Development of multiple linear regression models as predictive tools for fecal indicator concentrations in a stretch of the lower Lahn River, Germany

Integrating quantitative microbiological risk assessment and disability-adjusted life years to evaluate the effects of urbanization on health risks for river recreationists

Urban rivers provide an excellent opportunity for water recreation. This study probabilistically assessed health risks associated with water recreation in urban rivers in the Bitan Scenic Area, Taiwan, by employing quantitative microbial risk assessment and disability-adjusted life years (DALYs). Moreover, the effects of urbanization on the health risks of river recreation induced by waterborne pathogenic Escherichia coli (E. coli) were investigated. First, data on river E. coli levels were collected in both the Bitan Scenic Area and the upstream river section, and model parameters were obtained through a questionnaire administered to river recreationists. Monte Carlo simulation was then employed to address parameter uncertainty. Finally, DALYs were calculated to quantify the cumulative effects in terms of potential life lost and years lived with disability. The results indicated that the 90 % confidence intervals for the disease burden (DB) were 0.2-74.1 × 10-6, 0.01-94.0 × 10-6, and 0.3-128.9 × 10-6 DALY per person per year (pppy) for canoeing, swimming, and fishing, respectively, in the Bitan Scenic Area. Furthermore, urbanization near the Bitan Scenic Area approximately doubled the DB risks to river recreationists in upstream rural areas. At the 95th percentile, the DB risks exceeded the tolerances recommended by the World Health Organization (1 × 10-6) or U.S. Environmental Protection Agency (1 × 10-4). The findings suggest that the simultaneous implementation of effluent sewer systems and best management practices can reduce health risks to river recreationists by at least half, reducing the DALY levels below 1 × 10-4 or even 1 × 10-5 pppy.

Seasonal assessment of risks to canoeists' health in a Taiwanese recreational river

Canoeing is the most favorite recreational activity in several Taiwanese rivers. However, river water frequently contains elevated levels of pathogenic Escherichia coli, which has adverse effects on human health. This study adopted a quantitative microbial risk assessment to analyze seasonal risks to canoeists' health in the Dongshan River, Taiwan. First, river E. coli concentrations were statistically analyzed to determine the seasonal distributions. The exposure duration (ED) was determined by field observations. To propagate the parametric uncertainty, Monte Carlo simulation was employed to model the probability distributions of seasonal pathogenic E. coli levels, ingestion rates, and ED for athletes. Finally, the beta-Poisson dose-response model was implemented to determine seasonal health risks for canoeists. The study results indicated that the health risks in infection probability ranged from 0.5 × 10-3 to 8.8 × 10-3 illnesses/person/day for tourists and 1.2 × 10-3 to 7.7 × 10-3 illnesses/person/day for athletes. The health risks in the Lizejian Bridge area for tourists exceeded an acceptable level suggested by the U.S. Environmental Protection Agency, 8 × 10-3 illnesses/person/day, in spring for an ED of 2 h/day, and the health risks for tourists and athletes approached this level in spring and winter for an ED exceeding or equaling 1.5 h/day. According to sensitivity analysis, the geometric standard deviation of river E. coli levels was the most sensitive parameter affecting seasonal risks to canoeists' health. To protect canoeists' health, effluent sewer systems, best management practices, and total maximum daily loads should be promptly implemented in this watershed.

Development of AI-based hybrid soft computing models for prediction of critical river water quality indicators

Prediction of river water quality indicators (RWQIs) using artificial intelligence (AI)-based hybrid soft computing modeling techniques could provide essential predictions required for efficient river health planning and management. The study described the development of a novel AI-based relative weighted ensemble (AIRWE) hybrid model for predicting critical RWQIs, i.e., biochemical oxygen demand (BOD) and total coliform (TC). The study involved comprehensive water quality (WQ) monitoring from 30 locations along the Damodar River to establish the baseline data and delineate the WQ. The representative input features showing a strong association with BOD and TC were identified using Spearman's rank-coupled orthogonal linear transformation (SOT). The relative weighted ensemble (RWE) method was applied to determine the relative weights for base learners in the AIRWE model. The statistical analysis of the developed model revealed that it was most efficient and accurate for predicting BOD (R2, 0.97; RMSE, 0.06; MAE, 0.04) and TC (R2, 0.98; RMSE, 0.06; MAE, 0.05) over the traditional techniques. The tstat (BOD 0.02 and TC 0.47) was lesser than tcrit (1.672), confirming its unbiased predictions. The SOT technique removed the data noise and multicollinearity, whereas RWE curtailed the individual model's limitations and predicted more reliable results. The model resulted 97% accuracy with high precision (96%) in classifying the river water quality for various end uses. The study describes a novel approach for researchers, scientists, and decision-makers for modeling and predicting various environmental attributes.

Evaluation of the Role of Self-cleaning Capacity on Marine Environmental Carrying Capacity: A Case of Ganh Rai Bay, Vietnam

Economic activities are constantly increasing in the southern key economic region (SKER), especially in Ho Chi Minh City (HCMC), which leads to the influx of large amounts of wastewater from this region into Ganh Rai Bay (GRB). The problem of assessing the marine environmental carrying capacity (MECC) of coastal areas is urgent, and the role of self-cleaning must be elucidated. Four typical pollution parameters were selected: ammonium (NH4+), biological oxygen demand (BOD), phosphate (PO43-), and coliforms. The study aims to propose a framework to assess the impact of the role of self-cleaning on MECC and to apply the proposed framework to GRB as a case study. A series of models were used to simulate hydrodynamics, and an advection-diffusion model with an ecological parameter set was used for water quality modelling. The land-ocean interactions in the coastal zone model were used to calculate the GRB and East Sea retention time. Finally, a multiple linear regression model was used to clarify the relationship between the MECC and self-cleaning factors. Calculation results show that the self-cleaning factor increased the MECCAmmonium by 60.30% in the dry season and 22.75% in the wet season; similar to MECCBOD, MECCPhosphate increased by 5.26%, 0.21% (dry season), and 11.04%, 0.72% (wet season), respectively. MECCCColiforms in the dry season increased by 14.83%; in the wet season, MECCColiforms doubled. The results provide medium-and long-term solutions to improve the water quality of the GRB, especially the selection of activities that conserve the ecological system and improve the self-cleaning capacity of the bay.

Zerovalent Iron/Cu Combined Degradation of Halogenated Disinfection Byproducts and Quantitative Structure-Activity Relationship Modeling

Previous studies have reported that zerovalent iron (ZVI) can reduce several aliphatic groups of disinfection byproducts (DBPs) (e.g., haloacetic acids and haloacetamides) effectively, and the removal efficiency can be significantly improved by metallic copper. Information regarding ZVI/Cu combined degradation of different types of halogenated DBPs can help understand the fate of overall DBPs in drinking water distribution and storage systems consisting of unlined cast iron/copper pipes and related potential control strategies. In this study, we found that, besides aliphatic DBPs, many groups of new emerging aromatic DBPs formed in chlorinated and chloraminated drinking water can be effectively degraded by ZVI/Cu; meanwhile, total organic halogen and total ion intensity were reduced significantly after treatment. Moreover, a robust quantitative structure-activity relationship model was developed and validated based on the ZVI/Cu combined degradation rate constants of 14 typical aromatic DBPs; it can predict the degradation rate constants of other aromatic DBPs for screening and comparative purposes, and the optimized descriptors indicate that DBPs possessing a lower value of the lowest unoccupied molecular orbital energy and a higher value of dipole moment tend to present higher degradation rate constants. In addition, toxicity data of 47 DBPs (belonging to 18 groups) were predicted by two previously established toxicity models, demonstrating that, although most DBPs exhibit higher toxicity than their dehalogenated products, some DBPs show lower toxicity than their lowly halogenated analogs.

Comparative cytotoxicity analyses of disinfection byproducts in drinking water using dimensionless parameter scaling method: Effect of halogen substitution type and number

Up to date, over 700 disinfection byproducts (DBPs) have been detected and identified in drinking water. It has been recognized that cytotoxicity of DBPs varied significantly among groups. Even within the same group, cytotoxicity of different DBP species was also different due to different halogen substitution types and numbers. However, it is still difficult to quantitatively determine the inter-group cytotoxicity relationships of DBPs under the effect of halogen substitution in different cell lines, especially when a large number of DBP groups and multiple cytotoxicity cell lines are involved. In this study, a powerful dimensionless parameter scaling method was adopted to quantitatively determine the relationship of halogen substitution and the cytotoxicity of various DBP groups in three cell lines (i.e., the human breast carcinoma (MVLN), Chinese hamster ovary (CHO), and human hepatoma (Hep G2) cell cytotoxicity) with no need to consider their absolute values and other influences. By introducing the dimensionless parameters D_{x-orn-species}^cellline and D¯_{x-orn-species}^cellline, as well as their corresponding linear regression equation coefficients k_typeornumber^cellline and k¯_typeornumber^cellline, the strength and trend of halogen substitution influences on the relative cytotoxic potency could be determined. It was found that the effect of halogen substitution type and number on the cytotoxicity of DBPs followed the same patterns in the three cell lines. The CHO cell cytotoxicity was the most sensitive cell line to evaluate the effect of halogen substitution on the aliphatic DBPs, whereas the MVLN cell cytotoxicity was the most sensitive cell line to evaluate the effect of halogen substitution on the cyclic DBPs. Notably, seven quantitative structure activity relationship (QSAR) models were established, which could not only predict the cytotoxicity data of DBPs, but also help to explain and verify the patterns of halogen substitution effect on cytotoxicity of DBPs.

Predicting Fecal Indicator Bacteria Using Spatial Stream Network Models in A Mixed-Land-Use Suburban Watershed in New Jersey, USA

Good water quality safeguards public health and provides economic benefits through recreational opportunities for people in urban and suburban environments. However, expanding impervious areas and poorly managed sanitary infrastructures result in elevated concentrations of fecal indicator bacteria and waterborne pathogens in adjacent waterways and increased waterborne illness risk. Watershed characteristics, such as urban land, are often associated with impaired microbial water quality. Within the proximity of the New York-New Jersey-Pennsylvania metropolitan area, the Musconetcong River has been listed in the Clean Water Act's 303 (d) List of Water Quality-Limited Waters due to high concentrations of fecal indicator bacteria (FIB). In this study, we aimed to apply spatial stream network (SSN) models to associate key land use variables with E. coli as an FIB in the suburban mixed-land-use Musconetcong River watershed in the northwestern New Jersey. The SSN models explicitly account for spatial autocorrelation in stream networks and have been widely utilized to identify watershed attributes linked to deteriorated water quality indicators. Surface water samples were collected from the five mainstem and six tributary sites along the middle section of the Musconetcong River from May to October 2018. The log₁₀ geometric means of E. coli concentrations for all sampling dates and during storm events were derived as response variables for the SSN modeling, respectively. A nonspatial model based on an ordinary least square regression and two spatial models based on Euclidean and stream distance were constructed to incorporate four upstream watershed attributes as explanatory variables, including urban, pasture, forest, and wetland. The results indicate that upstream urban land was positively and significantly associated with the log₁₀ geometric mean concentrations of E. coli for all sampling cases and during storm events, respectively (p < 0.05). Prediction of E. coli concentrations by SSN models identified potential hot spots prone to water quality deterioration. The results emphasize that anthropogenic sources were the main threats to microbial water quality in the suburban Musconetcong River watershed. The SSN modeling approaches from this study can serve as a novel microbial water quality modeling framework for other watersheds to identify key land use stressors to guide future urban and suburban water quality restoration directions in the USA and beyond.

Deep learning-based prediction of effluent quality of a constructed wetland

Data-driven approaches that make timely predictions about pollutant concentrations in the effluent of constructed wetlands are essential for improving the treatment performance of constructed wetlands. However, the effect of the meteorological condition and flow changes in a real scenario are generally neglected in water quality prediction. To address this problem, in this study, we propose an approach based on multi-source data fusion that considers the following indicators: water quality indicators, water quantity indicators, and meteorological indicators. In this study, we establish four representative methods to simultaneously predict the concentrations of three representative pollutants in the effluent of a practical large-scale constructed wetland: (1) multiple linear regression; (2) backpropagation neural network (BPNN); (3) genetic algorithm combined with the BPNN to solve the local minima problem; and (4) long short-term memory (LSTM) neural network to consider the influence of past results on the present. The results suggest that the LSTM-predicting model performed considerably better than the other deep neural network-based model or linear method, with a satisfactory R². Additionally, given the huge fluctuation of different pollutant concentrations in the effluent, we used a moving average method to smooth the original data, which successfully improved the accuracy of traditional neural networks and hybrid neural networks. The results of this study indicate that the hybrid modeling concept that combines intelligent and scientific data preprocessing methods with deep learning algorithms is a feasible approach for forecasting water quality in the effluent of actual engineering.

Aquifer vulnerability assessment for fecal coliform bacteria using multi-threshold logistic regression

Assessing aquifer vulnerability is crucial for preventing groundwater pollution. In this study, aquifer vulnerability to fecal coliform (FC) pollution was assessed using auxiliary environmental data in the Pingtung Plain, Taiwan. Moreover, key environmental factors inducing different fecal pollution levels were determined. First, 23 explanatory variables on land uses, population density, livestock and poultry densities, sanitary condition, antecedent precipitation, groundwater quality, aquifer characteristics, and subsurface hydrology were obtained using geographic information systems in 2014. As dependent variables, groundwater FCs were also simultaneously obtained. Then, multi-threshold logistic regression (LR) was adopted to model aquifer vulnerability assessment after cross validation. The thresholds of aquifer vulnerability causing risks of incidental ingestion were analyzed by risk assessment. Risks to human health were acceptable for a low-level threshold and exceeded the acceptable level for medium- and high-level thresholds when residents incidentally ingested FC-polluted groundwater. Finally, key environmental factors inducing low, medium, and high levels of groundwater FC pollution were characterized. The key environmental factors for the LR with low- and medium-level thresholds were sand and gravel soil textures of unsaturated aquifers and antecedent 3-day cumulative precipitation, and those for the LR with high-level thresholds were chicken farming, urban land use, and ratio of tap water use. Thus, the multi-threshold LR indicated that environmental factors must be ranked for assessing aquifer vulnerability.

Risk assessment of Cryptosporidium intake in drinking water treatment plant by a combination of predictive models and event-tree and fault-tree techniques

Risk-informed decision making permits a more effective water safety management. In this framework, this article introduces the rationale and proposes a new approach to carry out a quantitative risk assessment along the water chain, from river source to tap water, by integrating predictive modelling combined with event-tree and fault-tree techniques. The model developed by this approach could not only account for normal but also for abnormal process conditions in the water treatment plant, as well as assess the real impact of the applied safety controls, such as turbidity control. A sensitivity study was conducted to determine the effect of considering a typical drinking water treatment plant (DWTP), i.e. coagulation, sedimentation and filtration with two turbidity controls (on intake and after filtration) on the risk of infection due to exposure to Cryptosporidium in tap water. The results showed that, with the current effectiveness of turbidity reduction in the DWTP, the first control did not minimise the annual risk of Cryptosporidium infection (3.6E-04) and only limiting turbidity after filtration to below 0.01NTU provided a clear reduction in risk (7.7E-05) at the cost of rejecting 60 % of the water after the control. The lowest risk was found when turbidity reduction was set at 4 logs (8.48E-06), although this means that the effectiveness of turbidity reduction should be greatly improved. It was therefore concluded that supplementing the current treatment with alternative barriers such as UV or ozone disinfection and/or implementing direct control of Cryptosporidium concentration should be considered.

Observations and Correlations from a 3-Year Study of Fecal Indicator Bacteria in the Mohawk River in Upstate NY

Fecal indicator bacteria (FIB), such as E. coli and Enterococci, are used to indicate the potential of fecal contamination in waterways. One known source of FIB in urbanized areas is the occurrence of combined sewer overflows (CSOs). To explore the impact of CSOs on local water quality and FIB presence, sampling was conducted during the summers of 2017–2019 of two cities, one with CSOs and one without, on the Mohawk River in upstate New York, USA. Sampling included in situ physiochemical parameters of pH, temperature, and dissolved oxygen and laboratory tests for E. coli, Enterococci, nitrates, and total organic carbon (TOC). Correlations between parameters were explored using the Wilcoxon rank sum test and Spearman’s Rank correlation with and without considerations of site and city location. Overall, positive correlations between FIB and rainfall were identified in one city but were less significant in the other, suggesting a buffering of FIB concentrations likely due to inflow contributions from a reservoir. Samples collected downstream from an active CSO reached the detection limit of the FIB tests, demonstrating a 2-log or greater increase in FIB concentrations from dry weather conditions. The city with CSOs demonstrated greater FIB concentrations, which are likely a combination of greater urban runoff, CSOs, and the potential resuspension of sediment during high flow events. Due to the widespread presence of FIB in the region, future research includes utilizing microbial source tracking to identify the sources of contamination in the region.

An improved adaptive neuro fuzzy inference system model using conjoined metaheuristic algorithms for electrical conductivity prediction

Precise prediction of water quality parameters plays a significant role in making an early alert of water pollution and making better decisions for the management of water resources. As one of the influential indicative parameters, electrical conductivity (EC) has a crucial role in calculating the proportion of mineralization. In this study, the integration of an adaptive hybrid of differential evolution and particle swarm optimization (A-DEPSO) with adaptive neuro fuzzy inference system (ANFIS) model is adopted for EC prediction. The A-DEPSO method uses unique mutation and crossover processes to correspondingly boost global and local search mechanisms. It also uses a refreshing operator to prevent the solution from being caught inside the local optimal solutions. This study uses A-DEPSO optimizer for ANFIS training phase to eliminate defects and predict accurately the EC water quality parameter every month at the Maroon River in the southwest of Iran. Accordingly, the recorded dataset originated from the Tange-Takab station from 1980 to 2016 was operated to develop the ANFIS-A-DEPSO model. Besides, the wavelet analysis was jointed to the proposed algorithm in which the original time series of EC was disintegrated into the sub-time series through two mother wavelets to boost the prediction certainty. In the following, the comparison between statistical metrics of the standalone ANFIS, least-square support vector machine (LSSVM), multivariate adaptive regression spline (MARS), generalized regression neural network (GRNN), wavelet-LSSVM (WLSSVM), wavelet-MARS (W-MARS), wavelet-ANFIS (W-ANFIS) and wavelet-GRNN (W-GRNN) models was implemented. As a result, it was apparent that not only was the W-ANFIS-A-DEPSO model able to rise remarkably the EC prediction certainty, but W-ANFIS-A-DEPSO (R = 0.988, RMSE = 53.841, and PI = 0.485) also had the edge over other models with Dmey mother in terms of EC prediction. Moreover, the W-ANFIS-A-DEPSO can improve the RMSE compared to the standalone ANFIS-DEPSO model, accounting for 80%. Hence, this model can create a closer approximation of EC value through W-ANFIS-A-DEPSO model, which is likely to act as a promising procedure to simulate the prediction of EC data.

Learning hierarchical Bayesian networks to assess the interaction effects of controlling factors on spatiotemporal patterns of fecal pollution in streams

The dynamics of fecal indicator bacteria, such as fecal coliforms (FC) in streams, are influenced by the interactions of a myriad of factors. To predict complex spatiotemporal patterns of FC in streams and assess the relative importance of numerous controlling factors, the adoption of a hierarchical Bayesian network (HBN) was proposed in this study. By introducing latent variables correlated to the observed variables into a Bayesian network, the HBN can represent causal relationships among a large set of variables with a multilevel hierarchy. The study area encompasses 215 sites across the watersheds of the four major rivers in South Korea. The monitoring data collected during the 2012-2019 period included 32 input variables pertaining to meteorology, geography, soil characteristics, land cover, urbanization index, livestock density, and point sources. As model endpoints, the exceedance probability of the FC standard concentration as well as two pollution characteristics (i.e., pollution degree and type), derived from FC load duration curves were used. The probability of exceeding an FC threshold value (200 CFU/100 mL) showed spatiotemporal variations, whereas pollution degree and type showed spatial variations that represent long-term severity and relative dominance of nonpoint and point source fecal pollution, respectively. The conceptual model was validated using structural equation modeling to develop the HBN. The results demonstrate that the HBN effectively simplified the model structure, while showing strong model performance (AUC = 0.81, accuracy = 0.74). The results of the sensitivity analysis indicate that land cover is the most important factor in predicting the probability of exceedance and pollution degree, whereas the urbanization index explains most of the variability in pollution type. Furthermore, the results of the scenario analysis suggest that the HBN provides an interpretable framework in which the interaction of controlling factors has causal relationships at different levels that can be identified and visualized.

Interpretable tree-based ensemble model for predicting beach water quality

Tree-based machine learning models based on environmental features offer low-cost and timely solutions for predicting microbial fecal contamination in beach water to inform the public of the health risk. However, many of these models are black boxes that are difficult for humans to understand, which may cause severe consequences such as unexplained decisions and failure in accountability. To develop interpretable predictive models for beach water quality, we evaluate five tree-based models, namely classification tree, random forest, CatBoost, XGBoost, and LightGBM, and employ a state-of-the-art explanation method SHAP to explain the models. When tested on the Escherichia coli (E. coli) concentration data collected from three beach sites along Lake Erie shores, LightGBM, followed by XGBoost, achieves the highest averaged precision and recall scores. For all three sites, both models suggest lake turbidity as the most important predictor, and elucidate the crucial role of accurate local data of wave height and rainfall in the model development. Local SHAP values further reveal the robustness of the importance of lake turbidity as its SHAP value increases nearly monotonically with its value and is minimally affected by other environmental factors. Moreover, we found an intriguing interaction between lake turbidity and day-of-year. This work suggests that the combination of LightGBM and SHAP has a promising potential to develop interpretable models for predicting microbial water quality in freshwater lakes.

Comparative Toxicity Analyses from Different Endpoints: Are New Cyclic Disinfection Byproducts (DBPs) More Toxic than Common Aliphatic DBPs?

In recent years, dozens of halogenated disinfection byproducts (DBPs) with cyclic structures were identified and detected in drinking water globally. Previous in vivo toxicity studies have shown that a few new cyclic DBPs possessed higher developmental toxicity and growth inhibition rate than common aliphatic DBPs; however, in vitro toxicity studies have proved that the latter exhibited higher cytotoxicity and genotoxicity than the former. Thus, to provide a more comprehensive toxicity comparison of DBPs from different endpoints, 11 groups of cyclic DBPs and nine groups of aliphatic DBPs were evaluated for their comparative in vitro and in vivo toxicity using human hepatoma cells (Hep G2) and zebrafish embryos. Notably, results showed that the in vitro Hep G2 cytotoxicity index of the aliphatic DBPs was nearly eight times higher than that of the cyclic DBPs, whereas the in vivo zebrafish embryo developmental/acute toxicity indexes of the cyclic DBPs were roughly 48-50 times higher than those of the aliphatic DBPs, indicating that the toxicity rank order differed when different endpoints were applied. For a broader comparison, a Pearson correlation analysis of DBP toxicity data from nine different endpoints was conducted. It was found that the observed Hep G2 cytotoxicity and zebrafish embryo developmental/acute toxicity in this study were highly correlated with the previously reported in vitro CHO cytotoxicity and in vivo toxicity in aquatic organisms (P < 0.01), respectively. However, the observed in vitro toxicity had no correlation with the in vivo toxicity (P > 0.05), suggesting that the toxicity rank orders obtained from in vitro and in vivo bioassays had large discrepancies. According to the observed toxicity data in this study and the candidate descriptors, two quantitative structure-activity relationship (QSAR) models were established, which help to further interpret the toxicity mechanisms of DBPs from different endpoints.

The assessment of emerging data-intelligence technologies for modeling Mg+2 and SO4-2 surface water quality

The concentration of soluble salts in surface water and rivers such as sodium, sulfate, chloride, magnesium ions, etc., plays an important role in the water salinity. Therefore, accurate determination of the distribution pattern of these ions can improve better management of drinking water resources and human health. The main goal of this research is to establish two novel wavelet-complementary intelligence paradigms so-called wavelet least square support vector machine coupled with improved simulated annealing (W-LSSVM-ISA) and the wavelet extended Kalman filter integrated with artificial neural network (W-EKF- ANN) for accurate forecasting of the monthly), magnesium (Mg⁺²), and sulfate (SO₄^-2) indices at Maroon River, in Southwest of Iran. The monthly River flow (Q), electrical conductivity (EC), Mg⁺², and SO₄^-2 data recorded at Tange-Takab station for the period 1980-2016. Some preprocessing procedures consisting of specifying the number of lag times and decomposition of the existing original signals into multi-resolution sub-series using three mother wavelets were performed to develop predictive models. In addition, the best subset regression analysis was designed to separately assess the best selective combinations for Mg⁺² and SO₄^-2. The statistical metrics and authoritative validation approaches showed that both complementary paradigms yielded promising accuracy compared with standalone artificial intelligence (AI) models. Furthermore, the results demonstrated that W-LSSVM-ISA-C1 (correlation coefficient (R) = 0.9521, root mean square error (RMSE) = 0.2637 mg/l, and Kling-Gupta efficiency (KGE) = 0.9361) and W-LSSVM-ISA-C4 (R = 0.9673, RMSE = 0.5534 mg/l and KGE = 0.9437), using Dmey mother that outperformed the W-EKF-ANN for predicting Mg⁺² and SO4^-2, respectively.

Machine learning methods for imbalanced data set for prediction of faecal contamination in beach waters

Predicting water contamination by statistical models is a useful tool to manage health risk in recreational beaches. Extreme contamination events, i.e. those exceeding normative are generally rare with respect to bathing conditions and thus the data is said to be imbalanced. Modeling and predicting those rare events present unique challenges. Here we introduce and evaluate several machine learning techniques and metrics to model imbalanced data and evaluate model performance. We do so by using a) simulated data-sets and b) a real data base with records of faecal coliform abundance monitored for 10 years in 21 recreational beaches in Uruguay (N ≈ 19000) using in situ and meteorological variables. We discuss advantages and disadvantages of the methods and provide a simple guide to perform models for a general audience. We also provide R codes to reproduce model fitting and testing. We found that most Machine Learning techniques are sensitive to imbalance and require specific data pre-treatment (e.g. upsampling) to improve performance. Accuracy (i.e. correctly classified cases over total cases) is not adequate to evaluate model performance on imbalanced data set. Instead, true positive rates (TPR) and false positive rates (FPR) are recommended. Among the 52 possible candidate algorithms tested, the stratified Random forest presented the better performance improving TPR in 50% with respect to baseline (0.4) and outperformed baseline in the evaluated metrics. Support vector machines combined with upsampling method or synthetic minority oversampling technique (SMOTE) performed well, similar to Adaboost with SMOTE. These results suggests that combining modeling strategies is necessary to improve our capacity to anticipate water contamination and avoid health risk.

Improving the robustness of beach water quality modeling using an ensemble machine learning approach

Microbial pollution of beach water can expose swimmers to harmful pathogens. Predictive modeling provides an alternative method for beach management that addresses several limitations associated with traditional culture-based methods of assessing water quality. Widely-used machine learning methods often suffer from high variability in performance from one year or beach to another. Therefore, the best machine learning method varies between beaches and years, making method selection difficult. This study proposes an ensemble machine learning approach referred to as model stacking that has a two-layered learning structure, where the outputs of five widely-used individual machine learning models (multiple linear regression, partial least square, sparse partial least square, random forest, and Bayesian network) are taken as input features for another model that produces the final prediction. Applying this approach to three beaches along eastern Lake Erie, New York, USA, we show that generally the model stacking approach was able to generate reliably good predictions compared to all of the five base models. The accuracy rankings of the stacking model consistently stayed 1^st or 2^nd every year, with yearly-average accuracy of 78%, 81%, and 82.3% at the three studied beaches, respectively. This study highlights the value of the model stacking approach in predicting beach water quality and solving other pressing environmental problems.

Using multi-threshold regression techniques to assess river fecal pollution in the highly urbanized Tamsui River watershed

Rivers are an important urban water resource. This study adopted multivariate linear regression (MLR) and logistic regression (LR) with multiple thresholds to assess river fecal pollution in the Tamsui River watershed using auxiliary environmental data. First, environmental data between 2015 and 2017 on land use, antecedent precipitation, population density, sewerage infrastructure, and river water quality were obtained using geographic information systems and served as explanatory variables. River fecal coliforms (FC), the dependent variable, were also collected for the same period. Then, MLR was used to establish an overall prediction model after validation, and to determine significant factors influencing the level of river fecal pollution. Finally, after stratifying the fecal pollution as low, medium, and high levels, LR with multiple thresholds was employed to explore key factors affecting different FC pollution levels. The study results revealed that land use type and river water quality (other than FC) strongly affected river FC pollution. The discharge of household sewage and wastewater from urban areas was a major source of river FC pollution, particularly for low and medium pollution levels, while farmland land use was negatively correlated with the medium and high levels of river FC pollution in the highly urbanized watershed. Biochemical oxygen demand and suspended solids were highly correlated with medium and high pollution levels in river water.

Exploring the biophysicochemical alteration of green alga Asterococcus superbus interactively affected by nanoparticles, triclosan and illumination

Toxic effects on Asterococcus superbus were studied based on different combinations of P25-TiO₂, nano-ZnO and triclosan under multiple illumination conditions. A full factorial design (2 × 2×2 × 3) was implemented to explore interactive effects, and to identify significant factors. The results showed illumination is the most important factor with significance and becomes one of the main reasons to affect chlorophyll pigments, photosynthesis activity, unsaturated fatty acids, mitochondria function, and cause oxidative stress. Triclosan considerably affects cell viability, photosynthesis activity, lipid peroxidation and protein structure, for which triclosan is more significant than nano-ZnO. P25 is significant for oxidative stress, antioxidant enzyme, and lipid peroxidation. P25 * nano-ZnO is the only significant interaction of pollutants, affecting macromolecules, lipid peroxidation, and photosynthesis activity. High-order interactions play significant roles in affecting multiple molecular components. Two groups of endpoints are best to reflect alga responses to interactively effects from P25, nano-ZnO, and triclosan. One is ROS, chlorophyll pigments, TBARS, area, MTT, and MMP, and the other one is chlorophyll pigments, ROS, TBARS, CAT, MTT and SOD. Our findings can be instructive for a comprehensive comparison among interactions of multiple pollutants and environmental factors in natural waters, such that more robust environmental toxicity analyses can be performed.

Comparative Quantitative Toxicology and QSAR Modeling of the Haloacetonitriles: Forcing Agents of Water Disinfection Byproduct Toxicity

The haloacetonitriles (HANs) is an emerging class of nitrogenous-disinfection byproducts (N-DBPs) present in disinfected drinking, recycled, processed wastewaters, and reuse waters. HANs were identified as primary forcing agents that accounted for DBP-associated toxicity. We evaluated the toxic characteristics of iodoacetonitrile (IAN), bromoacetonitrile (BAN), dibromoacetonitrile (DBAN), bromochloroacetonitrile (BCAN), tribromoacetonitrile (TBAN), chloroacetonitrile (CAN), dichloroacetonitrile (DCAN), trichloroacetonitrile (TCAN), bromodichloroacetonitrile (BDCAN), and chlorodibromoacetonitrile (CDBAN). This research generated the first quantitative, comparative analyses on the mammalian cell cytotoxicity, genotoxicity and thiol reactivity of these HANs. The descending rank order for HAN cytotoxicity was TBAN ≈ DBAN > BAN ≈ IAN > BCAN ≈ CDBAN > BDCAN > DCAN ≈ CAN ≈ TCAN. The rank order for genotoxicity was IAN ≈ TBAN ≈ DBAN > BAN > CDBAN ≈ BDCAN ≈ BCAN ≈ CAN ≈ TCAN ≈ DCAN. The rank order for thiol reactivity was TBAN > BDCAN ≈ CDBAN > DBAN > BCAN > BAN ≈ IAN > TCAN. These toxicity metrics were associated with membrane permeability and chemical reactivity. Based on their physiochemical parameters and toxicity metrics, we developed optimized, robust quantitative structure activity relationship (QSAR) models for cytotoxicity and for genotoxicity. These models can predict cytotoxicity and genotoxicity of novel HANs prior to analytical biological evaluation.

Prevalence and seasonal dynamics of blaCTX-M antibiotic resistance genes and fecal indicator organisms in the lower Lahn River, Germany

Antibiotic-resistant bacteria represent an emerging global health problem and are frequently detected in riverine environments. Analyzing the occurrence of corresponding antibiotic-resistant genes in rivers is of public interest as it contributes towards understanding the origin and dissemination of these emerging microbial contaminants via surface water. This is critical for devising strategies to mitigate the spread of resistances in the environment. Concentrations of bla_CTX-M antibiotic resistance genes were quantified weekly over a 12-month period in Lahn River surface water at two sampling sites using quantitative real-time PCR. Gene abundances were statistically assessed with regard to previously determined concentrations of fecal indicator organisms Escherichia coli, intestinal enterococci and somatic coliphages, as well as influential environmental factors. Similar seasonal patterns and strong positive correlations between fecal indicators and bla_CTX-M genes indicated identical sources. Accordingly, linear regression analyses showed that bla_CTX-M concentrations could largely be explained by fecal pollution. E. coli provided the best estimates (75% explained variance) at the upstream site, where proportions of bla_CTX-M genes in relation to fecal indicator organisms were highest. At this site, rainfall proved to be more influential, hinting at surface runoff as an emission source. The level of agricultural impact increased from downstream to upstream, linking increasing bla_CTX-M concentrations after rainfall events to the degree of agricultural land use. Exposure assessment revealed that even participants in non-swimming recreational activities were at risk of incidentally ingesting bla_CTX-M genes and thus potentially antibiotic resistant bacteria. Considering that bla_CTX-M genes are ubiquitous in Lahn River and participants in bathing and non-bathing water sports are at risk of exposure, results highlight the importance of microbial water quality monitoring with an emphasis on antibiotic resistance not only in designated bathing waters. Moreover, E. coli might serve as a suitable estimate for the presence of respective antibiotic resistant strains.

Prediction of recreational water safety using Escherichia coli as an indicator: case study of the Passaic and Pompton rivers, New Jersey

As contact with high concentrations of pathogens in a waterbody can cause waterborne diseases, Escherichia coli is commonly used as an indicator of water quality in routine public health monitoring of recreational freshwater ecosystems. However, traditional processes of detection and enumeration of pathogen indicators can be costly and are not time-sensitive enough to alarm recreational users. The predictive models developed to produce real-time predictions also have various methodological challenges, including arbitrary selection of explanatory variables, deterministic statistical approach, and heavy reliance on correlation instead of the more rigorous multivariate regression analyses, among others. The objective of this study is to address these challenges and develop a cost-effective and timely alternative for estimating pathogen indicators using real-time water quality and quantity data. As a case study we use New Jersey, where pathogens represent the most common cause of impairment for water quality, and Passaic and Pompton rivers, which are among the largest in the state and the country. We used Membrane Filtration Method and mColiblue24 media to enumerate Escherichia coli in a total of 69 water samples collected from April to November 2016 from the two rivers. We also collected data on environmental variables concurrently and performed stepwise and logistic regression analyses to address the said methodological challenges and determine the variables significantly predicting whether or not the Escherichia coli count was above prescribed levels for recreation activities. The results show that source water, higher specific conductance, lower pH, and cumulative rainfall for the 72 h antecedent the sampling significantly impacted the density of Escherichia coli. In addition to using the Bagging technique to validate the results, we also assessed Whole Model Tests, R², Entropy R², and Misclassification Rates. This approach improves the prediction of bacteria counts and their use in informing the potential safety/hazard of that waterbody for recreational activities.

Comparative cytotoxicity of halogenated aromatic DBPs and implications of the corresponding developed QSAR model to toxicity mechanisms of those DBPs: Binding interactions between aromatic DBPs and catalase play an important role

Halogenated aromatic disinfection byproducts (DBPs) are a new group of emerging DBPs identified recently. They have been detected in disinfected drinking water, wastewater effluents, recreational water and oil/gas produced water, at concentrations of ng/L to μg/L in general. Previously studies have demonstrated that most of them can induce developmental toxicity and growth inhibition in aquatic organisms based on in vivo bioassays. In this study, to further understand the adverse effects of aromatic DBPs to human health, the comparative cytotoxicity of 15 halogenated aromatic DBPs belonging to four subgroups (i.e., halophenols, halonitrophenols, halohydroxybenzaldehydes and halohydroxybenzoic acids) was evaluated with mammalian Chinese Hamster Ovary cells. The results indicated that the selected aromatic DBPs exhibited an in vitro toxicity rank order of halonitrophenols > halophenols > halohydroxybenzaldehydes > halohydroxybenzoic acids. The potential toxicity mechanisms involved with the antioxidant system were investigated by using molecular docking analysis between key antioxidant enzymes (i.e., catalase, superoxide dismutase, and glutathione S-transferase) and aromatic DBPs. Based on the observed cytotoxicity data and screening the candidate descriptors (including binding energies between the aromatic DBPs and key antioxidant enzymes as well as physical-chemical/quantum-chemical/topological descriptors), a QSAR model was developed as log (LC₅₀) ^-1 = - 1.050E_CAT + 0.300E_HOMO - 0.238E_LUMO- 0.164, indicating the importance of the interactions of aromatic DBPs towards catalase and the electrophilic/nucleophilic reactivity of aromatic DBPs in the toxicity mechanisms. In addition, the occurrence of the aromatic DBPs in tap water and finished water was studied in a mega city Shenzhen located in South China. Results showed that halogenated aromatic DBPs commonly existed in Shenzhen drinking water at ng/L levels, and three nitrogenous aromatic DBPs were detected in real drinking water for the first time. The major toxicity drivers among the target aromatic DBPs were identified through the integration of the measured concentrations and observed cytotoxicity; notably, DBPs with the highest concentrations may not contribute the highest proportions of overall toxicity.

Spatiotemporal prediction of Escherichia coli and Enterococci for the Commonwealth Games triathlon event using Bayesian Networks

A number of Bayesian Networks were developed in order to nowcast and forecast, up to 4 days ahead and in different locations, the likelihood of water quality within the 2018 Commonwealth Games Triathlon swim course exceeding the critical limits for Enterococci and Escherichia coli. The models are data-driven, but the identification of potential inputs and optimal model structure was performed through the parallel contribution of several stakeholders and experts, consulted through workshops. The models, whose main nodes were discretised with a customised discretisation algorithm, were validated over a test set of data and deployed in real-time during the Commonwealth Games in support to a traditional water quality monitoring program. The proposed modelling framework proved to be cost-effective and less time-consuming than process-based models while still achieving high accuracy; in addition, the added value of a continuous stakeholder engagement guarantees a shared understanding of the model outputs and its future deployment.

Relationship between Coliform Bacteria and Water Quality Factors at Weir Stations in the Nakdong River, South Korea

Artificial structures installed in rivers can change the natural physical, physiochemical, and biological characteristics of the rivers. Coliform bacteria are important water quality indicators, related to human health. This study investigated the relationship between coliform bacteria and water quality factors at eight weir stations constructed in the Nakdong River, a major river in South Korea. Fifteen water quality factors were analyzed at these sites from 2012 to 2016 using correlation and multiple regression analyses. The results for all stations confirmed the analytical validity, with high adjusted R2 values of approximately 0.6 and 0.8 on average for total and fecal coliforms, respectively. The results showed influential water quality factors affecting the concentration of coliform bacteria at weir stations. Specifically, total coliforms were mostly affected by organic matter and fecal coliforms were mostly affected by phosphate phosphorus and suspended solids. Rainfall was the most influential factor affecting both coliforms. Further, both coliforms were negatively affected by organic matter below the Dalseong weir in the mid- to downstream area of the Nakdong River. A positive relationship with phosphate phosphorus was indicated at all weir stations. To the authors’ knowledge, this kind of study has never been attempted so far. Thus, the study results can provide important information on influential water quality factors related to coliform bacteria, especially in the Nakdong River, creating a foundation for future water quality management.

Climate-driven QMRA model for selected water supply systems in Norway accounting for raw water sources and treatment processes

Formulating effective management intervention measures for water supply systems requires investigation of potential long-term impacts. This study applies an integrated multiple regression, random forest regression, and quantitative microbial risk assessment (QMRA) modelling approach to assess the effect of climate-driven precipitation on pathogen infection risks in three drinking water treatment plants (WTPs) in Norway. Pathogen removal efficacies of treatment steps were calculated using process models. The results indicate that while the WTPs investigated generally meet the current water safety guidelines, risks of Norovirus and Cryptosporidium infection may be of concern in the future. The pathogen infections attributable to current projections of average precipitation in the study locations may be low. However, the pathogen increases in the drinking water sources due to the occurrence of extreme precipitation events in the catchments could substantially increase the risks of pathogen infections. In addition, without optimal operation of the UV disinfection steps in the WTPs, both the present and potential future infection risks could be high. Therefore, the QMRA models demonstrated the need for improved optimization of key treatment steps in the WTPs, as well as implementation of stringent regulations in protecting raw water sources in the country. The variety of models applied and the pathogen: E. coli used in the study introduce some uncertainties in the results, thus, management decisions that will be based on the results should consider these limitations. Nevertheless, the integration of predictive models with QMRA as applied in this study could be a useful method for climate impact assessment in the water supply industry.

Statistical models of fecal coliform levels in Pacific Northwest estuaries for improved shellfish harvest area closure decision making

There is a substantial need for tools that effectively predict spatial and temporal fecal pollution patterns in estuarine waters. In this study, statistical models of exceedances of shellfish fecal coliform (FC) water quality criteria were developed using a 10-year dataset of FC levels and environmental data. Performance (sensitivity, specificity, and predictive capacity) of five different types of models was tested (MLR regression, Tobit (censored) regression, Firth's binary logistic regression (BLR), classification trees, and mixed-effects regression) for each of three conditionally managed shellfish-harvesting areas in Tillamook Bay, Oregon (USA). The most influential variables were related to precipitation and river stage height in the wet season and wind and tidal-stage in the dry season. Classification tree and Firth's BLR approaches better explained exceedances of shellfish water quality standards than the current closure thresholds. Findings demonstrate the utility of statistical modeling approaches for improved management of shellfish harvesting waters.

Factors influencing the fate of antibiotic resistance genes during thermochemical pretreatment and anaerobic digestion of pharmaceutical waste sludge

The prevalence of antibiotic resistance genes (ARGs) in waste sludge, especially for the pharmaceutical waste sludge, presents great potential risks to human health. Although ARGs and factors affecting their spreading are of major importance for human health, the factors influencing the fate of ARGs during sludge treatment, especially for pharmaceutical sludge treatment are not yet well understood. In order to be able to minimize ARGs spreading, it is important to find what is influencing their spreading. Therefore, certain factors, such as the sludge characteristics, bacterial diversity and community composition, and mobile genetic elements (MGEs) during the advanced AD of pharmaceutical sludge with different pretreatments were studied, and their affinity with ARGs was elucidated by Spearman correlation analysis. Furthermore, multiple linear regression was introduced to evaluate the importance of the various factors. Results showed that 59.7%-88.3% of the variations in individual ARGs and total ARGs can be explained by the corresponding factors. Bacterial diversity rather than specific bacterial community composition affected the fate of ARGs, whereas alkalinity was the most important factor on ARGs among all sludge characteristics investigated in this study. Besides, 66.4% of variation of total ARGs was driven by the changes of MGEs. Multiple linear regression models also reveal the collective effect of these factors on ARGs, and the contributions of each factor impact on ARGs. This study provides more comprehension about the factors impact on the fate of ARGs during pharmaceutical sludge treatment, and offers an approach to evaluate the importance of each factor, which method could be introduced for evaluation of factors influencing ARGs during other types of sludge or wastewater treatment.

On the implementation of reliable early warning systems at European bathing waters using multivariate Bayesian regression modelling

For ensuring microbial safety, the current European bathing water directive (BWD) (76/160/EEC 2006) demands the implementation of reliable early warning systems for bathing waters, which are known to be subject to short-term pollution. However, the BWD does not provide clearly defined threshold levels above which an early warning system should start warning or informing the population. Statistical regression modelling is a commonly used method for predicting concentrations of fecal indicator bacteria. The present study proposes a methodology for implementing early warning systems based on multivariate regression modelling, which takes into account the probabilistic character of European bathing water legislation for both alert levels and model validation criteria. Our study derives the methodology, demonstrates its implementation based on information and data collected at a river bathing site in Berlin, Germany, and evaluates health impacts as well as methodological aspects in comparison to the current way of long-term classification as outlined in the BWD.

Maxent estimation of aquatic Escherichia coli stream impairment

BACKGROUND

The leading cause of surface water impairment in United States' rivers and streams is pathogen contamination. Although use of fecal indicators has reduced human health risk, current approaches to identify and reduce exposure can be improved. One important knowledge gap within exposure assessment is characterization of complex fate and transport processes of fecal pollution. Novel modeling processes can inform watershed decision-making to improve exposure assessment.

METHODS

We used the ecological model, Maxent, and the fecal indicator bacterium Escherichia coli to identify environmental factors associated with surface water impairment. Samples were collected August, November, February, and May for 8 years on Sinking Creek in Northeast Tennessee and analyzed for 10 water quality parameters and E. coli concentrations. Univariate and multivariate models estimated probability of impairment given the water quality parameters. Model performance was assessed using area under the receiving operating characteristic (AUC) and prediction accuracy, defined as the model's ability to predict both true positives (impairment) and true negatives (compliance). Univariate models generated action values, or environmental thresholds, to indicate potential E. coli impairment based on a single parameter. Multivariate models predicted probability of impairment given a suite of environmental variables, and jack-knife sensitivity analysis removed unresponsive variables to elicit a set of the most responsive parameters.

RESULTS

Water temperature univariate models performed best as indicated by AUC, but alkalinity models were the most accurate at correctly classifying impairment. Sensitivity analysis revealed that models were most sensitive to removal of specific conductance. Other sensitive variables included water temperature, dissolved oxygen, discharge, and NO3. The removal of dissolved oxygen improved model performance based on testing AUC, justifying development of two optimized multivariate models; a 5-variable model including all sensitive parameters, and a 4-variable model that excluded dissolved oxygen.

DISCUSSION

Results suggest that E. coli impairment in Sinking Creek is influenced by seasonality and agricultural run-off, stressing the need for multi-month sampling along a stream continuum. Although discharge was not predictive of E. coli impairment alone, its interactive effect stresses the importance of both flow dependent and independent processes associated with E. coli impairment. This research also highlights the interactions between nutrient and fecal pollution, a key consideration for watersheds with multiple synergistic impairments. Although one indicator cannot mimic theplethora of existing pathogens in water, incorporating modeling can fine tune an indicator's utility, providing information concerning fate, transport, and source of fecal pollution while prioritizing resources and increasing confidence in decision making.

Multiple Linear Regression Models for Predicting Nonpoint-Source Pollutant Discharge from a Highland Agricultural Region

Sediment runoff from dense highland field areas greatly affects the quality of downstream lakes and drinking water sources. In this study, multiple linear regression (MLR) models were built to predict diffuse pollutant discharge using the environmental parameters of a basin. Explanatory variables that influence the sediment and pollutant discharge can be identified with the model, and such research could play an important role in limiting sediment erosion in the dense highland field area. Pollutant load per event, event mean concentration (EMC), and pollutant load per area were estimated from stormwater survey data from the Lake Soyang basin. During the wet season, heavy rains cause large amounts of suspended sediment and the occurrence of such rains is increasing due to climate change. The explanatory variables used in the MLR models are the percentage of fields, subbasin area, and mean slope of subbasin as topographic parameters, and the number of preceding dry days, rainfall intensity, rainfall depth, and rainfall duration as rainfall parameters. In the MLR modeling process, four types of regression equations with and without log transformation of the explanatory and response variables were examined to identify the best performing regression model. The performance of the MLR models was evaluated using the coefficient of determination (R2), root mean square error (RMSE), coefficient of variation of the root mean square error (CV(RMSE)), the ratio of the RMSE to the standard deviation of the observed data (RSR) and the Nash–Sutcliffe model efficiency (NSE). The performance of the MLR models of pollutant load except total nitrogen (TN) was good under the condition of RSR, and satisfactory for the NSE and R2. In the EMC and load/area models, the performance for suspended solids (SS) and total phosphorus (TP) was good for the RSR, and satisfactory for the NSE and R2. The standardized coefficients for the models were analyzed to identify the influential explanatory variables in the models. In the final performance evaluation, the results of jackknife validation indicate that the MLR models are robust.

Comparative predictive modelling of the occurrence of faecal indicator bacteria in a drinking water source in Norway

Presently, concentrations of fecal indicator bacteria (FIB) in raw water sources are not known before water undergoes treatment, since analysis takes approximately 24h to produce results. Using data on water quality and environmental variables, models can be used to predict real time concentrations of FIB in raw water. This study evaluates the potentials of zero-inflated regression models (ZI), Random Forest regression model (RF) and adaptive neuro-fuzzy inference system (ANFIS) to predict the concentration of FIB in the raw water source of a water treatment plant in Norway. The ZI, RF and ANFIS faecal indicator bacteria predictive models were built using physico-chemical (pH, temperature, electrical conductivity, turbidity, color, and alkalinity) and catchment precipitation data from 2009 to 2015. The study revealed that pH, temperature, turbidity, and electrical conductivity in the raw water were the most significant factors associated with the concentration of FIB in the raw water source. Compared to the other models, the ANFIS model was superior (Mean Square Error=39.49, 0.35, 0.09, 0.23CFU/100ml respectively for coliform bacteria, E. coli, Intestinal enterococci and Clostridium perfringens) in predicting the variations of FIB in the raw water during model testing. However, the model was not capable of predicting low counts of FIB during both training and testing stages of the models. The ZI and RF models were more consistent when applied to testing data, and they predicted FIB concentrations that characterized the observed FIB concentrations. While these models might need further improvement, results of this study indicate that ZI and RF regression models have high prospects as tools for the real-time prediction of FIB in raw water sources for proactive microbial risk management in water treatment plants.

Tracking the downstream impacts of inadequate sanitation in central Appalachia

Poor sanitation in rural infrastructure is often associated with high levels of fecal contamination in adjacent surface waters, which presents a community health risk. Although microbial source tracking techniques have been widely applied to identify primary remediation needs in urban and/or recreational waters, use of human-specific markers has been more limited in rural watersheds. This study quantified the human source tracking marker Bacteroides-HF183, along with more general fecal indicators (i.e. culturable Escherichia coli and a molecular Enterococcus marker), in two Appalachian streams above and below known discharges of untreated household waste. Although E. coli and Enterococcus were consistently recovered in samples collected from both streams, Bacteroides-HF183 was only detected sporadically in one stream. Multiple linear regression analysis demonstrated a positive correlation between the concentration of E. coli and the proximity and number of known waste discharge points upstream; this correlation was not significant with respect to Bacteroides-HF183, likely due to the low number of quantifiable samples. These findings suggest that, while the application of more advanced source targeting strategies can be useful in confirming the influence of substandard sanitation on surface waters to justify infrastructure improvements, they may be of limited use without concurrent traditional monitoring targets and on-the-ground sanitation surveys.

Monitoring bacterial indicators of water quality in a tidally influenced delta: A Sisyphean pursuit

The Sacramento-San Joaquin Delta Estuary (Delta) is the confluence of two major watersheds draining the Western Sierra Nevada mountains into the Central Valley of California, ultimately terminating into San Francisco Bay. We sampled 88 sites once a month for two years (2006-2008) over 87 separate sampling events for a total of 1740 samples. Water samples were analyzed for fecal indicator bacteria (Escherichia coli, enterococci and fecal coliforms), and 53 other physiochemical, land use, and environmental characteristics. The purpose of the study was to create a baseline of microbial water quality in the Delta and to identify various factors (climatic, land use, tidal, etc.) that were associated with elevated concentrations of indicator bacteria. Fecal indicator bacteria generally had weak to modest relationships to environmental conditions; the strength and direction of which varied for each microbial indicator, drainage region, and across seasons. Measured and unmeasured, site-specific effects accounted for large portions of variance in model predictions (ρ=0.086 to 0.255), indicating that spatial autocorrelation was a major component of water quality outcomes. The effects of tidal cycling and lack of connectivity between waterways and surrounding landscapes likely contributed to the lack of association between local land uses and microbial outcomes, though weak associations may also be indicative of mismatched spatiotemporal scales. The complex nature of this system necessitates continued monitoring and regular updates to statistical models designed to predict microbial water quality.