Machine learning approaches to identify spatial factors and their influential distances for heavy metal contamination in downstream sediment.

Contaminated sediments can adversely affect aquatic ecosystems, making the identification and management of pollutant sources extremely important. In this study, we proposed machine learning approaches to reveal sources and their influential distances for heavy metal contamination of downstream sediment. We employed classification models with artificial neural networks (ANN) and random forest (RF), respectively, to predict the heavy metal contamination of stream sediments using upland environmental variables as input features. A comprehensive Korean nationwide monitoring database containing 1546 datasets was used to train and test the models. These datasets encompass the concentrations of eight heavy metals (Ar, Cd, Cr, Cu, Hg, Ni, Pb, and Zn) in sediment samples collected from 160 stream sites across the nation from 2014 to 2018. Model's prediction accuracy was evaluated for input feature sets from different influential upland areas defined by different buffer radii and the watershed boundary for each site. Although both ANN and RF models were unsatisfactory in predicting heavy metal quartile classes, RF-classifiers with adaptive synthetic oversampling (ORFC) showed reasonably well-predicted classes of the sediment samples based on the Canada's Sediment Quality Guidelines (accuracy ranged from 0.67 to 0.94). The best influential distance (i.e., buffer radius) was determined for each heavy metal based on the accuracy of ORFC. The results indicated that Cd, Cu and Pb had shorter influential distances (1.5-2.0 km) than the other heavy metals with little difference in accuracy for different influential distances. Feature importance calculation revealed that upland soil contamination was the primary factor for Hg and Ni, while residential areas and roads were significant features associated with Pb and Zn contamination. This approach offers information on major contamination sources and their influential areas to be prioritized for managing contaminated stream sediments.

Wastewater-borne viruses and bacteria, surveillance and biosensors at the interface of academia and field deployment.

Wastewater is a complex, but an ideal, matrix for disease monitoring and surveillance as it represents the entire load of enteric pathogens from a local catchment area. It captures both clinical and community disease burdens. Global interest in wastewater surveillance has been growing rapidly for infectious diseases monitoring and for providing an early warning of potential outbreaks. Although molecular detection methods show high sensitivity and specificity in pathogen monitoring from wastewater, they are strongly limited by challenges, including expensive laboratory settings and prolonged sample processing and analysis. Alternatively, biosensors exhibit a wide range of practical utility in real-time monitoring of biological and chemical markers. However, field deployment of biosensors is primarily challenged by prolonged sample processing and pathogen concentration steps due to complex wastewater matrices. This review summarizes the role of wastewater surveillance and provides an overview of infectious viral and bacterial pathogens with cutting-edge technologies for their detection. It emphasizes the practical utility of biosensors in pathogen monitoring and the major bottlenecks for wastewater surveillance of pathogens, and overcoming approaches to field deployment of biosensors for real-time pathogen detection. Furthermore, the promising potential of novel machine learning algorithms to resolve uncertainties in wastewater data is discussed.

Identifying the acute toxicity of contaminated sediments using machine learning models.

Ecological risk assessment of contaminated sediment has become a fundamental component of water quality management programs, supporting decision-making for management actions or prompting additional investigations. In this study, we proposed a machine learning (ML)-based approach to assess the ecological risk of contaminated sediment as an alternative to existing index-based methods and costly toxicity testing. The performance of three widely used index-based methods (the pollution load index, potential ecological risk index, and mean probable effect concentration) and three ML algorithms (random forest, support vector machine, and extreme gradient boosting [XGB]) were compared in their prediction of sediment toxicity using 327 nationwide data sets from Korea consisting of 14 sediment quality parameters and sediment toxicity testing data. We also compared the performances of classifiers and regressors in predicting the toxicity for each of RF, SVM, and XGB algorithms. For all algorithms, the classifiers poorly classified toxic and non-toxic samples due to limited information on the sediment composition and the small training dataset. The regressors with a given classification threshold provided better classification, with the XGB regressor outperforming the other models in the classification. A permutation feature importance analysis revealed that Cr, Cu, Pb, and Zn were major contributors to toxicity prediction. The ML-based approach has the potential to be even more useful in the future with the expected increase in available sediment data.

Optimal allocation and operation of sewer monitoring sites for wastewater-based disease surveillance: A methodological proposal.

Wastewater-based epidemiology (WBE) is drawing increasing attention as a promising tool for an early warning of emerging infectious diseases such as COVID-19. This study demonstrated the utility of a spatial bisection method (SBM) and a global optimization algorithm (i.e., genetic algorithm, GA), to support better designing and operating a WBE program for disease surveillance and source identification. The performances of SBM and GA were compared in determining the optimal locations of sewer monitoring manholes to minimize the difference among the effective spatial monitoring scales of the selected manholes. While GA was more flexible in determining the spatial resolution of the monitoring areas, SBM allows stepwise selection of optimal sampling manholes with equiareal subcatchments and lowers computational cost. Upon detecting disease outbreaks at a regular sewer monitoring site, additional manholes within the catchment can be selected and monitored to identify source areas with a required spatial resolution. SBM offered an efficient method for rapidly searching for the optimal locations of additional sampling manholes to identify the source areas. This study provides strategic and technical elements of WBE including sampling site selection with required spatial resolution and a source identification method.

Improving the performance of machine learning models for early warning of harmful algal blooms using an adaptive synthetic sampling method.

Many countries have attempted to monitor and predict harmful algal blooms to mitigate related problems and establish management practices. The current alert system-based sampling of cell density is used to intimate the bloom status and to inform rapid and adequate response from water-associated organizations. The objective of this study was to develop an early warning system for cyanobacterial blooms to allow for efficient decision making prior to the occurrence of algal blooms and to guide preemptive actions regarding management practices. In this study, two machine learning models: artificial neural network (ANN) and support vector machine (SVM), were constructed for the timely prediction of alert levels of algal bloom using eight years' worth of meteorological, hydrodynamic, and water quality data in a reservoir where harmful cyanobacterial blooms frequently occur during summer. However, the proportion imbalance on all alert level data as the output variable leads to biased training of the data-driven model and degradation of model prediction performance. Therefore, the synthetic data generated by an adaptive synthetic (ADASYN) sampling method were used to resolve the imbalance of minority class data in the original data and to improve the prediction performance of the models. The results showed that the overall prediction performance yielded by the caution level (L1) and warning level (L2) in the models constructed using a combination of original and synthetic data was higher than the models constructed using original data only. In particular, the optimal ANN and SVM constructed using a combination of original and synthetic data during both training (including validation) and test generated distinctively improved recall and precision values of L1, which is a very critical alert level as it indicates a transition status from normalcy to bloom formation. In addition, both optimal models constructed using synthetic-added data exhibited improvement in recall and precision by more than 33.7% while predicting L-1 and L-2 during the test. Therefore, the application of synthetic data can improve detection performance of machine learning models by solving the imbalance of observed data. Reliable prediction by the improved models can be used to aid the design of management practices to mitigate algal blooms within a reservoir.

An empirical modeling approach to predicting pollutant loads and developing cost-effective stormwater treatment strategies for a large urban watershed.

Stormwater treatment strategies were evaluated for the upper Ballona Creek Watershed in Los Angeles, CA using an empirical model of stormwater runoff quantity and quality with zeroth-order regularization and a limited memory Broyden-Fletcher-Goldfarb-Shanno Bound constrained optimization routine. The model used landuse based estimation on the runoff volume, event mean concentration (EMC) and pollutant load employing ten different landuses, including highways and local roads. The model was validated by showing that its predictions were in reasonable agreement (r2 ~0.6 to 0.8) with total zinc (Zn), Total Kjeldahl Nitrogen (TKN), and Total Suspended Solids (TSS) loadings measured at the monitoring site at the bottom of the watershed. The developed model was used to demonstrate and quantify the benefits of the stormwater treatment practices (STPs) prioritized at specific landuses with high pollutant mass emission rates. For this demonstration, total Zn was selected as it is one of the most concerning pollutants in an extremely urbanized area such as the Ballona Creek Watershed. Transportation landuse including local roads and highways was found to be the best candidate for the STP applications due to their high percent load contribution per percent area. By focusing STPs for transportation landuse, the water quality goal of total Zn in the study watershed was expected be achieved at approximately 75% less cost.

Associating the spatial properties of a watershed with downstream Chl-a concentration using spatial analysis and generalized additive models.

We examined the relationship between downstream algal growth potential and the spatial environmental factors of both upland areas and stream buffer zones using spatial analysis and generalized additive models (GAMs). The models employed site-representative concentrations of chlorophyll a (Chl-a) from a total of 688 national water quality monitoring stations and the spatial factors of the corresponding 688 watersheds. The spatial environmental factors included topography, climate, land use class, soil type, and proximity of the monitoring station to the weir downstream and wastewater treatment plants (WWTPs). The explanatory power (adjusted R2 or Radj2) of the models was used to compare different spatial influential scales defined by stream buffers and upstream circular buffers. The spatial environmental factors of the entire watershed area better explained the inter-station variation in Chl-a than did those of the stream buffer and/or upstream circular buffer areas. However, the spatial environmental factors of watershed areas more than 25 km upstream circular buffer zones had only minor influence on the explainability of the models with regards to the inter-station variation in Chl-a levels. Generally, land use patterns were more strongly related to the inter-station Chl-a variation than were point sources of pollutants such as WWTPs. The two most influencing land uses on the inter-station Chl-a variation were urban and agricultural land uses, with varying relative contributions depending on the spatial influential scale: In general relative contribution of urban land use was larger at a larger spatial influential scale while that of agricultural land use showed an opposite trend. In addition, the proximity to the weir downstream explained high Chl-a concentrations in the stream water. Relative importance and causal effects of the spatial environmental variables to instream Chl-a were established based on this national scale correlative analysis, leading to decision-making with the goal of controlling instream algal growth.

Assessment of in vivo genotoxicity of citrated-coated silver nanoparticles via transcriptomic analysis of rabbit liver tissue.

BACKGROUND: Silver nanoparticles (AgNPs) are widely used in industrial and household applications, arousing concern regarding their safety in humans. The risks posed by stabilizer-coated AgNPs continue to be unclear, and assessing their toxicity is for an understanding of the safety issues involved in their use in various applications. PURPOSE: We aimed to investigated the long-term toxicity of citrate-coated silver nanoparticles (cAgNPs) in liver tissue using several toxicity tests and transcriptomic analysis at 7 and 28 days after a single intravenous injection into rabbit ear veins (n=4). MATERIALS AND METHODS: The cAgNPs used in this study were in the form of a 20% (w/v) aqueous solution, and their size was 7.9±0.95 nm, measured using transmission electron microscopy. The animal experiments were performed based on the principles of good laboratory practice. RESULTS: Our results showed that the structure and function of liver tissue were disrupted due to a single exposure to cAgNPs. In addition, in vivo comet assay showed unrepaired genotoxicity in liver tissue until 4 weeks after a single injection, suggesting a potential carcinogenic effect of cAgNPs. In our transcriptomic analysis, a total of 244 genes were found to have differential expression at 28 days after a single cAgNP injection. Carefully curated pathway analysis of these genes using Pathway Studio and Ingenuity Pathway Analysis tools revealed major molecular networks responding to cAgNP exposure and indicated a high correlation of the genes with inflammation, hepatotoxicity, and cancer. Molecular validation suggested potential biomarkers for assessing the toxicity of accumulated cAgNPs. CONCLUSION: Our investigation highlights the risk associated with a single cAgNP exposure with unrepaired damage persisting for at least a month.

Use of coliphages to investigate norovirus contamination in a shellfish growing area in Republic of Korea.

A number of severe norovirus outbreaks due to the consumption of contaminated shellfish have been reported recently. In this study, we evaluated the distribution of coliphage densities to determine their efficacy as fecal indicators of enteric viruses, including noroviruses, in water samples collected from a shellfish growing area in Republic of Korea over a period of approximately 1 year. Male-specific and somatic coliphages in water samples were analyzed using the single agar layer method, and norovirus genogroups I and II, which infect mainly humans, were analyzed using duplex reverse transcription quantitative PCR. Male-specific and somatic coliphages were detected widely throughout the study area. Several environmental parameters, including salinity, precipitation, temperature, and wind speed were significantly correlated with coliphage concentrations (P < 0.05). Moreover, the concentrations of male-specific coliphages were positively correlated with the presence of human noroviruses (r = 0.443; P < 0.01). The geospatial analysis with coliphage concentrations using a geographic information system revealed that densely populated residential areas were the major source of fecal contamination. Our results indicate that coliphage monitoring in water could be a useful approach to prevent norovirus contamination in shellfish.

Host-Specific Bacteroides Markers-Based Microbial Source Tracking in Aquaculture Areas.

Various waterborne pathogens originate from human or animal feces and may cause severe gastroenteric outbreaks. Bacteroides spp. that exhibit strong host- or group-specificities are promising markers for identifying fecal sources and their origins. In the present study, 240 water samples were collected from two major aquaculture areas in Republic of Korea over a period of approximately 1 year, and the concentrations and occurrences of four host-specific Bacteroides markers (human, poultry, pig, and ruminant) were evaluated in the study areas. Host-specific Bacteroides markers were detected widely in the study areas, among which the poultry-specific Bacteroides marker was detected at the highest concentration (1.0-1.2 log10 copies L-1). During the sampling period, high concentrations of host-specific Bacteroides markers were detected between September and December 2015. The host-specific Bacteroides marker-combined geospatial map revealed the up-to-downstream gradient of fecal contamination, as well as the effects of land-use patterns on host-specific Bacteroides marker concentrations. In contrast to traditional bacterial indicators, the human-specific Bacteroides marker correlated with human specific pathogens, such as noroviruses (r=0.337; P<0.001). The present results indicate that host-specific Bacteroides genetic markers with an advanced geospatial analysis are useful for tracking fecal sources and associated pathogens in aquaculture areas.

Photocatalytic removal of gaseous nitrogen oxides using WO3/TiO2 particles under visible light irradiation: Effect of surface modification.

Photocatalytic nanoparticles have been receiving considerable attention for their potential use in many environmental management applications, including urban air quality control. This paper investigates the performance of surface modified WO3/TiO2 composite particles in removing gaseous nitrogen oxides (NOx) under visible light irradiation. The WO3/TiO2 composite particles were synthesized using a modified wet chemical method with different concentrations of NaOH solution used as a surface modification agent for the host TiO2 particles. The NOx removal efficiency of the WO3/TiO2 particles was evaluated using a lab-scale continuous gas flow photo-reactor with a gas contact time of 1 min. Results showed that surface modification using NaOH can enhance the photocatalytic activity of the WO3/TiO2 particles. The NOx removal efficiency of the surface modified WO3/TiO2 was greater than 90%, while that of WO3/TiO2 particles prepared by the conventional wet chemical method was ∼75%. The enhanced removal efficiency might be attributed to the formation of oxygen vacancies on the TiO2 surface, providing sites for WO3 particles to effectively bind with TiO2. However, excess amount of NaOH >3 M deteriorated the photocatalytic performance due to the increased agglomeration of the host TiO2 particles.

Potential use of ionic species for identifying source land-uses of stormwater runoff.

Identifying critical land-uses or source areas is important to prioritize resources for cost-effective stormwater management. This study investigated the use of information on ionic composition as a fingerprint to identify the source land-use of stormwater runoff. We used 12 ionic species in stormwater runoff monitored for a total of 20 storm events at five sites with different land-use compositions during the 2012-2014 wet seasons. A stepwise forward discriminant function analysis (DFA) with the jack-knifed cross validation approach was used to select ionic species that better discriminate the land-use of its source. Of the 12 ionic species, 9 species (K+, Mg2+, Na+, NH4+, Br-, Cl-, F-, NO2-, and SO42-) were selected for better performance of the DFA. The DFA successfully differentiated stormwater samples from urban, rural, and construction sites using concentrations of the ionic species (70%, 95%, and 91% of correct classification, respectively). Over 80% of the new data cases were correctly classified by the trained DFA model. When applied to data cases from a mixed land-use catchment and downstream, the DFA model showed the greater impact of urban areas and rural areas respectively in the earlier and later parts of a storm event.

Structural improvements on hydrodynamic separators: a computational fluid dynamics approach.

Hydrodynamic separators (HDSs) have been used extensively to reduce stormwater pollutants from urbanized areas before entering the receiving water bodies. They primarily remove particulates and associated pollutants using gravity settling. Two types of HDSs with different structural configurations of the inner vortex-inducing components were presented in this study. One configuration consisted of a dip cylindrical plate with a center shaft while the other one has a hollow screen inside. With the help of computational fluid dynamics, the performance of these different types of HDSs have been evaluated and comparatively analyzed. The results showed that the particle removal efficiency was better with the cylindrical plate type HDSs than the screen type HDSs because of the larger swirling flow regime formed inside the device. Plate type HDSs were found more effective in removing fine particles (∼50 µm) than the screen type HDSs that were only efficient in removing large particles (≥250 µm). Structural improvements in a HDS such as increase in diameter and angle of the inlet pipe can enhance the removal efficiencies by up to 20% for plate type HDS while increase in the screen diameter can increase removal efficiencies of the screen type HDS.

Identification of environmental determinants for spatio-temporal patterns of norovirus outbreaks in Korea using a geographic information system and binary response models.

Although norovirus outbreaks are well-recognized to have strong winter seasonality relevant to low temperature and humidity, the role of artificial human-made features within geographical areas in norovirus outbreaks has rarely been studied. The aim of this study is to assess the natural and human-made environmental factors favoring the occurrence of norovirus outbreaks using nationwide surveillance data. We used a geographic information system and binary response models to examine whether the norovirus outbreaks are spatially patterned and whether these patterns are associated with specific environmental variables including service levels of water supply and sanitation systems and land-use types. The results showed that small-scale low-tech local sewage treatment plants and winter sports areas were statistically significant factors favoring norovirus outbreaks. Compactness of the land development also affected the occurrence of norovirus outbreaks; transportation, water, and forest land-uses were less favored for effective transmission of norovirus, while commercial areas were associated with an increased rate of norovirus outbreaks. We observed associations of norovirus outbreaks with various outcomes of human activities, including discharge of poorly treated sewage, overcrowding of people during winter season, and compactness of land development, which might help prioritize target regions and strategies for the management of norovirus outbreaks.

Characterization and source identification of pollutants in runoff from a mixed land use watershed using ordination analyses.

While identification of critical pollutant sources is the key initial step for cost-effective runoff management, it is challenging due to the highly uncertain nature of runoff pollution, especially during a storm event. To identify critical sources and their quantitative contributions to runoff pollution (especially focusing on phosphorous), two ordination methods were used in this study: principal component analysis (PCA) and positive matrix factorization (PMF). For the ordination analyses, we used runoff quality data for 14 storm events, including data for phosphorus, 11 heavy metal species, and eight ionic species measured at the outlets of subcatchments with different land use compositions in a mixed land use watershed. Five factors as sources of runoff pollutants were identified by PCA: agrochemicals, groundwater, native soils, domestic sewage, and urban sources (building materials and automotive activities). PMF identified similar factors to those identified by PCA, with more detailed source mechanisms for groundwater (i.e., nitrate leaching and cation exchange) and urban sources (vehicle components/motor oils/building materials and vehicle exhausts), confirming the sources identified by PCA. PMF was further used to quantify contributions of the identified sources to the water quality. Based on the results, agrochemicals and automotive activities were the two dominant and ubiquitous phosphorus sources (39-61 and 16-47 %, respectively) in the study area, regardless of land use types.

Evaluation of accuracy of linear regression models in predicting urban stormwater discharge characteristics.

Stormwater runoff has been identified as a source of pollution for the environment, especially for receiving waters. In order to quantify and manage the impacts of stormwater runoff on the environment, predictive models and mathematical models have been developed. Predictive tools such as regression models have been widely used to predict stormwater discharge characteristics. Storm event characteristics, such as antecedent dry days (ADD), have been related to response variables, such as pollutant loads and concentrations. However it has been a controversial issue among many studies to consider ADD as an important variable in predicting stormwater discharge characteristics. In this study, we examined the accuracy of general linear regression models in predicting discharge characteristics of roadway runoff. A total of 17 storm events were monitored in two highway segments, located in Gwangju, Korea. Data from the monitoring were used to calibrate United States Environmental Protection Agency's Storm Water Management Model (SWMM). The calibrated SWMM was simulated for 55 storm events, and the results of total suspended solid (TSS) discharge loads and event mean concentrations (EMC) were extracted. From these data, linear regression models were developed. R(2) and p-values of the regression of ADD for both TSS loads and EMCs were investigated. Results showed that pollutant loads were better predicted than pollutant EMC in the multiple regression models. Regression may not provide the true effect of site-specific characteristics, due to uncertainty in the data.

Developing a flow control strategy to reduce nutrient load in a reclaimed multi-reservoir system using a 2D hydrodynamic and water quality model.

Blocking the natural bi-directional flow in an estuarine system using an artificial dyke has commonly caused serious water quality problems. In the southwestern part of South Korea, a parallel triple-reservoir system was constructed by blocking the mouth of three different rivers (Yeongsan, Okcheon, and Kumja), which were then interconnected using two open channels. This system has experienced a deterioration in water quality due to pollutants accumulated from the upper watershed, and has continually discharged pollutant loads to the outer ocean. Therefore, the objective of this study is to establish an effective dam operation plan for reducing nutrient loads released from the integrated reservoir. In this study, the CE-QUAL-W2 model, which is a 2-dimentional hydrodynamic and water quality model, was applied to predict the pollutant load released from each reservoir in response to different flow scenarios for the interconnecting channel. The model was calibrated using two novel methods: a sensitivity analysis to determine meaningful model parameters, and a pattern search to optimize the parameters. From the scenario analysis using flow control, it was determined that the total nitrogen (TN) and total phosphorus (TP) loadings could be reduced by 27.2% and 6.6%, respectively, under the optimal channel flow scenario by regulating the chlorophyll-a concentration in the reservoir. The results confirm that effective dam operation could contribute to a decrease in pollutant loads in the receiving seawater body. As such, this study suggests operational strategies for a multi-reservoir system that can be used to reduce the nutrient load being discharged from reservoirs.

Accuracy of a stormwater monitoring program for urban landuses.

This study examined the accuracy of an urban stormwater monitoring program in estimating the annual discharge load (L(T)) and the annual reduction rate by a stormwater treatment device (R(T)) for total suspended solids. A calibrated stormwater management model was used to generate the entire stormwater runoff events in one year and was used to estimate L(T) and R(T) under different monitoring strategies having limited numbers of runoff events, including random, wet season, antecedent dry days (ADD)-based, monthly, and seasonally weighted. For random monitoring, 12 storms were required to estimate the values of L(T) and R(T) with mean relative errors of 13.98 and 0.24%, respectively. Monthly monitoring had slightly greater mean relative errors compared to random monitoring. Wet season and ADD-based monitoring under- or overestimated both L(T) and R(T). Monitoring with equal numbers of storms from the wet and dry seasons best estimated L(T) and R(T).

Optimal design of a hydrodynamic separator for treating runoff from roadways.

This study investigated the removal efficiency of target pollutants from an underground stormwater treatment device (a hydrodynamic separator), focusing on the overall performance of the devices of a catchment. An approach for sizing an underground stormwater treatment device was developed, in order to obtain the required reduction percentage of the total suspended solids (TSS) generated from a given impervious catchment. The United States Environmental Protection Agency's stormwater management model (SWMM) was used for developing contours to help determine the size of the device, with respect to the maximum inflow to the device (or bypass rate), and the catchment area served by the device. Additionally, three different configurations of underground stormwater treatment devices were examined. It was found that, for a given catchment area, a single large device provides slightly better performance than multiple small devices. The approach we propose here can be useful to determine the sizes, as well as to clarify the efficiencies, of different installation configurations of underground stormwater treatment device (e.g. a hydrodynamic separator) in relation to their bypass rates and site specific conditions, such as rainfall characteristics and the catchment area to be served.

Distribution of antibiotic resistance in urban watershed in Japan.

Antibiotic-resistant E. coli concentrations showed large spatial and temporal variations, with greater concentrations observed in tributaries and downstream than in the upstream and midstream. Twenty percent of the geometric mean concentrations of antibiotic-resistant E. coli in the Tama River basin (Japan) exceeded the maximum acceptable concentration of indicator E. coli established by the USEPA. The indicator E. coli concentrations were positively correlated with those of antibiotic-resistant E. coli and multiple-antibiotic-resistant E. coli (resistance to more than two kinds of antibiotics), respectively, but not the detection rate of antibiotic-resistant E. coli, implying that use of antibiotic-resistant E. coli concentration rather than the detection rate can be a better approach for water quality assessment. Multiple-antibiotic-resistant E. coli is a useful indicator for estimating the resistance diffusion, water quality degradation and public health risk potential. This assessment provides beneficial information for setting national regulatory or environmental standards and managing integrated watershed areas.