Water Quality Constituents Research Articles

Application of a reaction-based water quality model to the total dissolved solids concentration of the Pasig River.

With the goal to support effective water resource management, water quality models have gained popularity as tools for evaluating the distributions of pollutants and sediments. This work focuses on the application of the numerical solution of an advection-dispersion-reaction (ADR) water quality model for rivers and streams to a major Philippine waterway, the Pasig River. The water quality constituent is described by a system of reaction and advection-dispersion-reaction equations. The model and method are based on a previously used strategy where Guass-Jordan decomposition is applied to the matrix system and the resulting conservative form of the model is solved numerically using the fully implicit scheme and finite element method. The methodology is demonstrated by a case study in Pasig River involving the concentrations of total dissolved solids (TDS) obtained from the Department of Environment and Natural Resources (DENR) through the Pasig River Unified Monitoring Stations (PRUMS) report. Sensitivity analysis and parameter estimation are also applied to the model to assess which parameters influence the model output the most.

Evaluating a process‐guided deep learning approach for predicting dissolved oxygen in streams

AbstractDissolved oxygen (DO) is a critical water quality constituent that governs habitat suitability for aquatic biota, biogeochemical reactions and solubility of metals in streams. Recently introduced high‐frequency sensors have increased our ability to measure DO, but we still lack the capacity to understand and predict DO concentrations at high spatial resolutions or in unmonitored locations. Machine learning (ML) has been a commonly used approach for modelling DO, however, conventional ML models have no representation of the limnological processes governing DO dynamics. Here we implement and evaluate two process‐guided deep learning (PGDL) approaches for predicting daily minimum, mean and maximum DO concentrations in rivers from the Delaware River Basin, USA. In both cases, a multi‐task approach was taken in which the PGDL models predicted stream metabolism and gas exchange rates in addition to the DO concentrations themselves. Our results showed that for these sites, the PGDL approaches did not improve upon baseline predictions in temporal and spatially similar holdout experiments. One of the approaches did, however, improve predictions when applied to spatially dissimilar sites. Although this particular PGDL approach did not improve predictive accuracy in most cases, our results suggest that process guidance, perhaps a more constrained approach, could benefit a data‐driven DO model.

Stream chemical response is mediated by hydrologic connectivity and fire severity in a Pacific Northwest forest

AbstractLarge‐scale wildfires are becoming increasingly common in the wet forests of the Pacific Northwest (USA), with predicted increases in fire prevalence under future climate scenarios. Wildfires can alter streamflow response to precipitation and mobilize water quality constituents, which pose a risk to aquatic ecosystems and downstream drinking water treatment. Research often focuses on the impacts of high‐severity wildfires, with stream biogeochemical responses to low‐ and mixed‐severity fires often understudied, particularly during seasonal shifts in hydrologic connectivity between hillslopes and streams. We studied the impacts of the 2020 Holiday Farm Fire at the HJ Andrews Experimental Forest where rare pre‐fire stream discharge and chemistry data allowed us to evaluate the influence of mixed‐severity fire on stream water quantity and quality. Our research design focused on two well‐studied watersheds with low and low‐moderate burn severity where we examined long‐term data (pre‐ and post‐fire), and instantaneous grab samples collected during four rain events occurring immediately following wildfire and a prolonged dry summer. We analysed the impact of these rain events, which represent the transition from low‐to‐high hydrologic connectivity of the subsurface to the stream, on stream discharge and chemistry behaviour. Long‐term data revealed total annual flows and mean flows remained fairly consistent post‐fire, while small increases in baseflow were observed in the low‐moderately burned watershed. Stream water concentrations of nitrate, phosphate and sulfate significantly increased following fire, with variance in concentration increasing with fire severity. Our end member mixing models suggested that during rain events, the watershed with low‐moderate severity fire had greater streamflow inputs from soil water and groundwater during times of low connectivity compared to the watershed with low severity fire. Finally, differences in fire severity impacts on concentration‐discharge relationships of biogenic solutes were most expressed under low catchment connectivity conditions. Our study provides insights into post‐wildfire impacts to stream water quality, with the goal of informing future research on stream chemistry responses to low, moderate and mixed severity wildfire.

Potential Climate and Human Water-Use Effects on Water-Quality Trends in a Semiarid, Western U.S. Watershed: Fountain Creek, Colorado, USA

Nutrients, total dissolved solids (TDS), and trace elements affect the suitability of water for human and natural needs. Here, trends in such water-quality constituents are analyzed for 1999–2022 for eight nested monitoring sites in the 24,000 km2 Fountain Creek watershed in Colorado, USA, by using the weighted regressions on time, discharge, and season (WRTDS) methodology. Fountain Creek shares characteristics with other western U.S. watersheds: (1) an expanding but more water-efficient population, (2) a heavy reliance on imported water, (3) a semiarid climate trending towards warmer and drier conditions, and (4) shifts of water from agricultural to municipal uses. The WRTDS analysis found both upward and downward trends in the concentrations of nutrients that reflected possible shifts in effluent management, instream uptake, and water conservation by a watershed population that grew by about 40%. Selenium, other trace elements, and TDS can pose water-quality challenges downstream and their concentrations were found to have a downwards trend. Those trends could be driven by either a warming and drying of the local climate or decreased agricultural irrigation, as both would reduce recharge and subsequent mobilization from natural geologic sources via groundwater discharge. The patterns illustrate how changes in climate and water use may have affected water quality in Fountain Creek and demonstrate the patterns to look for in other western watersheds.

Longitudinal stream synoptic (LSS) monitoring to evaluate water quality in restored streams.

Impervious surface cover increases peak flows and degrades stream health, contributing to a variety of hydrologic, water quality, and ecological symptoms, collectively known as the urban stream syndrome. Strategies to combat the urban stream syndrome often employ engineering approaches to enhance stream-floodplain reconnection, dissipate erosive forces from urban runoff, and enhance contaminant retention, but it is not always clear how effective such practices are or how to monitor for their effectiveness. In this study, we explore applications of longitudinal stream synoptic (LSS) monitoring (an approach where multiple samples are collected along stream flowpaths across both space and time) to narrow this knowledge gap. Specifically, we investigate (1) whether LSS monitoring can be used to detect changes in water chemistry along longitudinal flowpaths in response to stream-floodplain reconnection and (2) what is the scale over which restoration efforts improve stream quality. We present results for four different classes of water quality constituents (carbon, nutrients, salt ions, and metals) across five watersheds with varying degrees of stream-floodplain reconnection. Our work suggests that LSS monitoring can be used to evaluate stream restoration strategies when implemented at meter to kilometer scales. As streams flow through restoration features, concentrations of nutrients, salts, and metals significantly decline (p < 0.05) or remain unchanged. This same pattern is not evident in unrestored streams, where salt ion concentrations (e.g., Na+, Ca2+, K+) significantly increase with increasing impervious cover. When used in concert with statistical approaches like principal component analysis, we find that LSS monitoring reveals changes in entire chemical mixtures (e.g., salts, metals, and nutrients), not just individual water quality constituents. These chemical mixtures are locally responsive to restoration projects, but can be obscured at the watershed scale and overwhelmed during storm events.

ASSESSMENT AND SPATIAL DISTRIBUTION OF WATER QUALITY CONSTITUENTS IN LAKE ECOSYSTEM USING SENITNEL-2 DATA AND MODELLING

An inland lake is an essential component of the freshwater environment and must be monitored to prevent the quality of the lake's water from deteriorating. Turbidity and total suspended matter (TSM) are crucial water quality assessment factors. A cutting-edge technical approach called remote sensing is utilized to keep an eye on the quality of lake water. To assess the lake water quality, the water quality constituents (WQC) model was developed using an empirical computation technique using spectral wavelengths (400-800 nm) from in-situ hyperspectral remote sensing measurements. The results show that the best appropriate models that are sensitive to the estimation of TSM (R2 = 0.8523) and Turbidity (R2 = 0.9115) were selected and incorporated into Sentinel-2MSI for WQC estimation. The spatial distribution of WQC is mapped, providing information on both the distribution and variations of WQC within the lake. The research has proven that it has the potential for regular monitoring of lake water through this technique.

Health Risk Assessment of Nitrate in Drinking Water with Potential Source Identification: A Case Study in Almaty, Kazakhstan.

Infant mortality in Kazakhstan is six times higher compared with the EU. There are several reasons for this, but a partial reason might be that less than 30% of Kazakhstan's population has access to safe water and sanitation and more than 57% uses polluted groundwater from wells that do not comply with international standards. For example, nitrate pollution in surface and groundwater continues to increase due to intensified agriculture and the discharge of untreated wastewater, causing concerns regarding environmental and human health. For this reason, drinking water samples were collected from the water supply distribution network in eight districts of Almaty, Kazakhstan, and water quality constituents, including nitrate, were analyzed. In several districts, the nitrate concentration was above the WHO and Kazakhstan's maximum permissible limits for drinking water. The spatial distribution of high nitrate concentration in drinking water was shown to be strongly correlated with areas that are supplied with groundwater, whereas areas with lower nitrate levels are supplied with surface water sources. Based on source identification, it was shown that groundwater is likely polluted by mainly domestic wastewater. The health risk for infants, children, teenagers, and adults was assessed based on chronic daily intake, and the hazard quotient (HQ) of nitrate intake from drinking water was determined. The non-carcinogenic risks increased in the following manner: adult < teenager < child < infant. For infants and children, the HQ was greater than the acceptable level and higher than that of other age groups, thus pointing to infants and children as the most vulnerable age group due to drinking water intake in the study area. Different water management options are suggested to improve the health situation of the population now drinking nitrate-polluted groundwater.

A water quality database for global lakes

Lakes with a small footprint in non-glaciated lands, are immense sources of our planet's biodiversity, which deliver undeniable benefits to human life and aquatic-terrestrial ecosystems. However, water quality in global lakes is threatened by unsustainable development and changing climate, with profound consequences for food-web dynamics and bio-geochemical processes. Here, we compiled a rich database of the primary water quality constituents based on their availability, morphometric characteristics, and geographical information of Earth's lakes through extensive searches in accessible repositories, published articles, publicly available maps, and both private and governmental reports. Our assembled database includes 866,390 water quality data measured in 13,876 lakes located in 77 different countries, as well as 55,629 datapoints of morphometric characteristics. In sum, our database includes 922,019 data associated with global lakes. These open-access data will be beneficial for managers and policymakers to improve water quality and ecological restoration in these invaluable aquatic ecosystems during the Anthropocene.

Impacts of droughts and heatwaves on river water quality worldwide

Droughts and heatwaves have a major impact on river water quality worldwide. However, previous studies of river water quality under these climate extremes are limited to a small number of river basins and regions, mainly located in North America, Europe or Australia. In this study, we estimate the large-scale effects of droughts, heatwaves and compound drought-heatwave events on river water quality for a total of 314,046 water quality monitoring stations worldwide over the period 1980–2021. We focus on 16 water quality constituents grouped into physical (e.g., temperature, salinity), chemical (e.g., pharmaceuticals, pesticides) and biological constituents (e.g., biochemical oxygen demand, faecal coliform). Further, we analyse the response of each constituent to droughts and heatwaves in relation to climate type, land use and level of wastewater treatment. We find a general deterioration in river water quality under droughts and heatwaves globally for most rivers and water quality constituents considered. For example, there is on average a 27 % increase in river water temperature, 17 % decrease in dissolved oxygen and 24 % increase in salinity under droughts and heatwaves. In addition, we find that climate type, land use and level of wastewater treatment have a significant effect on the magnitude of response in each water quality constituent during these extreme events. The median increase in river water temperature under compound drought-heatwaves is strongly driven by climate zone with higher warming rates at the polar climate zone (+4.5 °C) compared to the tropical zone (+2.1 °C). Increases in salinity under droughts are on average two times larger in irrigated regions compared to non-irrigated regions. The concentrations of nutrients (P and N) in rivers can either increase or decrease under droughts, depending on the nutrient form (dissolved vs. particulate) and land use (urban vs. rural). Higher levels of wastewater treatment contributed to a greater decrease in pathogenic concentrations under droughts and heatwaves (as indicated by faecal coliform). Pharmaceuticals show mixed responses mainly depending on the persistence of the constituent in surface waters, with for instance declines in diclofenac concentrations under droughts and heatwaves due to increased decay under higher water temperatures. The results of this study provide a broader understanding of how droughts and heatwaves affect river water quality compared to previous local and regional-scale analyses. In addition, this study could form the basis for large-scale modelling of river water quality under droughts and heatwaves.

A Decision Support System That Considers Risk and Site Specificity in the Assessment of Irrigation Water Quality (IrrigWQ)

Irrigators are increasingly challenged to maintain or even increase production using less water, sometimes of poorer quality, and often from unconventional sources. This paper describes the main features of a newly developed software-based Decision Support System (DSS), with which the fitness for use (FFU) of water for irrigation (IrrigWQ) can be assessed. The assessment considers site-specific factors, several non-traditional water constituents, and the risk of negative effects. The water balance components of a cropping system and the redistribution of solutes within a soil profile are assessed with a simplified soil water balance and chemistry model. User-friendly, colour-coded output highlights the expected effects of water constituents on soil quality, crop yield and quality, and irrigation infrastructure. Because IrrigWQ uses mainly internationally accepted cause–effect relationships to assess the effect of water quality constituents, it is expected to find universal acceptance and application among users. IrrigWQ also caters for calculating so-called Water Quality Requirements (WQRs). WQRs indicate the threshold levels of water quality constituents for irrigation at specified levels of acceptability or risk. WQRs assist water resource managers in setting site-specific maximum threshold levels of water quality constituents that can be tolerated in a water source before impacting negatively on successful irrigation.

The risk of pulmonary NTM infections and water-quality constituents among persons with cystic fibrosis in the United States, 2010-2019.

The prevalence of nontuberculous mycobacterial (NTM) pulmonary disease varies geographically in the United States. Previous studies indicate that the presence of certain water-quality constituents in source water increases NTM infection risk. To identify water-quality constituents that influence the risk of NTM pulmonary infection in persons with cystic fibrosis in the United States. We conducted a population-based case-control study using NTM incidence data collected from the Cystic Fibrosis Foundation Patient Registry during 2010-2019. We linked patient zip code to the county and associated patient county of residence with surface water data extracted from the Water Quality Portal. We used logistic regression models to estimate the odds of NTM infection as a function of water-quality constituents. We modeled two outcomes: pulmonary infection due to Mycobacterium avium complex (MAC) and Mycobacterium abscessus species. We identified 484 MAC cases, 222 M. abscessus cases and 2816 NTM-negative cystic fibrosis controls resident in 11 states. In multivariable models, we found that for every 1-standardized unit increase in the log concentration of sulfate and vanadium in surface water at the county level, the odds of infection increased by 39% and 21%, respectively, among persons with cystic fibrosis with MAC compared with cystic fibrosis-NTM-negative controls. When modeling M. abscessus as the dependent variable, every 1-standardized unit increase in the log concentration of molybdenum increased the odds of infection by 36%. These findings suggest that naturally occurring and anthropogenic water-quality constituents may influence the NTM abundance in water sources that supply municipal water systems, thereby increasing MAC and M. abscessus infection risk.

On the biodiversity of diatom communities in lakes of East China: Geochemical effects dominated after eutrophication

AbstractAimOver the last century, lakes all over the world have experienced significant eutrophication and become more susceptible to critical transitions. This has prompted a growing need to understand how community dynamics shift following ecological perturbations. This study aims to elucidate the present status, spatial patterns and determinants of biological diversity, and how they have changed in response to alterations driven by eutrophication.LocationLakes of East China, mostly in the Yangtze River Basin.TaxaBacillariophyceae (163 species in 33 genres, see taxa names in Appendix S1).MethodsThe environmental conditions of over 60 lakes were assessed based on parameters such as water quality, physical characteristics and biogeochemical constituents. Surface sediment samples and high‐resolution sediment records were collected to quantify the biodiversity of diatom communities, and principal component analysis (PCA) was used to determine the major environmental gradients. Meanwhile, linear regression, canonical correlation analysis and Mantel tests were used to quantify the correlations between variables.ResultsBiodiversity metrics showed no distinct geographic patterns, although areas near large lakes showed comparatively high evenness. Smaller pairwise dissimilarities were observed among hypereutrophic lakes in comparison to lakes with a lower trophic status. The past century has witnessed a decrease in both pairwise dissimilarity and the standard deviation of evenness. A significant statistical association was found between dissimilarity and ionic concentration and composition in the surface sediment of lakes, rather than nutrient levels.Main ConclusionsFollowing regional eutrophication and human disturbance, a significant reduction in biodiversity differentiation was observed in lakes located in East China. Currently, the dominant factor in determining the beta diversity of diatom communities is geochemical conditions rather than nutrient concentrations. Long‐term stress caused by cultural eutrophication may also have altered the dominant determinant of regional biodiversity and diatom community dynamics.

HARP: A suite of parameters to describe the hysteresis of streamflow and water quality constituents

Hysteresis occurs in ecological systems when there are internal dynamics that produce variable responses to an external stimuli or stressors. Understanding and quantifying hysteresis is important to inform understanding of system behaviour, support modelling, and provide comparisons within and between systems. We introduce a new tool, HARP (Hysteresis Area, Residual and Peaks) to describe and quantify hysteresis. HARP quantifies the key features of stream discharge–constituent concentration relationships, including the magnitude of the hysteresis (deviation from linearity), phase shifts in the system (end-state deviation from the pre-event state), and the timing of the turning points in the relationship of discharge and concentration. HARP is suitable to use on complex discharge–concentration data, including figure-eight loops, and supports statistical analysis to quantify the characteristics of independent events. The HARP metrics are introduced using discharge–water quality constituent concentration data, but are equally applicable to any system with properties of hysteresis. HARP provides a tool to develop quantitative observations of stream hydrological processes into quantitative enumeration statistics that facilitate comparisons of events and streams.

A complex interplay among agricultural land uses, urbanization, and landscape attributes shapes the concentration-discharge relationships in Ontario, Canada

The shape of a concentration-discharge (C-Q) relationship and its evolution in space and time can be used to elucidate the predominant hydrological pathways and residence time of a chemical of interest, to determine the rate and timing of biogeochemical production and uptake, and to evaluate how catchment characteristics (topography, land cover/soil vegetation, anthropogenic activities) modulate export dynamics. Being a powerful tool to address ecohydrology’s fundamental questions, the investigation of C–Q relationships for multiple elements at multiple spatial and temporal scales has been instrumental in the context of water-quality management. In this study, we use a statistical (Segmented, Classification and Regression Trees) modelling framework to establish linkages among watershed morphological attributes, weather variability, land-use patterns, and biogeochemical “hot spots” across Ontario, Canada. Our analysis signifies the presence of enriching behavior, either across the entire streamflow domain or at high flow regimes, for several water quality constituents, such as total phosphorus, suspended solids, nitrate, calcium, and dissolved organic carbon. Alkalinity, calcium, and dissolved inorganic carbon consistently displayed negative slopes at higher flows, indicative of source limitation for these constituents. Similarly, although many of our study watersheds are influenced by anthropogenic activities, the predominant C-Q relationship for chloride was suggestive of either a monotonically diluting pattern or a diluting one followed by a chemostatic response to increasing streamflows. Our results offer novel insights into the interplay between anthropogenic stressors and biogeochemical processes that shapes the severity of nonpoint-source pollution, as hydrological and nutrient cycles will be increasingly subjected to major disturbances by food-production systems and urbanization.

Impact of a natural disturbance on the performance and microbial communities in a full-scale constructed wetland for industrial wastewater treatment

Constructed Wetlands (CWs) are a cost-effective, versatile and sustainable choice for wastewater treatment. In these environments, microbial communities play a significant role in pollutant removal. However, little is known about how microbial communities in full-scale CWs contribute to maintaining water quality or how their dynamics change in response to pulse disturbances such as fire or freezes. Furthermore, few studies have examined the relationship between CW microbial community structure and performance in full-scale industrial operations. We characterized the water-column and leaf-litter layer microbial communities in a 110-acre free water surface CW that provides tertiary wastewater treatment to a plastics manufacturing plant. The CW’s sampling campaign was conducted over a 12-month period that included Winter Storm Uri, a 100-year freeze event. Analysis of 16S rRNA gene amplicon sequences revealed that the bacterial communities experienced a temporal shift. There was also a shift in microbial community structure between the influent and the first segment of the CW. However, no differences in microbial community structure were observed in the second segment of the CW. There was a negative association between microbial community diversity and chlorophyll a, as well as microbial community diversity and total suspended solids (TSS); demonstrating an increase in microbial biodiversity as water quality improved throughout the CW. Six months after the freeze, CW performance in terms of removal of water quality constituents began to return to former removal trends. Yet, there was still a significant difference in microbial community structure within the CW relative to the previous year. This suggests CW functional resilience despite a shift in microbial community structure in the wetland.

Improving detention ponds for effective stormwater management and water quality enhancement under future climate change: a simulation study using the PCSWMM model

Urban surfaces are often covered by impermeable materials such as concrete and asphalt which intensify urban runoff and pollutant concentration during storm events, and lead to the deterioration of the quality of surrounding water bodies. Detention ponds are used in urban stormwater management, providing two-fold benefits: flood risk reduction and pollution load minimization. This paper investigates the performance of nine proposed detention ponds (across the city of Renton, Washington, USA) under different climate change scenarios. First, a statistical model was developed to estimate the pollutant load for the current and future periods and to understand the effects of increased rainfall on stormwater runoff and pollutant loads. The Personal Computer Storm Water Management Model (PCSWMM) platform is employed to calibrate an urban drainage model for quantifying stormwater runoff and corresponding pollutant loads. The calibrated model was used to investigate the performance of the proposed nine (9) detention ponds under future climate scenarios of 100-year design storms, leading to identifying if they are likely to reduce stormwater discharge and pollutant loads. Results indicated significant increases in stormwater pollutants due to increases in rainfall from 2023 to 2050 compared to the historical period 2000–2014. We found that the performance of the proposed detention ponds in reducing stormwater pollutants varied depending on the size and location of the detention ponds. Simulations for the future indicated that the selected detention ponds are likely to reduce the concentrations (loads) of different water quality constituents such as ammonia (NH3), nitrogen dioxide (NO2), nitrate (NO3), total phosphate (TP), and suspended solids (SS) ranging from 18 to 86%, 35–70%, 36–65%, 26–91%, and 34–81%, respectively. The study concluded that detention ponds can be used as a reliable solution for reducing stormwater flows and pollutant loads under a warmer future climate and an effective adaptation option to combat climate change related challenges in urban stormwater management.

Advances in Catchment Science, Hydrochemistry, and Aquatic Ecology Enabled by High-Frequency Water Quality Measurements.

High-frequency water quality measurements in streams and rivers have expanded in scope and sophistication during the last two decades. Existing technology allows in situ automated measurements of water quality constituents, including both solutes and particulates, at unprecedented frequencies from seconds to subdaily sampling intervals. This detailed chemical information can be combined with measurements of hydrological and biogeochemical processes, bringing new insights into the sources, transport pathways, and transformation processes of solutes and particulates in complex catchments and along the aquatic continuum. Here, we summarize established and emerging high-frequency water quality technologies, outline key high-frequency hydrochemical data sets, and review scientific advances in key focus areas enabled by the rapid development of high-frequency water quality measurements in streams and rivers. Finally, we discuss future directions and challenges for using high-frequency water quality measurements to bridge scientific and management gaps by promoting a holistic understanding of freshwater systems and catchment status, health, and function.

Pecora 22: Remote Sensing for Freshwater and Marine Environments

Pecora 22: Remote Sensing for Freshwater and Marine Environments

Water Quality Modeling in Headwater Catchments: Comprehensive Data Assessment, Model Development and Simulation of Scenarios

Water quality is a major concern globally and in headwater catchments of developing countries it is often poorly managed. In these catchments, having scarce and heterogeneous information hinders the development of water quality assessments and predictive models to support management. To address this issue, the authors propose a framework of three stages that allows for: (i) conducting a comprehensive assessment of water quality; (ii) the development of a mountain stream water quality model based on said assessment; and (iii) the simulation of scenarios with the model to resolve conflicts between uses and quality of water. The framework involves multivariate analyses of principal components and clusters and follows a novel modeling protocol mainly designed for mountainous streams in developing countries. Applied to an Andean catchment in Colombia, the first stage of the framework revealed the catchment’s most significant water quality constituents and the most polluted season. The problematic constituents in this catchment include pathogens, nutrients, organic matter, and metals such as the highly toxic Cr and Pb, while water pollution is the highest during the driest months of the year (i.e., January to March). In the second stage, the model was calibrated reproducing the concentrations of pathogens, organic matter, and most nutrients, and showed a predictive capacity. This capacity was measured with an objective function to be minimized based on a normalized root mean square error. It increased only 14% when verified with a different dataset. In addition, during the third stage of the proposed framework, the simulation of alternative scenarios showed that centralized treatment is not sufficient to make water safe for potabilization and agriculture in the catchment. For this reason, improving water quality in the sub-basins at the highest altitudes is required. The proposed framework can be applied in other headwater catchments where information is limited, and where an improved management of water quality is needed.

A Machine Learning Approach to Predict Watershed Health Indices for Sediments and Nutrients at Ungauged Basins.

Effective water quality management and reliable environmental modeling depend on the availability, size, and quality of water quality (WQ) data. Observed stream water quality data are usuallEEy sparse in both time and space. Reconstruction of water quality time series using surrogate variables such as streamflow have been used to evaluate risk metrics such as reliability, resilience, vulnerability, and watershed health (WH) but only at gauged locations. Estimating these indices for ungauged watersheds has not been attempted because of the high-dimensional nature of the potential predictor space. In this study, machine learning (ML) models, namely random forest regression, AdaBoost, gradient boosting machines, and Bayesian ridge regression (along with an ensemble model), were evaluated to predict watershed health and other risk metrics at ungauged hydrologic unit code 10 (HUC-10) basins using watershed attributes, long-term climate data, soil data, land use and land cover data, fertilizer sales data, and geographic information as predictor variables. These ML models were tested over the Upper Mississippi River Basin, the Ohio River Basin, and the Maumee River Basin for water quality constituents such as suspended sediment concentration, nitrogen, and phosphorus. Random forest, AdaBoost, and gradient boosting regressors typically showed a coefficient of determination for suspended sediment concentration and nitrogen during the testing stage, while the ensemble model exhibited . Watershed health values with respect to suspended sediments and nitrogen predicted by all ML models including the ensemble model were lower for areas with larger agricultural land use, moderate for areas with predominant urban land use, and higher for forested areas; the trained ML models adequately predicted WH in ungauged basins. However, low WH values (with respect to phosphorus) were predicted at some basins in the Upper Mississippi River Basin that had dominant forest land use. Results suggest that the proposed ML models provide robust estimates at ungauged locations when sufficient training data are available for a WQ constituent. ML models may be used as quick screening tools by decision makers and water quality monitoring agencies for identifying critical source areas or hotspots with respect to different water quality constituents, even for ungauged watersheds.