Intraseasonal variation of E. coli and environmental covariates in two irrigation ponds in Maryland, USA

Using machine learning models to estimate Escherichia coli concentration in an irrigation pond from water quality and drone-based RGB imagery data

The rapid and efficient quantification of Escherichia coli concentrations is crucial for monitoring water quality. Remote sensing techniques and machine learning algorithms have been used to detect E. coli in water and estimate its concentrations. The application of these approaches, however, is challenged by limited sample availability and unbalanced water quality datasets. In this study, we estimated the E. coli concentration in an irrigation pond in Maryland, USA, during the summer season using demosaiced natural color (red, green, and blue: RGB) imagery in the visible and infrared spectral ranges, and a set of 14 water quality parameters. We did this by deploying four machine learning models - Random Forest (RF), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGB), and K-nearest Neighbor (KNN) - under three data utilization scenarios: water quality parameters only, combined water quality and small unmanned aircraft system (sUAS)-based RGB data, and RGB data only. To select the training and test datasets, we applied two data-splitting methods: ordinary and quantile data splitting. These methods provided a constant splitting ratio in each decile of the E. coli concentration distribution. Quantile data splitting resulted in better model performance metrics and smaller differences between the metrics for both the training and testing datasets. When trained with quantile data splitting after hyperparameter optimization, models RF, GBM, and XGB had R2 values above 0.847 for the training dataset and above 0.689 for the test dataset. The combination of water quality and RGB imagery data resulted in a higher R2 value (>0.896) for the test dataset. Shapley additive explanations (SHAP) of the relative importance of variables revealed that the visible blue spectrum intensity and water temperature were the most influential parameters in the RF model. Demosaiced RGB imagery served as a useful predictor of E. coli concentration in the studied irrigation pond.

Can water composition and weather factors predict fecal indicator bacteria removal in retention ponds in variable weather conditions?

Retention ponds provide benefits including flood control, groundwater recharge, and water quality improvement, but changes in weather conditions could limit the effectiveness in improving microbial water quality metrics. The concentration of fecal indicator bacteria (FIB), which is used as regulatory standards to assess microbial water quality in retention ponds, could vary widely based on many factors including local weather and influent water chemistry and composition. In this critical review, we analyzed 7421 data collected from 19 retention ponds across North America listed in the International Stormwater BMP Database to examine if variable FIB removal in the field conditions can be predicted based on changes in these weather and water composition factors. Our analysis confirms that FIB removal in retention ponds is sensitive to weather conditions or seasons, but temperature and precipitation data may not describe the variable FIB removal. These weather conditions affect suspended solid and nutrient concentrations, which in turn could affect FIB concentration in the ponds. Removal of total suspended solids and total P only explained 5% and 12% of FIB removal data, respectively, and TN removal had no correlation with FIB removal. These results indicate that regression-based modeling with a single parameter as input has limited use to predict FIB removal due to the interactive nature of their effects on FIB removal. In contrast, machine learning algorithms such as the random forest method were able to predict 65% of the data. The overall analysis indicates that the machine learning model could play a critical role in predicting microbial water quality of surface waters under complex conditions where the variation of both water composition and weather conditions could deem regression-based modeling less effective.

Intra-daily variation of Escherichia coli concentrations in agricultural irrigation ponds

Microbial water quality is determined by comparing observed Escherichia coli concentrations with regulatory thresholds. Measured concentrations can be expected to change throughout the course of a day in response to diurnal variation in environmental conditions, such as solar radiation and temperature. Therefore, the time of day at which samples are taken is an important factor within microbial water quality measurements. However, little is known about the diurnal variations of E. coli concentrations in surface sources of irrigation water. The objectives of this work were to evaluate the intra-daily dynamics of E. coli in three irrigation ponds in Maryland over several years and to determine the water quality parameters to which E. coli populations are most sensitive. Water sampling was conducted across the ponds at 0900, 1200, and 1500 h on a total of 17 dates in the summers of 2019-2021. One-way ANOVA revealed significant diurnal variability in E. coli concentrations in Pond (P)1 and P2, whereas no significant effects were observed in P3. Escherichia coli die-off rates calculated between sampling time points in the same day were significantly higher in P2 than in P1 and P3, and these rates ranged from 0.005 to 0.799 h^-1 across ponds. Concentrations of dissolved oxygen, pH, conductivity, and turbidity exerted the most control over E. coli populations. Results of this work demonstrate that sampling in the early-morning hours provides the most conservative assessment of the microbial quality of irrigation waters.

A preliminary evaluation of microbial water quality in the irrigation pond

This study aimed to determine the microbial water quality of Imrahor Pond by enumerating the coliform bacteria levels in the area. Water samples were collected biweekly from the surface and bottom waters at seven points in the pond. Samples were analyzed for total coliforms, Escherichia coli, physicochemical parameters (water temperature, conductivity, pH, turbidity, dissolved oxygen, nitrate) and 1-day rainfall. The average values of TC and E. coli were 1487.4 and 36.3 MPN/100 ml, respectively. TC concentrations/physicochemical parameters were met at least 2nd class water quality class and E. coli results were met "guideline value" (E. coli < 250 MPN/100 ml) of national regulation. Overall, among measured physicochemical parameters, rainfall had the strongest positive correlation (r = 0.377 for total coliforms and r = 0.466 for E. coli, p < 0.05) with both indicators, indicted that surface runoff due to rainfall is the main factor which effects microbial water quality in the study area. This study demonstrated the preliminary microbial water quality results (TC and E. coli) in the Imrahor Pond and can serve as a basis for developing more precise water quality monitoring and management studies in the future. PRACTITIONER POINTS: Prevalence of TC and E. coli in the surface and bottom waters of Imrahor Pond were investigated for the first time. Imrahor Pond was met guideline value of national regulations based on E. coli concentrations, in the study period. Surface runoff after rainfall was the main environmental factor which influenced the microbial water quality of the pond.

Persistence of E. coli O157:H7 in urban recreational waters from Spring and Autumn: a comparison analysis

People might get infected by pathogens found in urban recreational waters during water-contact activities, such as swimming, boating, bathing, and yachting. However, the persistence of pathogenic bacteria in those waters was not well documented. In this study, persistence of E. coli O157:H7 (EcO157) in 48 water samples (24 Spring samples and 24 Autumn samples) from the 3 urban recreational waters was investigated. Multivariate statistical analysis was performed to correlate survival data with water physicochemical properties and bacterial communities. Our data showed that EcO157 survived longer in Spring samples than in Autumn samples regardless of the lakes. Results revealed that recreational water physicochemical properties and bacterial community in Spring samples were different from those in Autumn samples. Mantel and Partial Mantel tests, as well as co-occurrence network analysis illustrated that EC salinity, TOC, and bacterial community were correlated with survival time (ttd) (p < 0.05). Variation partition analysis (VPA) indicated that bacterial community, EC, TOC, and TN explained about 64.81% of overall ttd variation in Spring samples, and bacterial community, EC, pH, and TP accounted for about 56.59% of overall ttd variation in Autumn samples. Structural equation model (SEM) illustrated that EC indirectly positively affected ttd through bacterial community. The correlation between bacterial community and ttd was negative in Spring samples and positive in Autumn samples. TN appeared a direct positive effect on ttd in Spring samples. TP displayed a direct negative effect on ttd in Autumn samples. Our results concluded that there was seasonal variation in environmental factors that directly or indirectly affected the survival of EcO157 in urban recreational waters.

Prediction of E. coli Concentrations in Agricultural Pond Waters: Application and Comparison of Machine Learning Algorithms

The microbial quality of irrigation water is an important issue as the use of contaminated waters has been linked to several foodborne outbreaks. To expedite microbial water quality determinations, many researchers estimate concentrations of the microbial contamination indicator Escherichia coli (E. coli) from the concentrations of physiochemical water quality parameters. However, these relationships are often non-linear and exhibit changes above or below certain threshold values. Machine learning (ML) algorithms have been shown to make accurate predictions in datasets with complex relationships. The purpose of this work was to evaluate several ML models for the prediction of E. coli in agricultural pond waters. Two ponds in Maryland were monitored from 2016 to 2018 during the irrigation season. E. coli concentrations along with 12 other water quality parameters were measured in water samples. The resulting datasets were used to predict E. coli using stochastic gradient boosting (SGB) machines, random forest (RF), support vector machines (SVM), and k-nearest neighbor (kNN) algorithms. The RF model provided the lowest RMSE value for predicted E. coli concentrations in both ponds in individual years and over consecutive years in almost all cases. For individual years, the RMSE of the predicted E. coli concentrations (log10 CFU 100 ml-1) ranged from 0.244 to 0.346 and 0.304 to 0.418 for Pond 1 and 2, respectively. For the 3-year datasets, these values were 0.334 and 0.381 for Pond 1 and 2, respectively. In most cases there was no significant difference (P > 0.05) between the RMSE of RF and other ML models when these RMSE were treated as statistics derived from 10-fold cross-validation performed with five repeats. Important E. coli predictors were turbidity, dissolved organic matter content, specific conductance, chlorophyll concentration, and temperature. Model predictive performance did not significantly differ when 5 predictors were used vs. 8 or 12, indicating that more tedious and costly measurements provide no substantial improvement in the predictive accuracy of the evaluated algorithms.

Evaluation of E. coli in sediment for assessing irrigation water quality using machine learning

Fresh produce irrigated with contaminated water poses a substantial risk to human health. This study evaluated the impact of incorporating sediment information on improving the performance of machine learning models to quantify E. coli level in irrigation water. Field samples were collected from irrigation canals in the Southwest U.S., for which meteorological, chemical, and physical water quality variables as well as three additional flow and sediment properties: the concentration of E. coli in sediment, sediment median size, and bed shear stress. Water quality was classified based on E. coli concentration exceeding two standard levels: 1 E. coli and 126 E. coli colony forming units (CFU) per 100 ml of irrigation water. Two series of features, including (FIS) and excluding (FES) sediment features, were selected using multi-variant filter feature selection. The correlation analysis revealed the inclusion of sediment features improves the correlation with the target standards for E. coli compared to the models excluding these features. Support vector machine, logistic regression, and ridge classifier were tested in this study. The support vector machine model performed the best for both targeted standards. Besides, incorporating sediment features improved all models' performance. Therefore, the concentration of E. coli in sediment and bed shear stress are major factors influencing E. coli concentration in irrigation water.

Data assimilation in surface water quality modeling: A review

Data assimilation (DA) techniques are powerful means of dynamic natural system modeling that allow for the use of data as soon as it appears to improve model predictions and reduce prediction uncertainty by correcting state variables, model parameters, and boundary and initial conditions. The objectives of this review are to explore existing approaches and advances in DA applications for surface water quality modeling and to identify future research prospects. We first reviewed the DA methods used in water quality modeling as reported in literature. We then addressed observations and suggestions regarding various factors of DA performance, such as the mismatch between both lateral and vertical spatial detail of measurements and modeling, subgrid heterogeneity, presence of temporally stable spatial patterns in water quality parameters and related biases, evaluation of uncertainty in data and modeling results, mismatch between scales and schedules of data from multiple sources, selection of parameters to be updated along with state variables, update frequency and forecast skill. The review concludes with the outlook section that outlines current challenges and opportunities related to growing role of novel data sources, scale mismatch between model discretization and observation, structural uncertainty of models and conversion of measured to simulated vales, experimentation with DA prior to applications, using DA performance or model selection, the role of sensitivity analysis, and the expanding use of DA in water quality management.

Intraseasonal variation of phycocyanin concentrations and environmental covariates in two agricultural irrigation ponds in Maryland, USA

Recently, cyanobacteria blooms have become a concern for agricultural irrigation water quality. Numerous studies have shown that cyanotoxins from these harmful algal blooms (HABs) can be transported to and assimilated into crops when present in irrigation waters. Phycocyanin is a pigment known only to occur in cyanobacteria and is often used to indicate cyanobacteria presence in waters. The objective of this work was to identify the most influential environmental covariates affecting the phycocyanin concentrations in agricultural irrigation ponds that experience cyanobacteria blooms of the potentially toxigenic species Microcystis and Aphanizomenon using machine learning methodology. The study was performed at two agricultural irrigation ponds over a 5-month period in the summer of 2018. Phycocyanin concentrations, along with sensor-based and fluorometer-based water quality parameters including turbidity (NTU), pH, dissolved oxygen (DO), fluorescent dissolved organic matter (fDOM), conductivity, chlorophyll, color dissolved organic matter (CDOM), and extracted chlorophyll were measured. Regression tree analyses were used to determine the most influential water quality parameters on phycocyanin concentrations. Nearshore sampling locations had higher phycocyanin concentrations than interior sampling locations and "zones" of consistently higher concentrations of phycocyanin were found in both ponds. The regression tree analyses indicated extracted chlorophyll, CDOM, and NTU were the three most influential parameters on phycocyanin concentrations. This study indicates that sensor-based and fluorometer-based water quality parameters could be useful to identify spatial patterns of phycocyanin concentrations and therefore, cyanobacteria blooms, in agricultural irrigation ponds and potentially other water bodies.

Hyperspectral Imaging from a Multipurpose Floating Platform to Estimate Chlorophyll-a Concentrations in Irrigation Pond Water

This study provides detailed information about the use of a hyperspectral imaging system mounted on a motor-driven multipurpose floating platform (MFP) for water quality sensing and water sampling, including the spatial and spectral calibration for the camera, image acquisition and correction procedures. To evaluate chlorophyll-a concentrations in an irrigation pond, visible/near-infrared hyperspectral images of the water were acquired as the MFP traveled to ten water sampling locations along the length of the pond, and dimensionality reduction with correlation analysis was performed to relate the image data to the measured chlorophyll-a data. About 80,000 sample images were acquired by the line-scan method. Image processing was used to remove sun-glint areas present in the raw hyperspectral images before further analysis was conducted by principal component analysis (PCA) to extract three key wavelengths (662 nm, 702 nm, and 752 nm) for detecting chlorophyll-a in irrigation water. Spectral intensities at the key wavelengths were used as inputs to two near-infrared (NIR)-red models. The determination coefficients (R2) of the two models were found to be about 0.83 and 0.81. The results show that hyperspectral imagery from low heights can provide valuable information about water quality in a fresh water source.

Accounting for the Three-Dimensional Distribution of Escherichia coli Concentrations in Pond Water in Simulations of the Microbial Quality of Water Withdrawn for Irrigation

Evaluating the microbial quality of irrigation water is essential for the prevention of foodborne illnesses. Generic Escherichia coli (E. coli) is used as an indicator organism to estimate the microbial quality of irrigation water. Monitoring E. coli concentrations in irrigation water sources is commonly performed using water samples taken from a single depth. Vertical gradients of E. coli concentrations are typically not measured or are ignored; however, E. coli concentrations in water bodies can be expected to have horizontal and vertical gradients. The objective of this work was to research 3D distributions of E. coli concentrations in an irrigation pond in Maryland and to estimate the dynamics of E. coli concentrations at the water intake during the irrigation event using hydrodynamic modeling in silico. The study pond is about 22 m wide and 200 m long, with an average depth of 1.5 m. Three transects sampled at 50-cm depth intervals, along with intensive nearshore sampling, were used to develop the initial concentration distribution for the application of the environmental fluid dynamic code (EFDC) model. An eight-hour irrigation event was simulated using on-site data on the wind speed and direction. Substantial vertical and horizontal variations in E. coli concentrations translated into temporally varying concentrations at the intake. Additional simulations showed that the E. coli concentrations at the intake reflect the 3D distribution of E. coli in the limited pond section close to the intake. The 3D sampling revealed E. coli concentration hot spots at different depths across the pond. Measured and simulated 3D E. coli concentrations provide improved insights into the expected microbial water quality of irrigation water compared with 1D or 2D representations of the spatial variability of the indicator concentration.

Prevalence of Salmonella and Listeria monocytogenes in non-traditional irrigation waters in the Mid-Atlantic United States is affected by water type, season, and recovery method

Irrigation water contaminated with Salmonella enterica and Listeria monocytogenes may provide a route of contamination of raw or minimally processed fruits and vegetables. While previous work has surveyed specific and singular types of agricultural irrigation water for bacterial pathogens, few studies have simultaneously surveyed different water sources repeatedly over an extended period of time. This study quantified S. enterica and L. monocytogenes levels (MPN/L) at 6 sites, including river waters: tidal freshwater river (MA04, n = 34), non-tidal freshwater river, (MA05, n = 32), one reclaimed water holding pond (MA06, n = 25), two pond water sites (MA10, n = 35; MA11, n = 34), and one produce wash water site (MA12, n = 10) from September 2016—October 2018. Overall, 50% (84/168) and 31% (53/170) of sampling events recovered S. enterica and L. monocytogenes, respectively. Results showed that river waters supported significantly (p < 0.05) greater levels of S. enterica than pond or reclaimed waters. The non-tidal river water sites (MA05) with the lowest water temperature supported significantly greater level of L. monocytogenes compared to all other sites; L. monocytogenes levels were also lower in winter and spring compared to summer seasons. Filtering 10 L of water through a modified Moore swab (MMS) was 43.5 (Odds ratio, p < 0.001) and 25.5 (p < 0.001) times more likely to recover S. enterica than filtering 1 L and 0.1 L, respectively; filtering 10 L was 4.8 (p < 0.05) and 3.9 (p < 0.05) times more likely to recover L. monocytogenes than 1L and 0.1 L, respectively. Work presented here shows that S. enterica and L. monocytogenes levels are higher in river waters compared to pond or reclaimed waters in the Mid-Atlantic region of the U.S., and quantitatively shows that analyzing 10 L water is more likely recover pathogens than smaller samples of environmental waters.