Modern space/time geostatistics using river distances: data integration of turbidity and E. coli measurements to assess fecal contamination along the Raritan River in New Jersey

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.

Reduction of pathogens in greywater with biological and sustainable treatments selected through a multicriteria approach

Non-potable reuse of greywater (GW) can represent a valid alternative to freshwater consumption, satisfying the Sustainable Development Goals promoted by United Nations. The Multi-Criteria Analysis (MCA) was applied to select the most suitable processes for the reduction of microbiological contamination in GW. A pilot plant, including horizontal flow constructed wetland (CW) and anaerobic filtration (AF) in parallel, best treatment options according to MCA results, was built to treat GW collected from a Venezuelan family. (i) The removal efficiency of microbiological parameters, and (ii) the turbidity as possible microbiological contamination indicator and possible influence factor of disinfection treatment, were investigated. Except for Escherichia coli (4.1 ± 0.9 log reduction with AF), CW achieved the best reductions yields for total coliforms, faecal coliforms, and Salmonella, respectively equal to 3.1 ± 0.5 log, 4.3 ± 0.5 log, and 2.9 ± 0.4 log. In accordance with Venezuelan legislation and WHO guidelines, GW treated with CW was found to be suitable for irrigation reuse for non-edible crops. However, the reduction of pathogens by CW should be considered as a preliminary and not complete disinfection treatment. To reuse GW, especially in the irrigation of edible crops, stronger disinfection treatment should be considered as a complement to the preliminary disinfection performed by CW, to avoid any kind of risk. No significant correlation was found for turbidity either as a possible predictor of microbiological contamination or as an influence on biological disinfection.

Microbial Find, Inform, and Test Model for Identifying Spatially Distributed Contamination Sources: Framework Foundation and Demonstration of Ruminant Bacteroides Abundance in River Sediments

Microbial pollution in rivers poses known ecological and health risks, yet causal and mechanistic linkages to sources remain difficult to establish. Host-associated microbial source tracking (MST) markers help to assess the microbial risks by linking hosts to contamination but do not identify the source locations. Land-use regression (LUR) models have been used to screen the source locations using spatial predictors but could be improved by characterizing transport (i.e., hauling, decay overland, and downstream). We introduce the microbial Find, Inform, and Test (FIT) framework, which expands previous LUR approaches and develops novel spatial predictor models to characterize the transported contributions. We applied FIT to characterize the sources of BoBac, a ruminant Bacteroides MST marker, quantified in riverbed sediment samples from Kewaunee County, Wisconsin. A 1 standard deviation increase in contributions from land-applied manure hauled from animal feeding operations (AFOs) was associated with a 77% (p-value <0.05) increase in the relative abundance of ruminant Bacteroides (BoBac-copies-per-16S-rRNA-copies) in the sediment. This is the first work finding an association between the upstream land-applied manure and the offsite bovine-associated fecal markers. These findings have implications for the sediment as a reservoir for microbial pollution associated with AFOs (e.g., pathogens and antibiotic-resistant bacteria). This framework and application advance statistical analysis in MST and water quality modeling more broadly.

Effect of different types of anthropogenic pollution on the bacterial community of urban rivers

The health of urban rivers is threatened by multiple anthropogenic stressors. Bacterial communities in rivers can quickly respond to different types of polluted environments, making them useful for water quality assessments and predictive insights. However, research on river bacterial communities has largely ignored interactions between these communities. Here, 16S rRNA amplicon sequencing analysis is used to comprehensively analyze the bacterial communities in the water and sediments in different types of anthropogenically impacted urban river. The results show that distinct differences occur in the bacterial communities in the river sediment and water with different pollution types. The changes in the bacterial communities in sediments were more pronounced than those in the water. A modular analysis further showed that the microbial co-occurrence network under different types of pollution had a nonrandom modular structure, and this structure was mainly driven by classification correlation and bacterial function. Genes identified for nitrogen cycling in all the river water and sediment samples included major functional genes for nitrogen fixation, assimilatory nitrogen reduction, nitrification, denitrification, and ammonification. Carbon degradation genes were mainly observed in the carbon cycle. Taken together, the above findings provide further insights into microbial communities in urban river ecosystems under anthropogenic contamination. PRACTITIONER POINTS: The physical and chemical indicators of the four types of pollution drive bacterial community structure. Bacterial community has C, N, P metabolic genes indicating its ecological effect. River bacteria were connected more frequently in the same or similar type of pollution in the co-occurrence network. Microbe-environment correlations and microbe-microbe interactions were combined to determine crucial indicators.

Interpretability Versus Accuracy: A Comparison of Machine Learning Models Built Using Different Algorithms, Performance Measures, and Features to Predict E. coli Levels in Agricultural Water

Since E. coli is considered a fecal indicator in surface water, government water quality standards and industry guidance often rely on E. coli monitoring to identify when there is an increased risk of pathogen contamination of water used for produce production (e.g., for irrigation). However, studies have indicated that E. coli testing can present an economic burden to growers and that time lags between sampling and obtaining results may reduce the utility of these data. Models that predict E. coli levels in agricultural water may provide a mechanism for overcoming these obstacles. Thus, this proof-of-concept study uses previously published datasets to train, test, and compare E. coli predictive models using multiple algorithms and performance measures. Since the collection of different feature data carries specific costs for growers, predictive performance was compared for models built using different feature types [geospatial, water quality, stream traits, and/or weather features]. Model performance was assessed against baseline regression models. Model performance varied considerably with root-mean-squared errors and Kendall's Tau ranging between 0.37 and 1.03, and 0.07 and 0.55, respectively. Overall, models that included turbidity, rain, and temperature outperformed all other models regardless of the algorithm used. Turbidity and weather factors were also found to drive model accuracy even when other feature types were included in the model. These findings confirm previous conclusions that machine learning models may be useful for predicting when, where, and at what level E. coli (and associated hazards) are likely to be present in preharvest agricultural water sources. This study also identifies specific algorithm-predictor combinations that should be the foci of future efforts to develop deployable models (i.e., models that can be used to guide on-farm decision-making and risk mitigation). When deploying E. coli predictive models in the field, it is important to note that past research indicates an inconsistent relationship between E. coli levels and foodborne pathogen presence. Thus, models that predict E. coli levels in agricultural water may be useful for assessing fecal contamination status and ensuring compliance with regulations but should not be used to assess the risk that specific pathogens of concern (e.g., Salmonella, Listeria) are present.

Soft sensor predictor of E. coli concentration based on conventional monitoring parameters for wastewater disinfection control

Real-time acquisition of indicator bacteria concentration at the inlet of disinfection unit is a fundamental support to the control of chemical and ultraviolet wastewater disinfection. Culture-based enumeration methods need time-consuming laboratory analyses, which give results after several hours or days, while newest biosensors rarely provide information about specific strains and outputs are not directly comparable with regulatory limits as a consequence of measurement principles. In this work, a novel soft sensor approach for virtual real-time monitoring of E. coli concentration is proposed. Conventional wastewater physical and chemical indicators (chemical oxygen demand, total nitrogen, nitrate, ammonia, total suspended solids, conductivity, pH, turbidity and absorbance at 254 nm) and flowrate were studied as potential predictors of E. coli concentration relying on data collected from three full-scale wastewater treatment plants. Different methods were compared: (i) linear modeling via ordinary least squares; (ii) ridge regression; (iii) principal component regression and partial least squares; (iv) non-linear modeling through artificial neural networks. Linear soft sensors reached some degree of accuracy, but performances of the artificial neural network based models were by far superior. Sensitivity analysis allowed to prioritize the importance of each predictor and to highlight the site-specific nature of the approach, because of the site-specific nature of relationships between predictors and E. coli concentration. In one case study, pH and conductivity worked as good proxy variables when the occurrence of intense rain events caused sharp increases in E. coli concentration. Differently, in other case studies, chemical oxygen demand, total suspended solids, turbidity and absorbance at 254 nm accounted for the positive correlation between low wastewater quality and E. coli concentration. Moreover, sensitivity analysis of artificial neural network models highlighted the importance of interactions among predictors, contributing to 25 to 30% of the model output variance. This evidence, along with performance results, supported the idea that nonlinear families of models should be preferred in the estimation of E. coli concentration. The artificial neural network based soft sensor deployment for control of peracetic acid disinfectant dosage was simulated over a realistic scenario of wastewater quality recorded by on-line sensors over 2 months. The scenario simulations highlighted the significant benefit of an E. coli soft sensor, which provided up to 57% of disinfectant saving.

Demonstration of Tryptophan-Like Fluorescence Sensor Concepts for Fecal Exposure Detection in Drinking Water in Remote and Resource Constrained Settings

Low-cost, field-deployable, near-time methods for assessing water quality are not available when and where waterborne infection risks are greatest. We describe the development and testing of a novel device for the measurement of tryptophan-like fluorescence (TLF), making use of recent advances in deep-ultraviolet light emitting diodes (UV-LEDs) and sensitive semiconductor photodiodes and photomultipliers. TLF is an emerging indicator of water quality that is associated with members of the coliform group of bacteria and therefore potential fecal contamination. Following the demonstration of close correlation between TLF and E. coli in model waters and proof of principle with sensitivity of 4 CFU/mL for E. coli, we further developed a two-LED flow-through configuration capable of detecting TLF levels corresponding to “high risk” fecal contamination levels (>10 CFU/100 mL). Findings to date suggest that this device represents a scalable solution for remote monitoring of drinking water supplies to identify high-risk drinking water in near-time. Such information can be immediately actionable to reduce risks.

A space-time geostatistical model for probabilistic estimation of harmful algal bloom biomass and areal extent

Harmful algal blooms (HABs) have been increasing in intensity worldwide, including the western basin of Lake Erie. Substantial efforts have been made to track these blooms using in situ sampling and remote sensing. However, such measurements do not fully capture HAB spatial and temporal dynamics due to the limitations of discrete shipboard sampling over large areas and the effects of clouds and winds on remote sensing estimates. To address these limitations, we develop a space-time geostatistical modeling framework for estimating HAB intensity and extent using chlorophyll a data sampled during the HAB season (June-October) from 2008 to 2017 by five independent monitoring programs. Based on the Bayesian information criterion for model selection, trend variables explain bloom northerly and easterly expansion from Maumee Bay, wind effects over depth, and variability among sampling methods. Cross validation results demonstrate that space-time kriging explains over half of the variability in daily, location-specific chlorophyll observations, on average. Conditional simulations provide, for the first time, comprehensive estimates of overall bloom biomass (based on depth-integrated concentrations) and surface areal extent with quantified uncertainties. These new estimates are contrasted with previous Lake Erie HAB monitoring studies, and deviations among estimates are explored and discussed. Overall, results highlight the importance of maintaining sufficient monitoring coverage to capture bloom dynamics, as well as the benefits of the proposed approach for synthesizing data from multiple monitoring programs to improve estimation accuracy while reducing uncertainty.

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.

Spatial dynamic assessment of health risks for urban river cruises

River cruising ships move along river courses, and thus health risks to passengers may vary spatially due to the accidental exposure of river fecal pollution. This study performed a spatial dynamic assessment of health risks for river cruises in the highly urbanized Tamsui River Basin. First, the spatial distributions of river Escherichia coli (E. coli) were probabilistically characterized using indicator kriging (IK). Moreover, the current river cruise information was surveyed to obtain cruise routes and transit times. Then, to explore the parametric uncertainty of quantitative microbial risk assessment (QMRA), the ingestion rate (IR) for boating was determined using Monte Carlo simulation (MCS). Moreover, river E. coli distributions were estimated using nonparametric MCS according to multi-threshold IK estimates. Eventually, after combining the distribution of the joint probability of the IR and E. coli in QMRA, the β-Poisson dose-response function was adopted to analyze risks to river cruise passengers at discretized segments of cruise routes. Health risks to river cruise passengers were integrated at the discretized segments to explore suitable recreational strategies for river cruises. The research results indicate that all health risks do not exceed a daily target level of 8 illnesses per 1000 exposures for single-trip cruise routes. However, health risks to passengers can exceed this level for round-trip cruise routes along highly polluted urban river courses.

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.

Geostatistical Prediction of Microbial Water Quality Throughout a Stream Network Using Meteorology, Land Cover, and Spatiotemporal Autocorrelation

Predictive modeling is promising as an inexpensive tool to assess water quality. We developed geostatistical predictive models of microbial water quality that empirically modeled spatiotemporal autocorrelation in measured fecal coliform (FC) bacteria concentrations to improve prediction. We compared five geostatistical models featuring different autocorrelation structures, fit to 676 observations from 19 locations in North Carolina's Jordan Lake watershed using meteorological and land cover predictor variables. Though stream distance metrics (with and without flow-weighting) failed to improve prediction over the Euclidean distance metric, incorporating temporal autocorrelation substantially improved prediction over the space-only models. We predicted FC throughout the stream network daily for one year, designating locations "impaired", "unimpaired", or "unassessed" if the probability of exceeding the state standard was ≥90%, ≤10%, or >10% but <90%, respectively. We could assign impairment status to more of the stream network on days any FC were measured, suggesting frequent sample-based monitoring remains necessary, though implementing spatiotemporal predictive models may reduce the number of concurrent sampling locations required to adequately assess water quality. Together, these results suggest that prioritizing sampling at different times and conditions using geographically sparse monitoring networks is adequate to build robust and informative geostatistical models of water quality impairment.

Faecal contamination and visual clarity in New Zealand rivers: correlation of key variables affecting swimming suitability

Swimming is a popular activity in Aotearoa-New Zealand (NZ). Two variables that strongly influence swimming suitability of waters are faecal contamination, as indicated by the bacterium Escherichia coli, and visual clarity as it affects aesthetics and safety with respect to submerged hazards. We show that E. coli and visual clarity are inversely related overall in NZ rivers (R = -0.54), and more strongly related in many individual rivers, while similar (but positive) correlations apply also to turbidity. This finding, apparently reflecting co-mobilisation of faecal contamination and fine sediment, suggests that visual clarity, measured or estimated from appearance of submerged features, should be a valuable indicator of faecal contamination status and (more generally) swimming suitability. If swimmers were to avoid river waters <1.6 m black disc visibility (a long-established NZ guideline for swimming) they would also avoid microbial hazards (E. coli <550 cfu/100 mL about 99% of the time in NZ rivers). However, urban-affected rivers might sometimes be microbially contaminated when still clear. Water management agencies should measure visual clarity together with E. coli in river surveillance. Real-time information on swimming suitability could then be based on continuous monitoring of turbidity locally calibrated to both visual clarity and E. coli.

Using spatial-stream-network models and long-term data to understand and predict dynamics of faecal contamination in a mixed land-use catchment

An 11year dataset of concentrations of E. coli at 10 spatially-distributed sites in a mixed land-use catchment in NE Scotland (52km²) revealed that concentrations were not clearly associated with flow or season. The lack of a clear flow-concentration relationship may have been due to greater water fluxes from less-contaminated headwaters during high flows diluting downstream concentrations, the importance of persistent point sources of E. coli both anthropogenic and agricultural, and possibly the temporal resolution of the dataset. Point sources and year-round grazing of livestock probably obscured clear seasonality in concentrations. Multiple linear regression models identified potential for contamination by anthropogenic point sources as a significant predictor of long-term spatial patterns of low, average and high concentrations of E. coli. Neither arable nor pasture land was significant, even when accounting for hydrological connectivity with a topographic-index method. However, this may have reflected coarse-scale land-cover data inadequately representing "point sources" of agricultural contamination (e.g. direct defecation of livestock into the stream) and temporal changes in availability of E. coli from diffuse sources. Spatial-stream-network models (SSNMs) were applied in a novel context, and had value in making more robust catchment-scale predictions of concentrations of E. coli with estimates of uncertainty, and in enabling identification of potential "hot spots" of faecal contamination. Successfully managing faecal contamination of surface waters is vital for safeguarding public health. Our finding that concentrations of E. coli could not clearly be associated with flow or season may suggest that management strategies should not necessarily target only high flow events or summer when faecal contamination risk is often assumed to be greatest. Furthermore, we identified SSNMs as valuable tools for identifying possible "hot spots" of contamination which could be targeted for management, and for highlighting areas where additional monitoring could help better constrain predictions relating to faecal contamination.

Assessment of carbon nanotubes and silver nanoparticles loaded clays as adsorbents for removal of bacterial contaminants from water sources

This work evaluated the antimicrobial efficacy of kaolin clay and its loaded forms with carbon nanotubes (CNTs) and silver nanoparticles (AgNPs) against bacterial isolates from different water supplies (tap, underground and surface water) in addition to wastewater. A total of 160 water samples were collected from different water sources in the investigated districts. Samples were cultured for isolation and serological identification of pathogenic bacteria. AgNPs were synthesized by a typical one-step synthesis protocol, where CNTs were carried out in a reactor employing the double bias-assisted hot filament chemical vapor deposition method. Both were characterized using transmission electron microscopy, infrared and X-ray fluorescence (XRF) spectroscopy. The antimicrobial efficacy of each of natural kaolin clay, AgNPs- and CNTs-loaded clays were evaluated by their application in four concentrations (0.01, 0.03, 0.05 and 0.1 ppm) at different contact times (5 min, 15 min, 30 min and 2 h). AgNPs-loaded clays at concentrations of 0.05 and 0.1 mg/l for 2 h contact time exhibited a higher bactericidal efficacy on Escherichia coli and Salmonella spp. (70, 70, 80 and 90%, respectively) compared to CNTs-loaded clay. Concluding, the application of AgNPs-loaded clay for removal of water bacterial contaminants at a concentration of 0.1 ppm for 2 h contact times resulted in highly effective removals.

Bayesian Maximum Entropy space/time estimation of surface water chloride in Maryland using river distances

Widespread contamination of surface water chloride is an emerging environmental concern. Consequently accurate and cost-effective methods are needed to estimate chloride along all river miles of potentially contaminated watersheds. Here we introduce a Bayesian Maximum Entropy (BME) space/time geostatistical estimation framework that uses river distances, and we compare it with Euclidean BME to estimate surface water chloride from 2005 to 2014 in the Gunpowder-Patapsco, Severn, and Patuxent subbasins in Maryland. River BME improves the cross-validation R2 by 23.67% over Euclidean BME, and river BME maps are significantly different than Euclidean BME maps, indicating that it is important to use river BME maps to assess water quality impairment. The river BME maps of chloride concentration show wide contamination throughout Baltimore and Columbia-Ellicott cities, the disappearance of a clean buffer separating these two large urban areas, and the emergence of multiple localized pockets of contamination in surrounding areas. The number of impaired river miles increased by 0.55% per year in 2005-2009 and by 1.23% per year in 2011-2014, corresponding to a marked acceleration of the rate of impairment. Our results support the need for control measures and increased monitoring of unassessed river miles.

Geospatial estimation of the impact of Deepwater Horizon oil spill on plant oiling along the Louisiana shorelines

Stranded oil covering soil and plant stems in fragile Louisiana marshes was one of the most visible impacts of the 2010 Deepwater Horizon (DWH) oil spill. As part of the assessment of marsh injury after the DWH spill, plant stem oiling was broken into five categories (0%, 0-10%, 10-50%, 50-90%, 90-100%) and used as the independent variable for estimating death of vegetation, accelerated erosion, and other metrics of injury. The length of shoreline falling into each of these stem oiling categories was therefore a key measure of the total extent of marsh injury, and its accurate estimation is the focus of this paper. First, we used geographically-weighted logistic regression (GWR) to explore and model spatially varying relationships between stem oiling field data and secondary information (oiling exposure category) collected during shoreline surveys. We then combined GWR probability estimates with field data using indicator cokriging to predict the probability of exceeding four stem oiling thresholds (0, 10, 50, and 90%) at 50 m intervals along the Louisiana shoreline. Cross-validation using Receiver Operating Characteristic (ROC) Curves demonstrate the greater prediction accuracy of the multivariate geostatistical approach relative to either aspatial regression or indicator kriging that ignores secondary information.

Estimating Escherichia coli loads in streams based on various physical, chemical, and biological factors

Microbes have been identified as a major contaminant of water resources. Escherichia coli (E. coli) is a commonly used indicator organism. It is well recognized that the fate of E. coli in surface water systems is governed by multiple physical, chemical, and biological factors. The aim of this work is to provide insight into the physical, chemical, and biological factors along with their interactions that are critical in the estimation of E. coli loads in surface streams. There are various models to predict E. coli loads in streams, but they tend to be system or site specific or overly complex without enhancing our understanding of these factors. Hence, based on available data, a Bayesian Neural Network (BNN) is presented for estimating E. coli loads based on physical, chemical, and biological factors in streams. The BNN has the dual advantage of overcoming the absence of quality data (with regards to consistency in data) and determination of mechanistic model parameters by employing a probabilistic framework. This study evaluates whether the BNN model can be an effective alternative tool to mechanistic models for E. coli loads estimation in streams. For this purpose, a comparison with a traditional model (LOADEST, USGS) is conducted. The models are compared for estimated E. coli loads based on available water quality data in Plum Creek, Texas. All the model efficiency measures suggest that overall E. coli loads estimations by the BNN model are better than the E. coli loads estimations by the LOADEST model on all the three occasions (three-fold cross validation). Thirteen factors were used for estimating E. coli loads with the exhaustive feature selection technique, which indicated that six of thirteen factors are important for estimating E. coli loads. Physical factors included temperature and dissolved oxygen; chemical factors include phosphate and ammonia; biological factors include suspended solids and chlorophyll. The results highlight that the LOADEST model estimates E. coli loads better in the smaller ranges, whereas the BNN model estimates E. coli loads better in the higher ranges. Hence, the BNN model can be used to design targeted monitoring programs and implement regulatory standards through TMDL programs.

Using river distance and existing hydrography data can improve the geostatistical estimation of fish tissue mercury at unsampled locations

Mercury in fish tissue is a major human health concern. Consumption of mercury-contaminated fish poses risks to the general population, including potentially serious developmental defects and neurological damage in young children. Therefore, it is important to accurately identify areas that have the potential for high levels of bioaccumulated mercury. However, due to time and resource constraints, it is difficult to adequately assess fish tissue mercury on a basin wide scale. We hypothesized that, given the nature of fish movement along streams, an analytical approach that takes into account distance traveled along these streams would improve the estimation accuracy for fish tissue mercury in unsampled streams. Therefore, we used a river-based Bayesian Maximum Entropy framework (river-BME) for modern space/time geostatistics to estimate fish tissue mercury at unsampled locations in the Cape Fear and Lumber Basins in eastern North Carolina. We also compared the space/time geostatistical estimation using river-BME to the more traditional Euclidean-based BME approach, with and without the inclusion of a secondary variable. Results showed that this river-based approach reduced the estimation error of fish tissue mercury by more than 13% and that the median estimate of fish tissue mercury exceeded the EPA action level of 0.3 ppm in more than 90% of river miles for the study domain.