Non-point source (NPS) pollution is an important problem that has been threatening freshwater resources throughout the world. Best Management Practices (BMPs) can reduce NPS pollution delivery to receiving waters. For economic reasons, BMPs should be placed at critical source areas (CSAs), which are the areas contributing most of the NPS pollution. The CSAs are the areas in a watershed where source coincides with transport factors, such as runoff, erosion, subsurface flow, and channel processes. Methods ranging from simple index-based to detailed hydrologic and water quality (HWQ) models are being used to identify CSAs. However, application of these methods for Canadian watersheds remains challenging due to the diversified hydrological conditions, which are not fully incorporated into most existing methods. The aim of this work is to review potential methods and challenges in identifying CSAs under Canadian conditions. As such, this study: (a) reviews different methods for identifying CSAs; (b) discusses challenges and the current state of CSA identification; and (c) highlights future research directions to address limitations of currently available methods. It appears that applications of both simple index-based methods and detailed HWQ models to determine CSAs are limited in Canadian conditions. As no single method/model is perfect, it is recommended to develop a ‘Toolbox’ that can host a variety of methods to identify CSAs so as to allow flexibility to the end users on the choice of the methods.

Water security is the capability of a community to have adequate access to good quality and a sufficient quantity of water as well as safeguard resources for the future generations. Understanding the spatial and temporal variabilities of water security can play a pivotal role in sustainable management of fresh water resources. In this study, a long-term water security analysis of the Grand River watershed (GRW), Ontario, Canada, was carried out using the soil and water assessment tool (SWAT). Analyses on blue and green water availability and water security were carried out by dividing the GRW into eight drainage zones. As such, both anthropogenic as well as environmental demand were considered. In particular, while calculating blue water scarcity, three different methods were used in determining the environmental flow requirement, namely, the presumptive standards method, the modified low stream-flow method, and the variable monthly flow method. Model results showed that the SWAT model could simulate streamflow dynamics of the GRW with ‘good’ to ‘very good’ accuracy with an average Nash–Sutcliffe Efficiency of 0.75, R² value of 0.78, and percentage of bias (PBIAS) of 8.23%. Sen’s slope calculated using data from over 60 years confirmed that the blue water flow, green water flow, and storage had increasing trends. The presumptive standards method and the modified low stream-flow method, respectively, were found to be the most and least restrictive method in calculating environmental flow requirements. While both green (0.4–1.1) and blue (0.25–2.0) water scarcity values showed marked temporal and spatial variabilities, blue water scarcity was found to be the highest in urban areas on account of higher water usage and less blue water availability. Similarly, green water scarcity was found to be highest in zones with higher temperatures and intensive agricultural practices. We believe that knowledge of the green and blue water security situation would be helpful in sustainable water resources management of the GRW and help to identify hotspots that need immediate attention.

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.

The detrimental impacts of agricultural subsurface tile flows and their associated pollutants on water quality is a major environmental issue in the Great Lakes region and many other places globally. A strong understanding of water quality indicators along with the contribution of tile-drained agriculture to water contamination is necessary to assess and reduce a significant source of non-point source pollution. In this study, DRAINMOD, a field-scale hydrology and water quality model, was applied to assess the impact of future climatic change on depth to water table, tile flow and associated nitrate loss from an 8.66 ha agricultural field near Londesborough, in Southwestern Ontario, Canada. The closest available climate data from a weather station approximately 10 km from the field site was used by the Ontario Ministry of Natural Resources and Forestry (MNRF) to generate future predictions of daily precipitation and maximum and minimum air temperatures required to create the weather files for DRAINMOD. Of the 28 models applied by MNRF, three models (CGCM3T47-Run5, GFDLCM2.0, and MIROC3.2hires) were selected based on the frequency of the models recommended for use in Ontario with SRA1B emission scenario. Results suggested that simulated tile flows and evapotranspiration (ET) in the 2071–2100 period are expected to increase by 7% and 14% compared to 1960–1990 period. Results also suggest that under future climates, significant increases in nitrate losses (about 50%) will occur along with the elevated tile flows. This work suggests that climate change will have a significant effect on field hydrology and water quality in tile-drained agricultural regions.

In the last decade, Lake Erie, one of the great lakes bordering Canada and the USA has been under serious threat due to increased phosphorus levels originating from agricultural fields. Large scale watersheds contributing to Lake Erie from the USA side are being simulated using hydrological and water quality (H/WQ) models such as the Soil and Water Assessment Tool (SWAT) and the results from the model are being used by policy and decision makers to implement better management decisions to solve emerging phosphorus issues. On the Canadian side, modeling applications are limited to either small watersheds or one major watershed contributing to Lake Erie. To the best of our knowledge, no efforts have been made to model the entire contributing watersheds to Lake Erie from Canada. This study applied the SWAT model for Northern Lake Erie Basin (NLEB; entire contributing basin to Lake Erie). Various provincial, national and global inputs of weather, land use and soil at various resolutions was assessed to evaluate the effects of input data types on the simulation of hydrological processes and streamflows. Twelve scenarios were developed using the input combinations and selected scenarios were evaluated at selected locations along the Grand and Thames Rivers using model performance statistics, and graphical comparisons of time variable plots and flow duration curves (FDCs). In addition, various hydrological components such as surface runoff, water yield, and evapotranspiration were also evaluated. Global level coarse resolution weather and soil did not perform better compared to fine resolution national data. Interestingly, in the case of land use, global and national/provincial land use were close, however, fine resolution provincial data performed slightly better. This study found that interpolated weather data from Environment Canada climate station observations performed slightly better compared to the measured data and therefore could be a good choice to use for large-scale H/WQ modeling studies.