Supervised Machine Learning for Estimation of Total Suspended Solids in Urban Watersheds

Modeling urban pollutant wash-off processes with ecological memory

Urbanization increases the extent of impervious surfaces, runoff, sediment, and nutrient loadings downstream, leading to the deterioration of urban surface waters. During pollutant wash-off from urban surfaces, the peak concentration of pollutants typically occurs after the rainfall peak. However, current urban wash-off models do not consider this time delay, assuming that the effect of rainfall on the wash-off process is immediate. Ecological memory (EM) is defined as the capacity of past states or experiences to influence the present or future ecological responses of a community ecosystem. In this study, ecological memory was calculated to reflect the lagging impact of rainfall on the wash-off process. Incorporating ecological memory into the original wash-off model improved its performance across all pollutant types and rainfall conditions, with the adjusted R-squared value increasing from -0.01-0.61 to -0.08-0.83. Among the factors impacting ecological memory, rainfall intensity significantly affected the distributions of ecological memory (p < 0.05 according to the Kolmogorov-Smirnov test). The first flush phenomenon, which involves a larger concentration or mass of pollutants in the initial portion of a storm event compared to the rest, showed no significant difference in the distributions of ecological memory between rain events with and without this phenomenon. The improved urban wash-off model can be applied for real-time simulation of urban pollutants and help develop site-specific measures for reducing urban nonpoint source pollution.

Machine learning models for prediction of nutrient concentrations in surface water in an agricultural watershed

Prediction and quantification of nutrient concentrations in surface water has gained substantial attention during recent decades because excess nutrients released from agricultural and urban watersheds can significantly deteriorate surface water quality. Machine learning (ML) models are considered an effective tool for better understanding and characterization of nutrient release from agricultural fields to surface water. However, to date, no systematic investigations have examined the implementation of different classification and regression ML models in agricultural settings to predict nutrient concentrations in surface water using a group of input variables including climatological (e.g., precipitation), hydrological (e.g., stream flow) and field characteristics (i.e., land and crop use). In the current study, multiple classification (e.g., decision trees) and regression (e.g., regression trees) ML models were applied on a dataset pertaining to surface water quality in an agricultural watershed in southern Ontario, Canada (i.e., Upper Parkhill watershed). The target variables of these models were the nutrient concentrations in surface water including nitrate, total phosphorus, soluble reactive phosphorus, and total dissolved phosphorus. These target variables were predicted using physical and chemical water parameters of surface water (e.g., temperature and DO), climatological, hydrological, and field conditions as the input variables. The performance of these different models was assessed using various evaluation metrics such as classification accuracy (CA) and coefficient of determination (R2) for classification and regression models, respectively. In general, both the ensemble bagged trees and logistic regression (CA ≥ 0.72), and exponential Gaussian process regression (R2≥ 0.93) models were the optimal classification and regression ML algorithms, respectively, where they resulted in the highest prediction accuracy of the target variables. The insights and outcomes of the current study demonstrates that ML models can be employed to effectively predict and quantify the nutrient concentrations in surface waters to supplement field-collected monitoring data in agricultural watersheds, assisting in maintaining high quality of the available surface water resources.

Application of classification machine learning algorithms for characterizing nutrient transport in a clay plain agricultural watershed

Excess nutrients in surface water and groundwater can lead to water quality deterioration in available water resources. Thus, the classification of nutrient concentrations in water resources has gained significant attention during recent decades. Machine learning (ML) algorithms are considered an efficient tool to describe nutrient loss from agricultural land to surface water and groundwater. Previous studies have applied regression and classification ML algorithms to predict nutrient concentrations in surface water and/or groundwater, or to categorize an output variable using a limited number of input variables. However, there have been no studies that examined the application of different ML classification algorithms in agricultural settings to classify various output variables using a wide range of input variables. In this study, twenty-four ML classification algorithms were implemented on a dataset from three locations within the Upper Parkhill watershed, an agricultural watershed in southern Ontario, Canada. Nutrient concentrations in surface water were classified using geochemical and physical water parameters of surface water and groundwater (e.g., pH), climate and field conditions as the input variables. The performance of these algorithms was evaluated using four evaluation metrics (e.g., classification accuracy) to identify the optimal algorithm for classifying the output variables. Ensemble bagged trees was found to be the optimal ML algorithm for classifying nitrate concentration in surface water (accuracy of 90.9%), while the weighted KNN was the most appropriate algorithm for categorizing the total phosphorus concentration (accuracy of 87%). The ensemble subspace discriminant algorithm gave the highest overall classification accuracy for the concentration of soluble reactive phosphorus and total dissolved phosphorus in surface water with an accuracy of 79.2% and 77.9%, respectively. This study exemplifies that ML algorithms can be used to signify exceedance of recommended concentrations of nutrients in surface waters in agricultural watersheds. Results are useful for decision makers to develop nutrient management strategies.

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.

Comparing stormwater quality and watershed typologies across the United States: A machine learning approach

Watersheds continue to be urbanized across different regions of the United States, increasing the number of impaired waterbodies due to urban stormwater. Using machine learning techniques, this study examined how stormwater quality and watershed characteristics are related at a national scale and compared stormwater quality across watersheds in diverse climates. We analyzed a selection of data from the National Stormwater Quality Database (NSQD) comprising 1,881 stormwater samples taken from 182 watersheds in 26 metropolitan areas in the United States between 1992 and 2003. Using an ensemble clustering algorithm, the stormwater quality in these samples was classified into "stormwater signatures," defined as distinct combinations of 9 contaminants including metals (Pb, Zn, Cu), particulates (TSS, TDS), and nutrients (BOD, TP, TKN, NOx). Next, multinomial logistic regression was applied to the NSQD data now classified by signature and combined with climate, weather, land use, and imperviousness data obtained from multiple sources. The results yielded 5 stormwater signatures with distinct aquatic toxicity implications and relationships to climate, weather, land use, and imperviousness: Signature 1 ("Ecotoxic and Eutrophic"), defined by high median concentrations of contaminants, likely represents the first flush in moderate-to-high imperviousness watersheds; Signature 2 ("Reduced Nitrates") represents a wet season signature, particularly for dry climates; Signature 3 ("Potentially Eutrophic") represents the first flush in low imperviousness watersheds; Signature 4 ("Elevated Particulates and Metals") represents a wet season signature, particularly on warmer days; finally, Signature 5 ("Most Dilute") is primarily a regional signature associated with the warm, wet climate of the southeastern US. This study serves as a proof-of-concept demonstrating how machine learning techniques can be used to identify patterns in high-dimensional and highly variable data. Applied to stormwater quality, these techniques identify major patterns in stormwater quality across the United States using a stormwater signature approach, which examines how contaminants co-occur and under what climate, weather, land use, and impervious conditions. The findings point to dominant processes driving stormwater generation and inform watershed monitoring, green infrastructure planning, stormwater quality under climate change, and opportunities for public engagement.

A Site-Scale Tool for Performance-Based Design of Stormwater Best Management Practices

The objective of this research is to develop a module for the design of best management practices based on percent pollutant removal. The module is a part of the site-scale integrated decision support tool (i-DSTss) that was developed for stormwater management. The current i-DSTss tool allows for the design of best management practices based on flow reduction. The new water quality module extends the capability of the i-DSTss tool by adding new procedures for the design of best management practices based on treatment performance. The water quality module can be used to assess the treatment of colloid/total suspended solid and dissolved pollutants. We classify best management practices into storage-based (e.g., pond) and infiltration-based (e.g., bioretention and permeable pavement) practices for design purposes. Several of the more complex stormwater tools require expertise to build and operate. The i-DSTss and its component modules including the newly added water quality module are built on an accessible platform (Microsoft Excel VBA) and can be operated with a minimum skillset. Predictions from the water quality module were compared with observed data, and the goodness-of-fit was evaluated. For percent total suspended solid removal, both R2 and Nash–Sutcliffe efficiency values were greater than 0.7 and 0.6 for infiltration-based and storage-based best management practices, respectively, demonstrating a good fit for both types of best management practices. For percent total phosphorous and Escherichia. coli removal, R2 and Nash–Sutcliffe efficiency values demonstrated an acceptable fit. To enhance usability of the tool by a broad range of users, the tool is designed to be flexible allowing user interaction through a graphical user interface.