Journal of the Academy of Marketing Science
40
(
3
),
414
433
.
Henseler
J.
Ringle
C. M.
Sinkovics
R. R.
2009
The use of partial least square path modeling in international marketing
.
Advances in International Marketing
20
,
277
319
.
Huong
H. T. L.
Pathirana
A.
2013
Urbanization and climate change impacts on future urban flooding in Can Tho city, Vietnam
.
Hydrology and Earth System Sciences
17
,
379
394
.
Hurlimann
A.
Hemphill
E.
McKay
J.
Geursen
G.
2008
Establishing components of community satisfaction with recycled water use through a structural equation model
.
Journal of Environmental Management
88
(
4
),
1221
1232
.
Hussein
A. O.
Shahid
S.
Basim
K. N.
Chelliapan
S.
2015
Modelling stormwater quality of an arid urban catchment
.
Applied Mechanics and Materials
735
,
215
219
.
IDRE
2015
.
Ji
D.
Xi
B.
Su
J.
Huo
S.
He
L.
Liu
H.
Yang
Q.
2013
A model to determine the lake nutrient standards for drinking water sources in Yunnan-Guizhou Plateau Ecoregion, China
.
Journal of Environmental Sciences
25
(
9
),
1773
1783
.
Kock Rasmussen
E.
Svenstrup Petersen
O.
Thompson
J. R.
Flower
R. J.
Ahmed
M. H.
2009
Hydrodynamic-ecological model analyses of the water quality of Lake Manzala (Nile Delta, Northern Egypt)
.
Hydrobiologia
622
,
195
220
.
Lee
S.
Yoon
C.
Jung
K.
Hwang
H.
2010
Comparative evaluation of runoff and water quality using HSPF and SWMM
.
Water Science & Technology
62
(
6
),
1401
1409
.
Lewis
B. R.
Templeton
G. F.
Byrd
T. A.
2005
A methodology for construct development in MIS research
.
European Journal of Information Systems
14
(
4
),
388
400
.
Llopart-Mascaró
A.
Farreny
R.
Gabarrell
X.
Rieradevall
J.
Gil
A.
Martínez
M.
Puertas
J.
Suárez
J.
del Rio
J.
Paraira
M.
2014
Storm tank against combined sewer overflow: operation strategies to minimize discharges impact to receiving waters
.
Urban Water Journal
12
(
3
),
219
228
.
Manerikar
V.
Manerikar
S.
2015
Cronbach's alpha
.
aWEshkar
XIX
(
1
),
117
119
.
Maruéjouls
T.
Lessard
P.
Vanrolleghem
P. A.
2014
Calibration and validation of a dynamic model for water quality in combined sewer retention tanks
.
Urban Water Journal
11
(
8
),
668
677
.
Mouri
G.
Shinoda
S.
Oki
T.
2012
Assessing environmental improvement options from a water quality perspective for an urban–rural catchment
.
Environmental Modelling & Software
32
,
16
26
.
Muserere
S. T.
Hoko
Z.
Nhapi
I.
2015
Fractionation of wastewater characteristics for modelling of Firle Sewage Treatment Works, Harare, Zimbabwe
.
Physics and Chemistry of the Earth, Parts A/B/C
76-78
,
124
133
.
Obaid
H. A.
Shamsuddin
S.
Basim
K. N.
Shreeshivadasan
C.
2014
Modeling sewer overflow of a city with a large floating population
.
Hydrology: Current Research
5
(
171
),
2
7
.
Pallant
J. F.
2007
SPSS Survival Manual: A Step-By-Step Guide to Data Analysis with SPSS
.
McGraw-Hill Education, Maidenhead, UK
Pantsar-Kallio
M.
Mujunen
S. P.
Hatzimihalis
G.
Koutoufides
P.
Minkkinen
P.
Wilkie
P. J.
Connor
M. A.
1999
Multivariate data analysis of key pollutants in sewage samples: a case study
.
Analytica Chimica Acta
393
(
1
),
181
191
.
Pathiratne
K. A. S.
De Silva
O. C. P.
Hehemann
D.
Atkinson
I.
Wei
R.
2007
Occurrence and distribution of polycyclic aromatic hydrocarbons (PAHs) in Bolgoda and Beira Lakes, Sri Lanka
.
Bulletin of Environmental Contamination and Toxicology
79
(
2
),
135
140
.
Ringle
C. M.
Sarstedt
M.
Schlittgen
R.
Taylor
C. R.
2013
PLS path modeling and evolutionary segmentation
.
Journal of Business Research
66
(
9
),
1318
1324
.
Rinschede
G.
1992
Forms of religious tourism
.
Annals of Tourism Research
19
,
51
67
.
Rossman
L. A.
2010
Storm water management model user's manual, version 5.0
.
National Risk Management Research Laboratory, Office of Research and Development, US Environmental Protection Agency, Cincinnati, OH, USA
.
Sharpley
R.
Sundaram
P.
2005
Tourism: a sacred journey? The case of ashram tourism, India
.
International Journal of Tourism Research
7
,
161
171
.
Shinde
K.
2012
Policy, planning, and management for religious tourism in Indian pilgrimage sites
.
Journal of Policy Research in Tourism, Leisure and Events
4
(
3
),
277
301
.
United Nations
2005
Millennium Development Goals Report 2005
.
UN
,
UGA
. .
Urbach
N.
Ahlemann
F.
2010
Structural equation modeling in information systems research using partial least squares
.
Journal of Information Technology Theory and Application
11
(
2
),
5
40
.
USEPA
2008
Analytical methods approved for compliance monitoring under the long term 2 enhanced surface water treatment rule
.
US Environmental Protection Agency
,
Washington, DC
. .
VandeWalle
J. L.
Goetz
G. W.
Huse
S. M.
Morrison
H. G.
Sogin
M. L.
Hoffmann
R. G.
Yan
K.
McLellan
S. L.
2012
Acinetobacter, Aeromonas and Trichococcus populations dominate the microbial community within urban sewer infrastructure
.
Environmental Microbiology
14
(
9
),
2538
2552
.
Vazquez-Prokopec
G. M.
Eng
J. L. V.
Kelly
R.
Mead
D. G.
Kolhe
P.
Howgate
J.
Kitron
U.
Burkot
T. R.
2010
The risk of West Nile virus infection is associated with combined sewer overflow streams in urban Atlanta, Georgia, USA
.
Environmental Health Perspectives
118
(
10
),
1382
1388
.
Vijayanand
S.
2012
Socio-economic impacts in pilgrimage tourism
.
Zenith International Journal of Multidisciplinary Research
2
(
1
),
329
343
.
Zeferino
J. A.
Antunes
A. P.
Cunha
M. C.
2012
Regional wastewater system planning under population dynamics uncertainty
.
Journal of Water Resources Planning and Management
140
(
3
),
322
331
.