Geospatial Analysis of Chloride Hot Spots and Groundwater Vulnerability in Southern Ontario, Canada

Abstract

Elevated chloride (Cl⁻) concentrations in surface water and groundwater are an increasing concern in cold region urban environments, largely due to long-term road salt application. This study investigates the Cl⁻ distribution across southern Ontario, Canada, using geospatial methods to identify contamination hot spots and assess groundwater vulnerability at both regional and watershed scales. Chloride data from 2001 to 2010 and 2011 to 2020 were compiled from public sources and interpolated using inverse distance weighting. A regional-scale vulnerability index was developed using slope (SL), surficial geology (SG), and land use (LU) (SL-SG-LU), and compared it to a more detailed DRASTIC-LU index within the Credit River watershed. Results show that Cl⁻ hot spots are concentrated in urbanized areas, including the Greater Toronto Area and Golden Horseshoe, with some rural zones also exhibiting elevated concentrations. Vulnerability mapping corresponded well with the observed Cl⁻ patterns and highlighted areas at risk for groundwater discharge to surface waters. While the DRASTIC-LU method offered finer resolution, the simplified SL-SG-LU index effectively captured broad vulnerability trends and is suitable for data-limited regions. This work provides a transferable framework for identifying Cl⁻ risk areas and supports long-term monitoring and management strategies in cold climate watersheds.

1. Introduction

In urban cold regions, elevated chloride (Cl⁻) concentrations in groundwater and surface waters have become a growing environmental concern. The intensive application of road salt (principally NaCl) during winter months is the primary source of Cl⁻ loading to the environment in urban areas across Canada and the northern United States [1,2,3]. Once applied, Cl⁻ is highly mobile and persistent [4], entering both surface and groundwater systems, where it can have long-lasting environmental and infrastructure impacts [1,5,6].

Within the province of Ontario in Canada, Cl⁻ concentrations have steadily increased in recent decades, particularly in urban areas, and frequently surpass the Canadian Water Quality Guidelines for aquatic health [7,8,9,10,11], which set chronic and acute thresholds at 120 mg/L and 640 mg/L, respectively. While management initiatives, such as the Code of Practice for Road Salt, were introduced in 2004 [12], long-term monitoring in the province suggests that Cl⁻ concentrations have continued to rise, likely due to both ongoing road salt applications and legacy accumulation in groundwater [13,14].

Legacy Cl⁻ contamination, the accumulation and retention of Cl⁻ in subsurface environments over time, has received less attention than the effects of more rapid pathways, such as surface runoff. However, this legacy Cl⁻ plays a crucial role in sustaining elevated concentrations in groundwater and influencing baseflow contributions to surface water systems [2,15]. Identifying where this legacy contamination may persist and intersect with vulnerable landscapes is critical for effective long-term water management.

Previous studies have documented spatial and temporal trends in Cl⁻ contamination and established broad relationships with land use [16,17,18,19]. However, few have combined these insights with geospatial analysis methods to develop a transferable vulnerability framework that also considers legacy effects. Overlay and index-based approaches, such as the DRASTIC method [20], offer a practical, data-efficient means to assess vulnerability, especially when detailed process-based modeling is infeasible [21,22].

This study maps and analyzes Cl⁻ concentrations in both groundwater and surface water across southern Ontario for two decadal periods (2001–2010 and 2011–2020) by plotting monitoring data and generating spatial interpolations using inverse distance weighting (IDW). This approach supports the visualization and investigation of spatio-temporal variability in Cl⁻ concentrations. This study further evaluates how identified Cl⁻ hotspots correspond to landscape characteristics by applying a novel composite vulnerability index (SL-SG-LU), which integrates slope (SL), surficial geology (SG), and land use (LU) across broad spatial scales. This index is assessed at both the regional scale (southern Ontario) and the watershed scale (Credit River), and its performance is compared to the widely used DRASTIC-LU (after [20]) method. In addition, this study explores relationships between baseflow contributions and surface water Cl⁻ concentrations, particularly in areas identified as having elevated groundwater vulnerability.

The geospatial techniques presented here offer a useful tool for identifying and prioritizing Cl⁻-vulnerable areas in cold climate regions. This approach can support watershed managers, municipalities, and policy makers in developing more effective strategies for monitoring, managing, and mitigating Cl⁻ contamination over both short- and long-term time frames.

2. Materials and Methods

Figure 1 presents a detailed flowchart of the methodology used in this study. The methodology will be explained in the next subsections.

2.1. Identification of the Study Area and Data Collection

Numerous studies and programs monitoring aquifer and surface water body quality exist across the province of Ontario. Given that Cl⁻ is a common ion, it is typically sampled under these monitoring programs, with the historical data ultimately publicly shared. However, there is a need for a comprehensive review, compilation, and geospatial analysis of these data.

Chloride concentration data for both surface water (streams and rivers) and groundwater have been collected from publicly available databases in Ontario. The data sources that were accessed to obtain historical Cl⁻ data can be found in Table 1, with the temporal range of data availability also presented.

2.2. Data Processing

To ensure consistency for geospatial visualization, all coordinates that were not already in UTM (Universal Transverse Mercator) format were converted accordingly. Many data sets were missing basic information, such as location coordinates or station names. Sampling locations where this could not be obtained were excluded from the comprehensive data set.

To minimize short-term variability due to seasonal conditions and highlight long-term contamination trends, average concentrations were computed for two decadal periods: 2001–2010 and 2011–2020. These 10-year averages help mitigate the effects of variable monitoring frequency and data sparsity and can help reveal legacy or historical Cl⁻ contamination in groundwater systems (e.g., [1]).

To organize the data, each data set was grouped into decades (Table 1) starting from the earliest available data range (1961) to virtually the present (2020). To represent the difference between general overburden and bedrock groundwater conditions, the data were separated based on aquifer type. Furthermore, the focus of this analysis is southern Ontario (Figure 2); therefore, any data collected north of this region were excluded. Groundwater data were further filtered based on a screen depth greater or less than 20 m below ground surface. This threshold was selected based on the classification of shallow wells per the Oak Ridges Moraine Groundwater Program [23] and was employed for consistency in the DRASTIC-LU analysis.

Table 1. Chloride data sets integrated into the geospatial analyses and the corresponding program details.

Water Source	Data Set	Available Data Periods	Sampling Occurrence	Temporal Coverage	Spatial Distribution	Reference
Surface water	Dorset Lakes	1971–2020	Annual to variable	Variable	Inland lakes, north-central Ontario	[24]
	Great Lakes Intake Program	1971–2020	Weekly to biweekly	Year-round	Nearshore Great Lakes and Lake Simcoe	[25]
	Lake Simcoe	1981–2020	Biweekly	Ice-free seasons (spring–fall)	Lake Simcoe	[26]
	Lake Partner Program	2011–2020	Biweekly to monthly	Ice-free seasons (spring–fall)	Provincial	[27]
	Provincial (Stream) Water Quality Monitoring Network (PWQMN)	1961–2020	Monthly	Ice-free seasons (spring–fall)	Provincial coverage	[28]
Groundwater	Ambient Groundwater Geochemistry	2001–2020	Predominantly annually	Ice-free seasons (spring–fall)	Provincial coverage	[29]
	Oak Ridges Moraine Groundwater Program	1971–2020	Variable	Year-round	Central to central east Ontario, Oak Ridges Moraine at the core	[23]
	Provincial Groundwater Monitoring Network (PGMN)	2001–2020	Annually	Ice-free seasons (spring–fall)	Provincial coverage	[30]
Surface water and groundwater	Drinking Water Surveillance Program	1991–2020	Minimum annually	Year-round	Provincial coverage; lakes (inland lakes and Great Lakes), rivers, and groundwater	[31]

Table 1. Chloride data sets integrated into the geospatial analyses and the corresponding program details.

Water Source	Data Set	Available Data Periods	Sampling Occurrence	Temporal Coverage	Spatial Distribution	Reference
Surface water	Dorset Lakes	1971–2020	Annual to variable	Variable	Inland lakes, north-central Ontario	[24]
	Great Lakes Intake Program	1971–2020	Weekly to biweekly	Year-round	Nearshore Great Lakes and Lake Simcoe	[25]
	Lake Simcoe	1981–2020	Biweekly	Ice-free seasons (spring–fall)	Lake Simcoe	[26]
	Lake Partner Program	2011–2020	Biweekly to monthly	Ice-free seasons (spring–fall)	Provincial	[27]
	Provincial (Stream) Water Quality Monitoring Network (PWQMN)	1961–2020	Monthly	Ice-free seasons (spring–fall)	Provincial coverage	[28]
Groundwater	Ambient Groundwater Geochemistry	2001–2020	Predominantly annually	Ice-free seasons (spring–fall)	Provincial coverage	[29]
	Oak Ridges Moraine Groundwater Program	1971–2020	Variable	Year-round	Central to central east Ontario, Oak Ridges Moraine at the core	[23]
	Provincial Groundwater Monitoring Network (PGMN)	2001–2020	Annually	Ice-free seasons (spring–fall)	Provincial coverage	[30]
Surface water and groundwater	Drinking Water Surveillance Program	1991–2020	Minimum annually	Year-round	Provincial coverage; lakes (inland lakes and Great Lakes), rivers, and groundwater	[31]

2.3. Spatial Interpolation

Due to the scarcity of available groundwater Cl⁻ monitoring data pre-2000, only the two most recent decades (2001–2010 and 2011–2020) were selected to complete the vulnerability analysis.

2.3.1. Groundwater Chloride Interpolation

Given that the compiled concentration data are point-based and not represented by a homogenous map, the data sets were interpolated to create a systematic representation of groundwater Cl⁻ concentrations across southern Ontario in 2001–2010 (n = 2000) and 2011–2020 (n = 1852). A separate interpolation was conducted individually for the overburden and bedrock groundwater concentrations for the specified decade. Concentration data were also separated for wells less than or greater than 20 m deep (based on a shallow well classification per [23]) to evaluate any patterns with depth. Data that were coincident at a spatial location were averaged for the specified time period. Inverse Distance Weighting interpolation in Geospatial Information System (GIS) software, ArcGIS Pro 3.5.1, was applied to the groundwater Cl⁻ concentration data to develop continuous spatial representations of the Cl⁻ distribution across southern Ontario for both decades. Inverse Distance Weighting is a deterministic spatial interpolation technique that assigns estimated values to unsampled locations based on the proximity of surrounding data points, with greater influence given to observations that are closer [32]. It is well-suited for water quality mapping where data are unevenly distributed across an area, and its methodological simplicity and ease of implementation render it suitable for regional-scale screening analyses [33]. This method has been employed in previous groundwater and surface water quality studies, as evidenced by [34,35,36], among others.

To evaluate potential artifacts arising from changing monitoring locations over time, a spatial (point) density analysis of monitoring locations for each decade was conducted (see Supplementary Material Figure S1 for maps). While IDW cannot eliminate uncertainty from data gaps, it supports the visualization of general patterns and a comparison over time to identify emerging or persistent hotspots in Cl⁻ concentrations [37].

2.3.2. Surface Water Chloride Interpolation

Surface water Cl⁻ concentrations (n = 501 for 2001–2010; n = 564 for 2011–2020) were also interpolated using IDW; however, they were constrained to major surface water features in the study area. This was completed by extracting (via the mask function) the interpolated surface to the water courses in the 1M data set produced by the Ontario Hydro Network [38]. Only samples associated with flowing water bodies were included. While IDW interpolation across watershed boundaries introduces limitations, it was employed here primarily as a visualization tool, with interpolated values extracted along major river courses. This constraint helps maintain physiographic relevance and reduces the risk of misinterpreting interpolated patterns.

Interpolated surface water layers were not used in the statistical validation of the vulnerability indices. Instead, they serve to support qualitative assessments of spatial correspondence between the baseflow index (BFI), groundwater contamination, and potential areas of surface water impact.

2.3.3. Visualization of Chloride Patterns and Assessing Spatial Relationships

The change in overburden concentrations from 2001–2010 to 2011–2020 was calculated using the compute change function in the GIS software. This function calculates the mathematical difference between the pixels in both rasters, with a positive value indicating an increase in concentration, and a negative value suggesting a decrease in concentrations has occurred.

Furthermore, in addition to the interpolated concentration data, to illustrate Cl⁻ hot spots, the average concentrations for both decades and water sources at each of the monitoring locations were represented graphically. Analyzing the point and interpolated Cl⁻ data enhances the understanding of spatial patterns and variations in concentrations across the study period.

To preliminarily assess the spatial relationships between Cl⁻ concentrations and road proximity, Pearson’s correlation analyses were conducted in Excel between Cl⁻ concentrations in groundwater and surface water and GIS-derived metrics of distance to the nearest road. Analyses were performed separately for the 2001–2010 and 2011–2020 time periods.

2.4. Baseflow Data

To examine groundwater–surface water (GW-SW) interactions, a BFI raster was developed for southern Ontario using data from [39]. BFI values for watersheds in Ontario were estimated in [39] through hydrographic separation of historical streamflow records at gauged sites in the GLB. For ungauged areas, the BFI was estimated via a regression model that related the BFI to surficial geology and surface water coverage, though specific site counts for the regression model were not detailed [39]. Baseflow indices for quaternary watersheds in this study were presented as ranges; thus, to include these data as a geospatial layer, the median of each range was used. These data were then geospatially referenced and converted into a raster layer, which is available in the Supplementary Material (Figure S2).

The BFI indicates the proportion of streamflow derived from groundwater and was overlaid with Cl⁻ and vulnerability maps to explore spatial coincidence. High-BFI areas with high Cl⁻ and vulnerability scores suggest locations where legacy groundwater contamination may impact surface water via baseflow. Baseflow indices were used qualitatively to guide the interpretation of the risk to streams from subsurface Cl⁻ transport.

2.5. Vulnerability Index Development (Large-Scale)

To evaluate regional groundwater vulnerability to Cl⁻ contamination, a simplified composite index referred to as SL-SG-LU was developed, based on three spatial predictors: land use [40], surficial geology [41], and slope [42]. The data for these layers are publicly accessible, available for the entire study area, and are utilized in existing index-based methods, including DRASTIC [20].

Previous research findings indicate that the key factors affecting the dynamics of Cl⁻ in the subsurface include the permeability of the soil and vegetation cover, in addition to the presence of roadside drainage [43], which was not available for the current study given the scale. A workflow diagram is illustrated in Figure 3 with an overview of the process for this and subsequent analyses.

The weighted overlay analysis was carried out using the raster calculator (spatial analyst tool) in ArcGIS Pro software. To preprocess the data for the analysis, the cells in the three raster layers were reclassified based on their relative weight from 1 to 10. This ensures that the data are scored to the same scale. Higher scores reflect a stronger influence of that parameter on Cl⁻ vulnerability in this region. The reclassified map layers can be found in the Supplementary Material Figure S3.

The land use raster layer was reclassified using four grouped categories: vegetated/natural, agriculture, impervious built-up, and transportation. All vegetated and natural land uses (e.g., forests and grasslands) were scored with a weight of 1, agriculture 4, impervious built-up 6, and transportation 10, based on Cl⁻ uses and application in various land uses (e.g., [44]). While road salt is the primary source of Cl⁻ in urbanized areas [2], other sources such as fertilizer (e.g., KCl) runoff and discharge from water softeners to septic systems can contribute significantly to Cl⁻ loads in rural watersheds [44,45].

The surficial geology raster was categorized using the permeability of the units. When overburden permeability is low, less water is able to infiltrate into the soil, increasing the likelihood of surface runoff. Permeability categories were rated as follows [46]: low = 1, medium–low = 4, medium = 6, medium–high = 8, high = 10, variable = 8.

Slope was calculated from a 10 m resolution provincial digital elevation model (DEM) [42] and reclassified using the following formula (Equation (1)):

$$\left(1-{\displaystyle \frac{{SL}_i-{SL}_{min}}{{SL}_{max}-{SL}_{min}}}\right)\times 10=SL$$

(1)

This reclassification assigns higher values to flatter slopes to reflect increased vulnerability due to slower runoff and higher infiltration potential [20,47].

Following this reclassification, a numerical weighting factor (% of influence weight) was assigned to each layer according to its relative influence in relation to all the other layers. The weighting factor for each layer is detailed as follows: land use 5, surficial geology 3, and slope 1. These weightings were established by combining the original DRASTIC model [20], which assigns weightings for surficial geology and slope based on their conductivity and runoff influence, and recent GIS-based vulnerability studies that incorporate land use/land cover factors (e.g., [46,48,49]). Specifically, land use was assigned the highest weight (5) because anthropogenic factors such as road salt application in urbanized and transportation areas significantly influence Cl⁻ contamination [46,49]. Surficial geology received a weight of 3, reflecting its role in determining the permeability of the landscape, which governs how easily Cl⁻, a mobile contaminant, can infiltrate groundwater [20]. Slope, while important in influencing runoff and infiltration, was assigned the lowest weight (1), as it has a less direct influence on Cl⁻ migration compared to land use and geology. These weightings reflect the relative of each parameter based on available data and previous studies, rather than a strictly quantitative measure.

To complete the weighted overlay analysis, each reclassified layer was multiplied by the assigned weighting factor using the raster calculator tool (Equation (2)). The output of this analysis is a raster illustrating relative spatial vulnerability of groundwater-based contamination using slope (SL), surficial geology (SG), and land use (LU).

$$\mathrm{SL}\text{-}\mathrm{SG}\text{-}\mathrm{LU} \mathrm{Vulnerability}=\mathrm{SL}·1+\mathrm{SG}·3+\mathrm{LU}·5$$

(2)

2.6. DRASTIC-LU Vulnerability Mapping

A comprehensive province-wide DRASTIC analysis was not feasible due to the inconsistent availability of key data layers (e.g., water table depth and recharge). Therefore, DRASTIC-LU was applied at a watershed scale within the Credit River watershed, where sufficient hydrogeological data were available.

The Credit River watershed, located in the western portion of the Lake Ontario Basin within the Greater Toronto Area (GTA), was selected due to persistently high and increasing Cl⁻ concentrations in both surface water and groundwater [50,51]. The region is undergoing rapid urbanization, leading to increased impervious surfaces and greater road salt application [52].

The DRASTIC index evaluates groundwater vulnerability using seven hydrogeological parameters: depth to water (D), net recharge (R), aquifer media (A), soil media (S), topography (T), impact of the vadose zone (I), and hydraulic conductivity (C) [20]. A land use (LU) modifier layer was added to reflect Cl⁻-specific anthropogenic pressures (e.g., from road salting, agriculture, and industrial activities), following [53]. Each parameter was weighted and rated based on standard DRASTIC values, the details of which are summarized in

Table 2. Input parameters for the DRASTIC-LU analysis, including the class, rating, and source of each layer.

Weight	Layer	Class	Rating	Source
5	Depth to water (D)	Min-max	1–10	[23]
4	Recharge (R)	Min-max	1–10	[54]
3	Aquifer media (A)	Bedrock	2	[41]
		Glaciofluvial ice contact deposits	8
		Glaciofluvial outwash deposits	8
		Glaciolacustrine coarse deposits	8
		Glaciolacustrine fine deposits	6
		Tills	5
2	Soil media (S)	Clay	2	[55]
		Clay loam	3
		Fine sandy loam	4
		Loam	5
		Loamy sand	6
		Organic	7
		Silt loam	4
		Sandy loam	6
		Variable	5
		N/A	3
1	Topography (T)	Min-max	1–10	[42], estimated from the provincial DEM
5	Impact of vadose zone (I)	Bedrock	3	[41]
		Clay and silt	3
		Clay, silt, sand, and gravel	5
		Diamicton	5
		Gravel	9
		Organic deposits	6
		Sand	7
		Sand and gravel	8
		Silt, sand, and gravel	7
3	Conductivity (C)	Min-max	1–10	[23]
5	Land Use (LU)	Vegetated/natural	1	[40]
		Agriculture	4
		Impervious built-up	6
		Transportation	10

. The LU layer was classified using the same categories and scores as in the SL-SG-LU method, reflecting the influence of anthropogenic activity on Cl⁻ contamination, particularly in urbanized areas. All parameter weights were based on DRASTIC guidance [20]. The resulting vulnerability index was computed as follows (Equation (3)):

$$\mathrm{DRASTIC}\text{-}\mathrm{LU} \mathrm{Vulnerability} = \mathrm{D}·5 + \mathrm{R}·4 + \mathrm{A}·3 + \mathrm{S}·2 + \mathrm{T}·1 + \mathrm{I}·5 + \mathrm{C}·3 + \mathrm{LU}·5$$

(3)

Each parameter was reclassified to simplify the raw data into representative categories aligned with relative vulnerability. The scoring scale from 1 to 10 is based on the expected impact of each factor on groundwater vulnerability to Cl⁻ contamination, where higher scores reflect a greater potential for Cl⁻ infiltration or accumulation. Reclassified maps of the eight input layers for the watershed can be referenced in the Supplementary Material Figure S4.

2.7. Sensitivity Analysis

To examine the impact of each input layer on the vulnerability assessment, a map removal sensitivity analysis was conducting following the approach of [56]. This method was applied to both the regional (large-scale) and watershed-scale overlay analyses. The goal is to evaluate how the inclusion or exclusion of specific parameters influences the overall vulnerability index, helping to identify which layers are most influential in the study area.

The analysis was conducted using Equation (4), where Sen is the sensitivity associated with the removal of one map layer (X), R is the original risk index value (computed based on Equations (2) and (3)) using N map layers, and Rx is the risk index calculated when excluding one map layer.

$$Sen=\left|{\displaystyle \frac{R}{N}}-{\displaystyle \frac{R_x}{N-1}}\right|$$

(4)

2.8. Validation

The relationship between the observed groundwater Cl⁻ concentrations (2011–2020, n = 101) and the vulnerability indices generated from the two analyses of the Credit River watershed was assessed using Pearson’s correlation coefficient. This coefficient, which ranges from −1 to +1, indicates both the strength and direction of the correlation. It is important to note these indices are intended as screening tools, not precise predictive models.

3. Results

3.1. Spatial and Temporal Patterns of Regional Groundwater Chloride Concentrations

The highest average groundwater concentrations (250 to 5000 mg/L) within southern Ontario for both decades (Figure 4a,b) were typically observed near main urban centres (e.g., GTA, Barrie, Sarnia, and Niagara region) and the central east (Kingston–Pembroke) region of Ontario (Figure 4). Between 2001 and 2010 (Figure 4a), concentrations above 250 mg/L were recorded predominantly in southwest (London and Windsor–Sarnia), central west regions (Kitchener–Waterloo–Barrie and Toronto), and the GTA. Notably, elevated concentrations measured in the southwest, particularly between Sarnia and Chatham, occurred in areas characterized predominantly by rural land use.

The spatial distribution of groundwater monitoring locations shifted notably between the two decades. In 2001–2010, the southwest and central west regions (e.g., Windsor–Sarnia, London, and Kitchener–Waterloo–Barrie) had more extensive monitoring, whereas in 2011–2020, monitoring became more concentrated in the GTA and central east (e.g., Kingston–Pembroke), with sparser coverage in the southwest. A spatial density analysis (S1) further illustrates this transition. Although the total number of sampling points decreased slightly from 2000 (2001–2010) to 1852 (2011–2020), the later decade exhibits greater point density and broader spatial coverage, especially around the GTA, Kitchener, and Barrie, highlighting a focused monitoring effort in these key urban areas.

Because of this shift, direct point-to-point temporal comparisons were not possible. Instead, the interpolated surfaces represent decadal patterns derived from two differing monitoring networks. Nevertheless, the persistence of elevated Cl⁻ concentrations in the southwest, despite fewer monitoring locations in 2011–2020, suggests that these trends are not solely artifacts of spatial sampling variability.

To summarize the changing distribution of concentrations, monitoring points were categorized into concentration ranges (Figure 4a,b), and the percentage of points in each range was computed (

Table 3. Groundwater monitoring points categorized by concentration range and decade.

Concentration Range (mg/L)	2001–2010	Percentage (%) of Locations	2011–2020	Percentage (%) of Locations
0–15	989	49	829	45
15–50	430	22	446	24
50–100	193	10	219	12
100–150	105	5	115	6
150–200	50	3	51	3
200–250	49	2	41	2
>250	184	9	151	8

). This proportion-based approach helps mitigate the influence of shifting monitoring locations by evaluating relative distributions rather than fixed-point changes. Most monitoring points (49%) in 2001–2010 had Cl⁻ concentrations in the 0–15 mg/L range, decreasing slightly to 45% in 2011–2020. The 15–50 mg/L range was consistently represented in both decades (22% and 24%), while the 50–100 mg/L range rose from 10% to 12%. Higher concentration ranges exhibited only minor fluctuations.

To evaluate temporal changes more robustly, a subset analysis was conducted using only the 720 groundwater monitoring wells with data available in both decades. Among these consistently monitoring sites, 50.3% exhibited increased average Cl⁻ concentrations. The mean and median changes were 4.5 and 0.02 mg/L, respectively, indicating a slight overall upward trend. The full range of changes in average concentrations between the two decades spanned from −922.4 to 1266.5 mg/L.

Separating the overburden from bedrock monitoring locations enabled a comparison of Cl⁻ trends across different geological materials. Overburden concentrations in 2001–2010 mirrored general trends, with guideline exceedances near Sarnia, Chatham, Niagara Falls, the Golden Horseshoe, and Kingston (Figure 5a). By 2011–2020, these zones expanded, with intensified hot spots in the GTA, Lake Ontario corridor, Sarnia–Chatham, and Niagara (Figure 5b). The compute-change GIS tool estimated a decadal range from a 2422 mg/L decrease in some locations to a 1317 mg/L increase (Figure 5c).

Bedrock aquifer patterns in 2001–2010 were comparable to those in the overburden, with locally elevated zones in Kingston–Pembroke, GTA, and southwestern Ontario (Figure 6a). In 2011–2020, bedrock concentrations became more variable with a greater number of scattered hotspots (Figure 6b). Estimated changes showed a maximum 1731 mg/L increase and a maximum 9350 mg/L decrease (Figure 6c). However, the apparent decrease in concentrations during 2011–2020 may partially reflect changes in monitoring density and spatial coverage. While high concentrations persisted in certain areas, the addition of new monitoring wells, particularly in surrounding regions with lower Cl⁻ levels, may have reduced the influence of earlier, more localized hotspots in the interpolated surfaces. As a result, observed declines in regional averages or hotspot intensity may not reflect real improvements in water quality, but instead reflect changes in how and where sampling was conducted.

Analyses of wells less than and greater than 20 m deep showed no clear difference in Cl⁻ concentrations. Spatial trends were also broadly similar between the two depth categories.

3.2. Spatial and Temporal Patterns of Regional Surface Water Chloride Concentrations

Following the pattern observed in groundwater, average surface water Cl⁻ concentrations over the two decades were highest near urban centres (Figure 7a,b). Elevated concentrations were consistently observed in the GTA, Kitchener–Waterloo, Hamilton, and Windsor areas in both 2001–2010 and 2011–2020.

A key difference between the two decades is the growing number of monitoring locations where average Cl⁻ concentrations now exceed the Canadian freshwater quality guidelines: 120 mg/L for chronic (long-term) exposure and 640 mg/L for acute (short-term) exposure (

Table 4. Surface water monitoring points categorized by concentration range and decade.

Concentration Range (mg/L)	2001–2010	Percentage (%) of Locations	2011–2020	Percentage (%) of Locations
0–10	55	11	121	21
10–50	304	61	309	55
50–120	100	20	87	15
120–250	34	7	31	5
250–380	4	1	11	2
380–640	3	1	2	0.4
>640	1	0.2	3	0.5

). The largest increases in exceedances occurred in highly urbanized areas such as the GTA and Kitchener region.

A subset analysis of surface water monitoring sites with data available in both decades (n = 383) was conducted to assess changes independent of shifting monitoring locations. Of the consistently monitored sites, 57% exhibited increased Cl⁻ concentrations between 2001–2010 and 2011–2020. The mean and median change was 6.2 and 0.48 mg/L, respectively, with values ranging from −65.0 to 600.0 mg/L.

Interpolated surface water concentrations, constrained to major water features, are shown in Figure 8a,b. Estimated changes in average surface water Cl⁻ concentrations between the two decades ranged from a decrease of 159 mg/L to an increase of 270 mg/L (Figure 8c). The greatest decreases (>100 mg/L) occurred near Sarnia, along Lake Erie, and north of the GTA, whereas the most significant increases (~100–120 mg/L) were observed in the GTA, Kitchener, and Windsor.

3.3. Groundwater–Surface Water Hot Spots and Spatial Drivers

The highest coinciding Cl⁻ concentrations were observed in the GTA. Between 2001–2010 and 2011–2020, hot spot locations shifted from more localized clusters of elevated concentrations to a broader zone along the western Lake Ontario shoreline.

Additional areas with both elevated groundwater and surface water Cl⁻ concentrations were identified across the broader Golden Horseshoe region and in southwestern Ontario, particularly between Sarnia and Chatham.

Some locations, such as areas near Belleville and the Bay of Quinte, parts of southwestern Ontario, and sections of the Niagara region, exhibited elevated Cl⁻ concentrations in groundwater while surface water concentrations remained below or near guideline thresholds.

The combined raster of the BFI and groundwater Cl⁻ concentrations (2011–2020; Figure 9) indicates that stream vulnerability to Cl⁻-rich groundwater discharge is highest in southeastern Ontario (e.g., Kingston area) and along the western Lake Ontario shoreline (e.g., GTA). A Pearson correlation analysis was conducted between surface water Cl⁻ concentrations and the combined BFI and groundwater raster. The results (r = 0.103, n = 559, p = 0.015) suggests a slight positive relationship between high baseflow contributions and elevated stream Cl⁻ concentrations.

Potential spatial drivers of elevated Cl⁻ concentrations were further explored, and the Pearson correlation analysis revealed very weak and mostly non-significant correlations between Cl⁻ concentrations and proximity to roads. Groundwater correlations were near zero in both decades (2001–2010: R = −0.029, p = 0.22; 2011–2010: R= −0.032, p = 0.16), and surface water showed a weak but statistically significant negative correlation only in 2001–2010 (R = −0.16, p < 0.001), but not thereafter in 2011–2020 (R = 0.0005, p = 0.99).

3.4. Regional Vulnerability Assessment via the SL-SG-LU Overlay Method

Using the large-scale weighted overlay analysis incorporating slope, surficial geology, and land use, a spatially continuous map of estimated groundwater vulnerability was generated (Figure 10a). The greatest vulnerability was found to be concentrated along the western Lake Ontario shoreline, extending from the GTA through the Golden Horseshoe to the Niagara region. High vulnerability areas match well with elevated groundwater and surface water Cl⁻ concentrations (Figure 10b and Figure 10c, respectively). Additional high-vulnerability zones were observed near Barrie, Kitchener–Waterloo, London, Sarnia, and Chatham; areas commonly associated with urban development and dense roadways.

Across the study area, 25% of land was classified as low vulnerability, 58% as medium, 15% as high, and 2% as very high vulnerability (

Table 5. Vulnerability classes and associated indices from the two methodologies, adapted from [56].

Assigned Vulnerability Class	Vulnerability Indices with the SL-SG-LU Method	Vulnerability Indices with the DRASTIC-LU Method
Very low	<10	<75
Low	10–30	75–100
Medium	30–50	100–125
High	50–70	125–150
Very high	70–100	>150

). No areas were categorized as very low vulnerability.

3.5. Comparing Groundwater Vulnerability in the Credit River Watershed: DRASTIC-LU vs. SL-SG-LU

To examine how the choice of method affects the vulnerability classification, both SL-SG-LU and DRASTIC-LU analyses were applied to the Credit River watershed, a region in southern Ontario that was identified as having Cl⁻ hotspots, as well as comprehensive hydrogeological data sets (Figure 11).

The SL-SG-LU map of the Credit River watershed (Figure 11a) shows a broader, more generalized pattern with large adjoining areas of medium and high vulnerability. In contrast, the DRASTIC-LU output (Figure 11b) presents a more detailed and spatially heterogeneous vulnerability landscape.

These differences are further illustrated in Figure 12, which compares the proportion of land in each vulnerability class. The DRASTIC-LU approach identified a small fraction of land as very low vulnerability; a class absent from the SL-SG-LU results. Despite these differences, both methods identified a similar proportion of land as very high vulnerability (<6%).

3.6. Sensitivity AnalysisOutcomes

The map removal sensitivity analysis, based on [57] assessed the impact of excluding specific map layers on the vulnerability indices estimated using the two methodologies. For southern Ontario, the removal of surficial geology, slope, and land use produced distinct sensitivity values. Surficial geology exhibited the highest sensitivity (7.42), followed by slope (5.90) and land use (4.73). At the Credit River watershed scale, slope (9.30) emerged as the most critical parameter, with surficial geology (7.67) and land use (5.13) following.

Within the DRASTIC-LU framework for the watershed, conductivity showed the highest sensitivity (2.82), while the vadose zone and recharge also displayed notable sensitivities (2.64 and 2.25). Land use (2.19) and depth to the water table (1.70) contributed meaningfully, whereas soil media, topography, and aquifer media had lower sensitivities.

3.7. Validation Metrics

Correlation coefficients between the observed groundwater Cl⁻ concentrations (2011–2020) and the two vulnerability assessments in the Credit River watershed were examined. The SL-SG-LU analysis showed a weak positive correlation (r = 0.176, R² = 0.03). In contrast, the DRASTIC-LU method demonstrated a negative correlation (r = −0.222, R² = 0.05).

4. Discussion

4.1. Spatial and Temporal Patterns of Groundwater Chloride Concentrations

While rural or less urbanized areas typically have lower Cl⁻ concentrations, prior studies [2,8] have documented Cl⁻ accumulation in rural aquifers, suggesting legacy salt impacts or increased infiltration and reduced export of Cl⁻ from any source via surface water pathways. Elevated concentrations within the GTA and central west regions across both decades align with the localized trends of elevated groundwater Cl⁻ concentrations in urban areas reported in previous studies (e.g., [18,58,59,60]), with road salt repeatedly identified as the principal source [61]. These patterns are further supported by [15], which documented elevated and increasing Cl⁻ concentrations in GTA groundwater, attributing them primarily to road salt use and urban landscape factors. Rural Cl⁻ hot spots, observed particularly in Figure 4, Figure 5 and Figure 6, may also reflect non-urban sources, including agricultural fertilizer (e.g., KCl), septic systems, water softeners, and geogenic sources [2,44,45,61]. However, confirming these sources requires additional investigation, such as isotopic or geochemical analysis, to distinguish their contributions from other anthropogenic Cl⁻ sources [18].

Higher concentration ranges exhibited only minor fluctuations over time, suggesting that concentrations above the drinking water guideline (250 mg/L) have persisted but not significantly increased. These stable proportions highlight the importance of continued monitoring and potential mitigation strategies, particularly in regions where high concentrations subsist. The Kingston region’s thin drift cover (<25 m) suggests that bedrock aquifers are more vulnerable to surface contamination [62,63], similar to patterns observed in Minnesota, where shallow bedrock under a thin overburden yielded elevated Cl⁻ concentrations [64,65].

Groundwater Cl⁻ data separated into shallow (<20 m) and deep (>20 m) well classifications (per [23]) showed no consistent depth-related trends in Cl⁻ concentrations. This may reflect the influence of multiple Cl⁻ sources (e.g., both anthropogenic and geogenic contributions) across diverse hydrostratigraphic contexts or diffuse infiltration of Cl⁻ at the surface. It may also result from vertical migration via preferential pathways, such as fractures or comprised well casings. The absence of detailed surficial geology at individual well locations limited the ability to relate depth to specific geological units. Future work incorporating well-specific stratigraphic data and Cl⁻ concentrations would improve the understanding of how concentrations vary with depth.

Apparent decreases in concentrations during 2011–2020 may partially reflect changes in monitoring density and spatial coverage. While high concentrations persisted in certain areas, the addition of new monitoring wells, particularly in surrounding regions with lower Cl⁻ levels, may have reduced the influence of earlier, more localized hotspots in the interpolated surfaces. As a result, observed declines in regional averages or hotspot intensity may not reflect real improvements in water quality but rather changes in how and where sampling was conducted, suggesting that contamination is not limited to shallow systems and that vertical migration or legacy sources may play a role. These maps illustrate decadal changes across non-identical networks and should be interpreted as general regional shifts, not precise local trends.

The subset analysis of wells consistently monitored across both decades reflects both improvement and degradation across the study area. These results suggest that interpolated decadal patterns are broadly consistent with trends at sites sampled in both decades, supporting the reliability of the observed general regional changes despite the temporal variation in the monitoring site distribution. This suggests that the observed shifts may indicate broader regional dynamics, rather than solely reflecting changes in the monitoring network distribution.

Long-term studies document increases in Cl⁻ concentrations across southern Ontario. For example, Ref. [8] found increasing rates of 0.02–2.8 mg/L/year at 24 locations, with higher rates in developed watersheds. Ref. [66] recorded 21–34% increases in one year at four GTA sites, while Ref. [17] found a 6 mg/L/year rise in the Highland Creek watershed. Ref. [11] observed increasing trends of 20.4% and 17.1% at overburden and bedrock PGMN wells, respectively, between 2002 and 2020. The results from the present study support a broad trend of increasing Cl⁻ concentrations in overburden aquifers, though the rate and magnitude of increase may differ due to the inclusion of wells beyond the PGMN network and use of decadal averages. These differences underscore that this study’s findings should be interpreted as a regional assessment rather than a point-specific temporal trend, especially given the variability in data coverage and resolution across space and time.

A small but noteworthy increase (0.3%) in locations initially below or near guideline thresholds that now surpass them may pose risks to aquatic ecosystems. This aligns with the existing literature showing that Cl⁻ concentrations in urban watercourses often exceed 1000 mg/L and that surface waters draining urbanized areas consistently maintain these elevated concentrations [67,68,69]. These observations highlight the need for targeted strategies to better understand and manage Cl⁻ loading to surface water systems, particularly in urban areas.

4.2. Surface Water Chloride Trends and Monitoring Challenges

Surface water Cl⁻ concentrations followed a pattern similar to groundwater, with the western Lake Ontario basin, especially the GTA, remaining an area of concern due to elevated concentrations relative to aquatic ecosystem guidelines. These patterns are consistent with long-term monitoring data from the GTA and other urban centers in southern Ontario, where Cl⁻ concentrations continue to rise in response to intensified road salt use [10,11,70]. Ref. [69] similarly documented increasing Cl⁻ trends in eight of ten Lake Simcoe tributaries from 1993 to 2007, especially those draining heavily urbanized areas. Ref. [71] also found that median stream Cl⁻ concentrations at 20 GTA sites were significantly higher in 2012 than 2002, with 58% of samples exceeding the chronic guideline.

It is important to note that provincial monitoring programs such as the PWQMN typically sample surface water only during ice-free months (March/April to October/November). Consequently, trends inferred from these data may underestimate peak Cl⁻ concentrations, especially those associated with winter road salt applications. While winter sampling presents logistical and safety challenges, excluding this season omits critical data on short-term Cl⁻ loading via surface runoff. For example, Ref. [72] found that winter Cl⁻ concentrations in the Highland Creek, Humber River, and Don River watersheds (GTA) often exceeded 1500 mg/L and surpassed 230 mg/L in up to 80% of samples. Short-term pulses up to 3355 and 6000 mg/L have also been observed in a Credit River tributary [73] and Highland Creek [17,72], respectively.

The elevated Cl⁻ concentrations observed here, despite being based on non-salting season data, suggest contributions from slower, sustained sources such as groundwater discharge or potentially other anthropogenic inputs like wastewater or landfill leachate. Evidence of groundwater-derived Cl⁻ inputs aligns with prior findings across Ontario [8,74,75,76]. Surface water Cl⁻ loads in winter and early spring are largely driven by surface runoff and shallow groundwater contributions, while deeper groundwater discharge plays a more prominent role during summer and fall [17,74,77]. Therefore, the concentrations and modest increases observed in this study likely reflect persistent inputs from Cl⁻-contaminated groundwater and other sources, further reinforcing the need for year-round monitoring and mitigation.

These interpretations are supported by the subset analysis using only surface water sites monitored in both decades (n = 383), which showed that 57.4% of sites exhibited increased average Cl⁻ concentrations. These findings strengthen the broader decadal trends observed in the interpolated data, suggesting that increases in Cl⁻ concentrations are not solely a consequence of changes in the coverage of the monitoring network.

Once Cl⁻ reaches an aquifer, its transport depends on groundwater flow and residence times, which vary widely across the study area due to diverse hydrogeological conditions. Ref. [59] projected that residual Cl⁻ could take decades to flush from an aquifer in southern Ontario, even if road salt applications ceased completely. This long-term accumulation, combined with sustained baseflow contributions, suggests that reducing road salt application rates would result in notable delays in stream health and quality, particularly in high-BFI areas. These lags reinforce the need for long-term monitoring to capture legacy Cl⁻ dynamics.

4.3. Implications of Groundwater–Surface Water Interactions

Regions where elevated groundwater Cl⁻ levels overlap with high surface water concentrations pose heightened risks to aquatic ecosystems, increased potential for drinking water contamination, and accelerated infrastructure deterioration. Long-term trends indicate increased Cl⁻ loading to Lake Ontario, which is primarily linked to urbanization but also influenced by emerging point sources and groundwater inputs [7,13].

The western Lake Ontario basin, an area with both elevated groundwater and surface water Cl⁻ concentrations, continues to experience rapid population growth and urban development. Between 1996 and 2021, the Greater Golden Horseshoe population increased 57% from 6.5 to 10.2 million and is projected to grow another 45% by 2051 [78]. This intensifying urbanization has led to the widespread replacement of natural and agricultural landscapes with low-density urban sprawl [78], increasing the vulnerability of regional water resources [52].

Surface water Cl⁻ levels may be diluted during precipitation or snowmelt events [79,80,81], whereas Cl⁻ infiltrating into aquifers can accumulate over time, resulting in chronically elevated concentrations [82,83]. Variability in groundwater Cl⁻ concentrations may also reflect the geological formations sampled. Natural Cl⁻ sources such as geogenic saline aquifers or weathering of Cl⁻-bearing rocks have been documented in wells drilled into the Silurian Salina Formation and Devonian and Ordovician bedrock units in the Niagara Peninsula and parts of southern Ontario [84,85,86]. More detailed source characterization and geochemical analysis are needed to confirm these origins.

4.4. Influences of Landscape and Hydrology on Chloride Vulnerability

In addition to geological controls, land cover and hydrologic setting influence Cl⁻ dynamics. Areas with permeable soils, vegetated land cover, and gentle topography, conditions favorable for groundwater recharge, may enhance the vertical transport of Cl⁻ to aquifers if present in infiltrating water [20].

While road salt is a known major source of Cl⁻ in southern Ontario [2,11,17,61], a preliminary exploratory analysis of road proximity and Cl⁻ concentrations revealed a limited correlation. This suggests that proximity alone may not adequately reflect Cl⁻ loading patterns. Future work incorporating detailed municipal salt application data and winter maintenance practices is recommended to better understand spatial drivers of Cl⁻ concentrations.

Combined patterns of groundwater Cl⁻ concentrations and the BFI suggest that southeastern Ontario (e.g., Kingston area) and the western Lake Ontario shoreline (e.g., GTA) are particularly vulnerable to Cl⁻-rich groundwater discharge. These areas consistently exhibit moderate to high Cl⁻ concentrations alongside elevated BFI values, indicating a stronger reliance on groundwater contributions to streamflow. This overlap contributes to the highest vulnerability scores observed in the study, reinforcing the need for targeted management in regions where sustained baseflow may act as a long-term Cl⁻ source to surface waters.

The role of baseflow in Cl⁻ transport varies by hydrogeologic context. In the current study, an elevated BFI tends to co-occur with higher groundwater Cl⁻ concentrations, suggesting that baseflow may be a dominant pathway for Cl⁻ export to surface water in these areas. While the observed correlation between surface water Cl⁻ concentrations and the BFI x groundwater Cl⁻ composite was relatively weak, it was statistically significant and suggests groundwater plays a role in sustaining elevated stream Cl⁻ concentrations. This supports the idea that baseflow can act as a long-term delivery mechanism for legacy Cl⁻ stored in aquifers. Although runoff drives short-term winter peaks, baseflow likely maintains elevated stream Cl⁻ concentrations year-round, highlighting the need for integrated groundwater–surface water monitoring.

However, in some settings, particularly when groundwater residence times are long and flushing of the subsurface is limited, an elevated BFI could instead reflect Cl⁻ accumulation and attenuation, with slower transport to surface water. Thus, a high BFI may result in either persistent Cl⁻ loading or delayed delivery, depending on subsurface dynamics, including current concentrations and residence times. A conceptual model (Supplementary Material Figure S5) illustrates the potential roles of baseflow of similar magnitudes in Cl⁻ transport.

Nonetheless, elevated groundwater Cl⁻ inputs do not always translate into high surface water concentrations. For example, while the GTA has a high surface water Cl⁻ concentration, it ranks only moderately in vulnerability, likely reflecting a higher proportion of Cl⁻ delivery via surface runoff rather than baseflow. Conversely, southeastern Ontario may receive proportionally more Cl⁻ via groundwater discharge, with surface water concentrations potentially moderated by dilution (e.g., non-winter high flow events) or lower surface runoff inputs.

While a formal comparison of the BFI across different land use categories was not conducted, general trends suggest that higher BFI values tend to occur in more rural or forested settings, which often support greater infiltration and subsequent groundwater recharge. In contrast, lower BFI values are more common in urban catchments, likely reflecting increased impervious cover, which reduces infiltration and enhances surface runoff. Future work with finer-scale BFI estimates and detailed land use data could help clarify how land cover influences baseflow contributions and Cl⁻ transport.

Overall, spatial variability in Cl⁻ loading appears to reflect the relative influence of groundwater versus surface runoff. Surface runoff dominates in the GTA (heavily urbanized), whereas groundwater discharge may play a more prominent role in southeastern Ontario (less urban, more rural), especially in areas with elevated groundwater Cl⁻ concentrations and high BFI values. These findings highlight the importance of considering both pathways when assessing risks and developing management strategies for Cl⁻ loading.

4.5. Groundwater Vulnerability Assessment and Method Comparison

The SL-SG-LU method enables the identification of regional groundwater susceptibility where traditional hydrogeologic data sets are limited. Vulnerability classes, based on composite index scores derived from slope, surficial geology, and land use, reflect the combined effects of land use intensity, surficial materials conducive to infiltration, and slope. Urbanized areas with high-permeability soils, dense road networks, and slight gradients often show high SL-SG-LU scores, reinforcing prior findings that anthropogenic features increase the infiltration potential and susceptibility to surface-applied contaminants, including road salt [19,85].

Overlaying groundwater and surface water Cl⁻ concentrations on the SL-SG-LU vulnerability layer (Figure 10) reveals that groundwater sites exceeding the Canadian drinking water guideline (250 mg/L, represented by red triangles) and surface water sites exceeding 380 mg/L (orange to red circles) cluster predominantly within medium to high vulnerability zones, especially across the GTA, central east Ontario, and southwestern regions. This suggests a broad spatial relationship between land characteristics and elevated Cl⁻ levels. Despite the variability in sampling density, the overall spatial correspondence supports the utility of SL-SG-LU for predicting areas at an elevated Cl⁻ contamination risk. The higher data density in southwestern Ontario likely reflects targeted monitoring in known contamination hotspots, reinforcing the observed spatial alignment between vulnerability predictions and measured Cl⁻ concentrations.

While designed for groundwater vulnerability, surface water Cl⁻ patterns mirror SL-SG-LU spatial trends, likely due to shared landscape controls such as impervious cover, road density, and land use. However, surface water Cl⁻ dynamics are also shaped by seasonal and event-based conditions (e.g., snowmelt and runoff) not accounted for in static GIS layers, and so surface water patterns should be considered correlational rather than strictly causal.

Despite limitations, SL-SG-LU is a practical, scalable screening tool for broad regional assessments where hydrogeologic datasets are incomplete. Incorporating subsurface data can refine vulnerability mapping, as seen with DRASTIC-LU, which integrates water table depth, recharge, and aquifer media [20]. However, inconsistent availability of these data at provincial and watershed scales limited DRASTIC-LU application over this broad of a study area. Developing comprehensive, province-wide hydrogeologic data sets would significantly enhance the accuracy and consistency of future vulnerability assessments.

DRASTIC-LU produced more spatially heterogeneous vulnerability maps, capturing finer-scale variation. It classified a larger share of land as high vulnerability, while SL-SG-LU concentrated classifications in the medium range. These differences highlight the importance of hydrogeologic data in refining assessments, distinguishing low- and high-risk areas, and reducing the over- or underestimation inherent to surface-focused methods.

Nevertheless, SL-SG-LU remains valuable for jurisdictions lacking comprehensive subsurface data, effectively highlighting regions of concern and guiding monitoring or management efforts.

4.6. Sensitivity and Validation of Vulnerability Models

The map removal sensitivity analysis identified surficial geology as having the greatest impact on vulnerability estimates, followed by slope and land use. These results were consistent across the regional and watershed scales, emphasizing these parameters’ importance in groundwater vulnerability assessments.

Correlation analysis in the Credit River watershed showed a weak positive correlation between SL-SG-LU and groundwater Cl⁻ concentrations (r = 0.176, R² = 0.03) and a negative correlation for DRASTIC-LU (r = −0.222, R² = 0.05). These relationships were not statistically significant (p > 0.05), indicating limited predictive power for groundwater Cl⁻ concentrations. This likely reflects fundamental differences in what the indices represent, as the two methods are based on relatively static landscape and hydrogeological features (e.g., slope, surficial geology, and conductivity), while groundwater Cl⁻ concentrations are strongly influenced by dynamic, temporally variable factors such as road salt application rates and variability in recharge. Additionally, the point and interpolated Cl⁻ data may be affected by spatial smoothing and uneven monitoring coverage, which can bias correlations. For example, some areas classified as highly vulnerable (e.g., London and Sarnia; Figure 10a) exhibit low to moderate concentrations (Figure 10b,c), suggesting that the vulnerability indices are more suitable for as screening tools to identify potentially at-risk areas or regions, rather than precise predictors of current contaminations. This highlights the complexity of hydrogeological conditions, anthropogenic inputs, and data limitations in understanding the Cl⁻ distribution.

The validation highlights the importance of selecting or combining vulnerability assessment methods suited to the specific environmental and contamination context.

5. Conclusions

Elevated Cl⁻ concentrations in groundwater and surface water pose a growing challenge in cold regions, with no economically viable solution for large-scale removal once contamination occurs. Therefore, prevention through an improved understanding of Cl⁻ contributions, particularly from groundwater to surface water, is critical. This study identified and characterized hotspots of elevated Cl⁻ concentrations in southern Ontario, revealing consistent contamination in urban areas such as the GTA, as well as unexpected elevations in rural southwestern Ontario. These findings underscore the persistence of legacy contamination alongside ongoing inputs, highlighting the need for targeted management strategies.

Spatial analyses showed that land use, surficial geology, and slope are major factors influencing vulnerability to Cl⁻ contamination. The novel composite vulnerability index effectively identified areas of concern at the regional and watershed scales, with medium to very high vulnerability zones coinciding with most monitoring wells exceeding drinking water guidelines. However, a comparison with the more detailed DRASTIC-LU assessment in the Credit River watershed revealed discrepancies, underscoring the complexity of hydrogeological and anthropogenic factors and the need for tailored vulnerability assessments. While this study offers key insights, limitations related to monitoring density, data availability, and spatial interpolation highlight the need for ongoing data improvement and method refinement. Future work should consider refining weighting schemes and integrating continuous parameters to improve objectivity and accuracy.

In examining GW-SW interactions, it was revealed that areas with high BFIs, such as the GTA and southeastern Ontario, correspond closely with Cl⁻ hot spots in both groundwater and surface water (Figure 9). This spatial alignment suggests that regions with high groundwater discharge contribute notably to surface water Cl⁻ contamination via baseflow, reflecting legacy inputs from aquifers. These findings highlight the importance of baseflow dynamics in Cl⁻ transport and the need for long-term monitoring.

Strategic placement of monitoring wells within identified vulnerable zones is strongly recommended, not only in highly urbanized regions like the GTA but also in other susceptible areas such as the Golden Horseshoe and Kitchener–Waterloo. Given the observed spatial variability, deploying a denser network of wells across different aquifers will better capture localized Cl⁻ dynamics. Moreover, monitoring programs should incorporate regular, seasonal sampling, ideally at least quarterly, to capture temporal variations influenced by de-icing practices, with a particular focus on the late fall to spring period when Cl⁻ inputs peak.

This study also revealed challenges in accessing and using open-access water quality data in Ontario due to fragmentation, inconsistent reporting, and incomplete metadata. Establishing standardized, comprehensive data management and reporting protocols will enhance data reliability, support more robust analyses, and improve stakeholder confidence. Such improvements are essential for informed decision-making to mitigate Cl⁻ contamination risks.

Overall, the geospatial framework and vulnerability assessment presented here provide a transferable tool for prioritizing areas at risk of Cl⁻ contamination in cold climate regions. The framework presented here can be adapted for use in other cold climate regions facing similar salinization concerns, especially where hydrogeological data are limited. These insights can guide policy makers, watershed managers, and practitioners in developing more effective road salt application strategies and water resource protection measures. Future research should aim to integrate more detailed hydrogeological data, enhance method sensitivity, and expand monitoring efforts to better understand and manage Cl⁻ contamination in freshwater systems.

6. Recommendations

The findings of this study highlight several opportunities to improve Cl⁻ monitoring, management, and vulnerability assessments in cold region settings. Based on the observed patterns and procedural insights, the following are recommended:

• Refine and validate vulnerability indices

Future research should focus on refining the weighting schemes used in index-based methods like SL-SG-LU by incorporating continuous variables such as the depth to water and impervious surface area. Validating these indices against measured Cl⁻ concentrations and comparing them with alternative methods, like machine learning, will improve their reliability and applicability.

2.

Expand groundwater monitoring in vulnerable areas

Monitoring networks should be expanded in both urban and rural areas identified as vulnerable, with an emphasis on shallow wells (less than 20 m depth) that better represent impacts from Cl⁻ introduced via surface pathways. Expanding coverage will help capture localized patterns of contamination and provide a stronger basis for decision-making.

3.

Incorporate seasonal sampling to capture temporal variability

Given that Cl⁻ inputs peak during late fall to spring from road salt applications, monitoring programs should include regular seasonal sampling, at minimum quarterly but ideally more frequently, to accurately capture temporal changes and improve the understanding of patterns during various climatic periods.

4.

Standardize and centralize water quality data

Addressing challenges with fragmented and inconsistent water quality data in Ontario requires standardized reporting protocols and centralized, publicly available data repositories. This would improve data quality and accessibility, enabling more robust regional analyses.

5.

Investigate the influence of land use change on Cl⁻ dynamics

Future studies should leverage available land use datasets to analyze how transitions in land use, particularly from natural or agricultural to urban land uses, and increasing impervious surface cover affect spatial and temporal Cl⁻ patterns. Quantifying these relationships would support predictive land use planning and targeted management to mitigate Cl⁻ vulnerability in expanding urban areas.

6.

Advance research on groundwater–surface water interactions

Further investigation is needed to quantify groundwater’s role as a source of Cl⁻ to surface waters. Using a combination of field observations (e.g., tracer studies) and desk-based analyses (e.g., hydrographic separation) can help infer the relative contribution of legacy versus ongoing or recent Cl⁻ inputs and support integrated watershed management.