Classification machine learning to detect de facto reuse and cyanobacteria at a drinking water intake.

Harmful algal blooms (HABs) or higher levels of de facto water reuse (DFR) can increase the levels of certain contaminants at drinking water intakes. Therefore, the goal of this study was to use multi-class supervised machine learning (SML) classification with data collected from six online instruments measuring fourteen total water quality parameters to detect cyanobacteria (corresponding to approximately 950 cells/mL, 2900 cells/mL, and 8600 cells/mL) or DFR (0.5, 1 and 2 % of wastewater effluent) events in the raw water entering an intake. Among 56 screened models from the caret package in R, four (mda, LogitBoost, bagFDAGCV, and xgbTree) were selected for optimization. mda had the greatest testing set accuracy, 98.09 %, after optimization with 7 false alerts. Some of the most important water parameters for the different models were phycocyanin-like fluorescence, UVA254, and pH. SML could detect algae blending events (estimated <9000 cells/mL) due in part to the phycocyanin-like fluorescence sensor. UVA254 helped identify higher concentrations of DFR. These results show that multi-class SML classification could be used at drinking water intakes in conjunction with online instrumentation to detect and differentiate HABs and DFR events. This could be used to create alert systems for the water utilities at the intake, rather than the finished water, so any adjustment to the treatment process could be implemented.

Sources of per- and polyfluoroalkyl substances in an arid, urban, wastewater-dominated watershed.

Per- and polyfluoroalkyl substances (PFAS) enter surface waters from various sources such as wastewater treatment plants, fire-fighting sites, and PFAS-producing and PFAS-using industries. The Las Vegas Wash in Southern Nevada of the United States (U.S.) conveys wastewater effluent from the Las Vegas metropolitan area to Lake Mead, a drinking water source for millions of people in the U.S. Southwest. PFAS have previously been detected in the Las Vegas Wash, but PFAS sources were not identified. In this study, upstream wash tributaries, wastewater treatment effluents, and shallow groundwater wells were sampled in multiple campaigns during dry-weather conditions to investigate possible PFAS sources. Out of 19 PFAS, two short-chain PFAS-perfluoropentanoic acid (48 % of the total molar concentration) and perfluorohexanoic acid (32 %)-comprised the majority of PFAS loading measured in the Las Vegas Wash, followed by perfluorooctanoic acid (9 %). On a mass loading basis, the majority of total measured PFAS (approximately 90 %) and at least 48 % of each specific PFAS in the Las Vegas Wash likely entered via municipal wastewater effluents, of which the main source was likely residential wastewater. One of the drainage areas with a major civilian airport was identified as a potential source of relatively enriched perfluorosulfonic acids to a small wash tributary and shallow groundwater samples. Nonetheless, that tributary contributed at most 15 % of any specific PFAS to the mainstem of the Las Vegas Wash. Total PFAS concentrations were relatively low for the small tributary associated with an urban smaller airport and the lack of flow in the tributary channel immediately downgradient of an Air Force base indicates the smaller airport and base were unlikely significant PFAS sources to the Las Vegas Wash. Overall, this study demonstrated effective PFAS source investigation methodology and the importance of wastewater effluent as a PFAS environmental pathway.

Per- and polyfluoroalkyl substances and organofluorine in lakes and waterways of the northwestern Great Basin and Sierra Nevada.

Per- and polyfluoroalkyl substances (PFAS) are anthropogenic chemicals that occur ubiquitously in the environment and have been linked to numerous adverse health effects in humans and aquatic organisms. Although numerous environmental monitoring studies have been conducted, only one has evaluated PFAS in surface waters of the northwestern Great Basin, which features unique topography that results in dozens of endorheic basins and terminal lakes with no natural outlet, where PFAS may accumulate. To close this knowledge gap, we evaluated the occurrence of PFAS in grab samples from 15 lakes (headwater and terminal lakes) and 10 rivers in the Great Basin located in Nevada and California of the United States. PFAS and organofluorine were quantified by liquid chromatography tandem mass spectroscopy (LC-MS/MS) and combustion ion chromatography, respectively. The highest concentrations of PFAS occurred in samples taken near sites with known or suspected prior aqueous film forming foam (AFFF) application (~20 to 4754 ng/L). Samples near wastewater treatment plants and in urban areas also tended to have PFAS concentrations greater than those measured in remote, less anthropogenically influenced areas (~2 to 15 ng/L, <3 ng/L respectively). In limited snapshot sampling events PFAS appeared to accumulate in terminal lakes to some extent; in-lake concentrations were two to five times greater than those of their inflows. Fluorotelomer sulfonates were present downstream of a known AFFF application area likely to have had fluorotelomer-based foams applied to it, and the concentrations decayed in a predictable manner, suggesting they may be used as an indicator of PFAS transport away from an AFFF source. In all but two samples, organofluorine concentrations were greater than the sum of targeted PFAS (on a F basis) (median of 0.6 % of organofluorine identified via LC-MS/MS), although there was considerable variability in organofluorine measured in replicate samples.

Critical Review on Bromate Formation during Ozonation and Control Options for Its Minimization.

Ozone is a commonly applied disinfectant and oxidant in drinking water and has more recently been implemented for enhanced municipal wastewater treatment for potable reuse and ecosystem protection. One drawback is the potential formation of bromate, a possible human carcinogen with a strict drinking water standard of 10 µg/L. The formation of bromate from bromide during ozonation is complex and involves reactions with both ozone and secondary oxidants formed from ozone decomposition, i.e., hydroxyl radical. The underlying mechanism has been elucidated over the past several decades, and the extent of many parallel reactions occurring with either ozone or hydroxyl radicals depends strongly on the concentration, type of dissolved organic matter (DOM), and carbonate. On the basis of mechanistic considerations, several approaches minimizing bromate formation during ozonation can be applied. Removal of bromate after ozonation is less feasible. We recommend that bromate control strategies be prioritized in the following order: (1) control bromide discharge at the source and ensure optimal ozone mass-transfer design to minimize bromate formation, (2) minimize bromate formation during ozonation by chemical control strategies, such as ammonium with or without chlorine addition or hydrogen peroxide addition, which interfere with specific bromate formation steps and/or mask bromide, (3) implement a pretreatment strategy to reduce bromide and/or DOM prior to ozonation, and (4) assess the suitability of ozonation altogether or utilize a downstream treatment process that may already be in place, such as reverse osmosis, for post-ozone bromate abatement. A one-size-fits-all approach to bromate control does not exist, and treatment objectives, such as disinfection and micropollutant abatement, must also be considered.

Occurrence of quantifiable and semi-quantifiable poly- and perfluoroalkyl substances in united states wastewater treatment plants.

Both quantifiable and semi-quantifiable poly- and perfluoroalkyl substances (PFAS) were evaluated in the influent, effluent, and biosolids of 38 wastewater treatment plants. PFAS were detected in all streams at all facilities. For the means of the sums of detected, quantifiable PFAS concentrations were 98 ± 28 ng/L, 80 ± 24 ng/L, and 160,000 ± 46,000 ng/kg (dry weight basis) in the influent, effluent, and biosolids (respectively). In the aqueous influent and effluent streams this quantifiable PFAS mass was typically associated with perfluoroalkyl acids (PFAAs). In contrast, quantifiable PFAS in the biosolids were primarily polyfluoroalkyl substances that potentially serve as precursors to the more recalcitrant PFAAs. Results of the total oxidizable precursor (TOP) assay on select influent and effluent samples showed that semi-quantified (or, unidentified) precursors accounted for a substantial portion (21 to 88%) of the fluorine mass compared to that associated with quantified PFAS, and that this fluorine precursor mass was not appreciably transformed to perfluoroalkyl acids within the WWTPs, as influent and effluent precursor concentrations via the TOP assay were statistically identical. Evaluation of semi-quantified PFAS, consistent with results of the TOP assay, showed the presence of several classes of precursors in the influent, effluent, and biosolids; perfluorophosphonic acids (PFPAs) and fluorotelomer phosphate diesters (di-PAPs) occurred in 100 and 92% of biosolid samples, respectively. Analysis of mass flows showed that, for both quantified (on a fluorine mass basis) and semi-quantified PFAS, the majority of PFAS exited WWTPs through the aqueous effluent compared to the biosolids stream. Overall, these results highlight the importance of semi-quantified PFAS precursors in WWTPs, and the need to further understand the impacts of their ultimate fate in the environment.

Characterizing the chemical and microbial fingerprint of unsheltered homelessness in an urban watershed.

Unsheltered homelessness is rapidly becoming a critical issue in many cities worldwide. The worsening situation not only highlights the socioeconomic plight, but it also raises awareness of ancillary issues such as the potential implications for urban water quality. The objective of this study was to simultaneously leverage diverse source tracking tools to develop a chemical and microbial fingerprint describing the relative contribution of direct human inputs into Las Vegas' tributary washes. By evaluating a wide range of urban water matrices using general water quality parameters, fecal indicator bacteria (FIB), human-associated microbial markers [e.g., HF183, crAssphage, and pepper mild mottle virus (PMMoV)], 16S rRNA gene sequencing data, and concentrations of 52 anthropogenic trace organic compounds (TOrCs), this study was able to differentiate principal sources of these constituents, including contributions from unsheltered homelessness. For example, HF183 (31% vs. 0%), crAssphage (61% vs. 5%), and PMMoV (72% vs. 55%) were more frequently detected in tributary washes with higher homeless census counts vs. 'control' tributary washes. Illicit drugs or their metabolites (e.g., heroin, acetylmorphine, amphetamine, and cocaine) and select TOrCs (e.g., acetaminophen, caffeine, ibuprofen, and naproxen) were also detected more frequently and at higher concentrations in the more anthropogenically-impacted washes. These data can be used to raise awareness of the shared interests between the broader community and those who are experiencing homelessness, notably the importance of protecting environmental health and water quality. Ultimately, this may lead to more rapid adoption of proven strategies for achieving functional zero homelessness, or at least additional resources for unsheltered individuals.

Pretreatment strategies for ion exchange to control brominated disinfection byproducts in potable reuse.

The application of ion exchange (IX) resins to remove disinfection byproduct (DBP) precursors in wastewater effluents is challenging due to relatively high concentrations of competing anions. This study examined various pretreatment strategies to target competing ions to improve IX removal of DBP precursors, bromide and dissolved organic matter (DOM), measured as trihalomethane and haloacetic acid formation potentials (THMFP and HAAFP). IX batch experiments were performed with four commercial anion exchange (AIX) resins selective for bromide (BrP), DOM (A860), sulfate (MTA) and PFOA/PFOS (PFA), and one cation exchange (CIX) resin selective for iodide (CT). For single AIX treatments the bromide removal ranking was the following: PFA (58%) > MTA (51%) > BrP (43%) > A860 (16%), which corresponded with decreasing brominated THMFP removals and increasing bromine incorporation factors. For dual AIX combinations (PFA and BrP, MTA and BrP), either simultaneous or sequential treatments had the highest bromide (PFA + BrP [69%], MTA + BrP [67%], (PFA→BrP [77%], MTA→BrP [74%]) and Br-THMFP (THMFP [∼80%]) and Br-HAAFP (HAAFP [∼77%]) removals, and therefore the lowest fractions of brominated DBPs (Br-DBPs). Despite ozone (O3), biological active carbon (BAC), and granular activated carbon (GAC) pretreatments reducing the overall DOM concentration (33%), these pretreatment steps did not improve the bromide removals of the resins, although it did increase the Br-THMFP and Br-HAAFP removals by 2-38% and 13-20%, respectively. Nanofiltration (NF) pretreatment significantly removed sulfate (97%) resulting in an increased bromide removal of 19% by the AIX resins, which led to increased removal of Br-THMFP and Br-HAAFP by 93% and 96%, respectively. Among all the IX resins the CT resin had the highest bromide removal (83%) and lowest fraction of Br-DBPs. The results reveal pretreatment with existing technologies including AIX, O3/BAC/GAC, or NF can potentially enhance the removal of brominated DBP precursors by IX resins during potable reuse applications.

Using machine learning classification to detect simulated increases of de facto reuse and urban stormwater surges in surface water.

Water quality events such as increases in stormwater or wastewater effluent in drinking water sources pose hazards to drinking water consumers. Stormwater and wastewater effluent enter Lake Mead-an important drinking water source in the southwest USA-via the Las Vegas Wash. Previous studies have applied machine learning and online instruments to detect contamination in water distribution systems. However, alert systems at drinking water intakes would provide more time for corrective action. An array of online instruments measuring pH, conductivity, redox potential, turbidity, temperature, tryptophan-like fluorescence, UV absorbance (UVA254), TOC, and chlorophyll-a was fed raw water directly from Lake Mead. Wastewater effluent, dry weather Las Vegas Wash, and storm-impacted Las Vegas Wash samples were blended into the instrument inlets at known ratios to simulate three types of adverse water quality events. Data preprocessing was conducted to correct for diurnal patterns or instrument drift. Supervised machine learning was conducted using previously published models in R. Ninety-nine models were screened on the raw data. Eight high-performing models were evaluated in-depth and optimized. Weighted k-Nearest Neighbors, Single C5.0 Ruleset, Mixture Discriminant Analysis, and an ensemble of these three models had accuracy over 97% when assigning test set data among three classes (Normal, Event, or Maintenance). The ensemble detected all event types at the earliest timepoint and had one false positive that was not a lag error (i.e., consecutively following a true positive). Omitting Maintenance, the Adaboost model had over 99% test set accuracy and zero false positives that were not lag errors. Data preprocessing was beneficial, but the optimal methods were model-specific. All nine water quality variables were useful for most models, but UVA254 and turbidity were most important.

Biochar Water Treatment for Control of Organic Micropollutants with UVA Surrogate Monitoring.

Biochar adsorbent can be produced in low-resource settings using local materials and simple pyrolysis technology, and it has shown promise for uptake of micropollutants (MPs) such as pesticides, pharmaceuticals, industrial compounds, and chemicals released from consumer goods present in water at ng/L to µg/L levels. Accordingly, the use of biochar in water treatment applications where granular activated carbon (GAC) is economically or logistically infeasible is gaining interest. Monitoring treatment systems for individual MPs require laboratory analytical techniques that are typically cost-prohibitive and impractical for low-resource settings. Therefore, identification of surrogate parameters(s) for adsorbent bed life that can be measured inexpensively and in the field is a high priority. Background dissolved organic matter (DOM) is ubiquitous in natural and anthropogenic waters at concentrations typically 1,000 to 100,000 that of MPs. Some constituents of DOM foul the adsorbent and reduce bed life for removal of target contaminants. Aromatic DOM foulants absorb ultraviolet light at a wavelength of 254 nm (UVA254). Because DOM fouling directly affects MP adsorption capacity and DOM is a bulk water parameter that can be quantified using relatively inexpensive and portable instruments, it could be exploited as a surrogate for monitoring biochar adsorber bed life under field conditions. The objective of this study was to quantify removal of MPs from waters containing different types and concentrations of background DOM (surface water, wastewater, dump leachate) and thus exhibiting different UVA254 breakthrough profiles in bench-scale column experiments. Breakthrough profiles of weakly to moderately adsorbing MPs, including herbicides, pharmaceuticals and personal care products, and perfluoroalkyl acids, were collected using biochars generated under different pyrolysis conditions and a commercial GAC as a performance benchmark. Optimal conditions for biochar water treatment include using biochar produced from wood at ≥850°C under slightly aerobic conditions, empty bed contact times of ≥30 min, and upstream treatment processes to reduce DOM. Relative UVA254 breakthrough (C/C 0) up to 0.6-0.9 corresponded to ≥90% MP removal for most MP-water combinations studied.

Assessment of PFAS fate, transport, and treatment inhibition associated with a simulated AFFF release within a WASTEWATER treatment plant.

Sequencing batch reactors (SBRs) were operated for 36 days to simulate the potential wastewater treatment impacts as well as fate and transport of per- and polyfluoroalkyl substances (PFAS) that could be associated with a release of alcohol resistant aqueous film forming foam (AR-AFFF) from on-site methanol fire suppression systems. The results of this study indicate that two days of exposure to AFFF were associated with small reductions in mixed liquor solids content and nitrification rates. No impacts on denitrification or biological phosphorus removal were observed. The addition of AFFF was associated with increases in 6:2 fluorotelomer sulfonate (6:2 FTS) in influent, effluent, and solids samples in the SBR. The following biotransformation pathway is proposed: an unidentified fluorotelomer precursor quickly degraded to 6:2 FTS, which then slowly degraded to several identified degradation intermediates and terminal, short-chain perfluorocarboxylic acid products. Data for 6:2 FTS, which was used as a proxy for AFFF-associated PFAS, were extrapolated to estimate that a removal of approximately 70% of AFFF via effluent and solids wasting would occur after 4 days at a full-scale treatment plant. This information can be used to better understand potential impacts on downstream processes, including potable reuse and biosolids production.