https://orcid.org/0009-0001-0933-2779
Figures
Abstract
Post-wildfire contamination poses a serious threat to drinking water in forested watersheds, with implications for long-term resilience of drinking water treatment plants. Infrastructure such as pumping equipment and transportation piping systems can also be exposed to wildfire-related impacts. Wildfires significantly alter water quality by increasing sediment loads, dissolved organic carbon, nutrients, and heavy metals. Fire suppression efforts further introduce contaminants, such as per- and polyfluoroalkyl substances, which persist in the environment and pose long-term risks to drinking water safety. This study presents a broadly applicable framework to assess the vulnerability of drinking water intake to the post-wildfire runoff and erosion response. It is designed for use across diverse geographic and environmental contexts, draws on readily available data, incorporates key post-wildfire runoff and erosion parameters, and accounts for climate change to address future contamination threats to water supplies. The framework calculates a vulnerability index incorporating information about forest cover, runoff potential, fire regime under climate change, and rainfall under climate change. The index is further refined accounting for drinking water intake exposure. Its design enables prioritization of sub-watersheds despite data availability while remaining adaptable to more refined datasets when available. The results indicate that forest cover and runoff potential are the dominant variables influencing final index scores, while rainfall projections under climate change amplify post-wildfire water contamination. Although wildfire threats driven by climate change continue to increase, drinking water managers often fail to integrate potential climate-driven hazards to surface water supply into their long-term adaptation strategies. Ensuring drinking water treatment plant resilience will require both adaptable assessment tools and scalable protection planning. In response, this framework supports informed decision-making to enforce targeted land use regulations, and develop emergency response strategies. These measures help mitigate post-wildfire impacts on drinking water intakes and support infrastructure adaptation accordingly.
Citation: Pouliot E, Bichai F, Kammoun R, Dorner S (2025) Source water protection and wildfire threats: A simplified vulnerability assessment framework for drinking water intakes. PLOS Water 5(1): e0000491. https://doi.org/10.1371/journal.pwat.0000491
Editor: Daniel Reddythota, Faculty of Water Supply & Environmental Engineering, ArbaMinch Water Technology Institute (AWTI), ETHIOPIA
Received: April 29, 2025; Accepted: December 9, 2025; Published: January 5, 2026
Copyright: © 2026 Pouliot et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are included in the paper or its Supporting information.
Funding: This work was supported by the NSERC OPSIDIAN CREATE grant (EP and SD), the City of Gatineau (SD), the Industrial Chair on Drinking Water with funding from NSERC and its municipal partners (SD) as well as Québec’s Ministère de l’Environnement, de la Lutte contre les changements climatiques, de la Faune et des Parcs (SD). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
High-severity wildfires significantly impact surface water quantity by disrupting a broad range of hydrological processes such as interception, infiltration, evaporation, and storage [1]. These modifications can induce rapid connections between disturbed landscapes and water bodies, resulting in higher annual water yields due to increased peak and low-water flows [2,3]. Vegetation loss can accelerate snowmelt due to increased solar radiation exposure, further altering runoff timing [3]. Furthermore, intense post-wildfire rainfall events compromise surface water quality by increasing runoff [4] and mobilizing ash, sediments, nutrients, chlorophyll-a, terrestrial debris, and contaminants—such as heavy metals— into water bodies and reservoirs [2,5–9]. This post-wildfire water quality impairment can persist for periods ranging from weeks to several years [10,11], even in large, low-relief, wetland-dominated watersheds [12]. Thus, the integrity of drinking water treatment is compromised and threatens the ability to produce safe and clean drinking water [2,6,9,13], resulting in an increase in chemical usage and sludge production to meet turbidity and disinfection by-product requirements [6,14]. Consequently, there is a growing need for adequate tools to assess post-wildfire water contamination and facilitate the mitigation of threats from post-wildfire on drinking water intake (DWI) vulnerability. Examples of recent record-breaking wildfires and their impacts on drinking water systems are provided in S1 Text.
Healthy forested watersheds play a vital role in safeguarding drinking water sources by acting as a natural filtering system, often providing high-quality water for downstream municipalities. This not only protects public health but also reduces drinking water treatment costs [15–17]. A study led by the Trust for Public Land in collaboration with the American Water Works Association showed that operating treatment costs decrease as forest cover in a source area increases, with approximately 50–55% of the cost variation attributed to the percentage of forest cover [17]. However, forested watersheds are becoming increasingly vulnerable to wildfires, which can exert substantial pressure on drinking water treatment plants (DWTPs) [6,18]. While low-severity wildfires can help preserve forest ecosystem health and sustain ecological services—such as wood production, biodiversity, and recreational activities [19]—they also help maintain complex and productive aquatic ecosystems [20]. In contrast, high-severity wildfires challenge water treatment processes and pose significant risks to public health, as extensively documented in the literature [2,6,7,21].
Identifying and evaluating the potential hazards to DWTPs from post-wildfire water contamination is essential. Such evaluations support more effective landscape management and help mitigate wildfire-related threats to ensure the protection and distribution of safe drinking water [22]. These evaluations also contribute to strengthening the long-term resilience of water supply systems—especially in communities that rely on forested watersheds as their primary source of drinking water. Few studies describe systematic approaches to assess the vulnerability of DWIs from post-wildfire impacts in a simple and rapid manner—without the need for complex modeling or machine learning—enabling stakeholders to engage in effective risk mitigation and emergency planning. In 2018, Robinne et al. [23] proposed a global-scale assessment of wildfire risks to drinking water security using a spatially explicit index. However, applying the index is challenging due to its reliance on over 30 variables, including conditions triggering post-wildfire effects (Driving forces), hydrogeomorphic changes (Pressure), hazard (State), impacts on water supply and ecosystem (Impact), and the effectiveness of mitigation strategies (Responses) [24]. While the global scale provides a broad overview, it may serve only as an initial step that leads to further investigation at finer scales for management purposes. In 2019, Robinne et al. [22] published a simple spatial index tailored for regional-scale applications to assess the exposure of forested watersheds to wildfire hazards in Alberta, Canada. This framework accounts for the post-wildfire water contamination hazard to municipal treatment systems by integrating both the distance between forested watersheds and DWIs, and the estimated water yield of each forested watershed as a proxy for its relative contribution to downstream drinking water supply. In 2023, Robichaud et al. [25] developed two complementary indices—one addressing wildfire hazard and the other evaluating water utility preparedness—to identify locations in the western United States where drinking water operations may be more at risk from wildfire threats. The assessment incorporates water utility preparedness as an evaluative criterion, which takes source diversity, source redundancy, treatment technology and diversion point into account. This added information can strengthen vulnerability assessments; however, the classification scores may not always reflect real-case scenarios. For instance, the assumption that reservoir withdrawal points are inherently safer than stream withdrawal points is contradicted by the case of Sydney (Australia) where the largest water supply reservoir experienced wildfire water contamination, necessitating an alternative water source [26]. Other cases of reservoir contamination following wildfires have been reported in other regions [13]. However, as these works primarily focus on wildfire hazards, this focus can downplay the influence of post-wildfire responses—such as post-wildfire rainfall events and snowmelt events—on surface water contamination. Incorporating hydrological and geomorphological responses in an assessment framework can improve management prioritization efforts at local and regional scales [8,12]. These factors can help identify locations where adverse effects on downstream DWIs are more likely. An organizational framework to assess post-wildfire processes within three quantifiable domains—fire regime, precipitation regime and hydro-geomorphic regime—was proposed in 2013 to improve conceptual and computer models of post-wildfire runoff and erosion processes [4]. Using these domains to develop a framework offers the capacity to account for post-wildfire erosion and runoff impacts, thereby enabling the prioritization of management risks for watersheds that most influence water contamination at DWIs—an aspect not addressed by the studies mentioned, or only partially addressed based on proximity in a previous study [22].
Climate change should also be accounted for within post-wildfire assessment frameworks to integrate shifts in fire behavior and accurately predict and manage future threats from wildfire on DWIs. While Robichaud et al. [25] incorporate climate change in fuel hazard, the other frameworks have yet to account for this evolving driver. However, climate change not only influences wildfire hazards [27,28] but also alters precipitation patterns, potentially affecting post-wildfire hydrologic and geomorphic processes [29,30]. Climate change projections indicate that wildfire-impacted watersheds will experience higher sediment and organic carbon yields due to intensified precipitation and increased surface runoff, as demonstrated in a recent study [29]. These modifications can exacerbate contaminant transport and amplify potential hazards to DWTP operations. Neglecting climate change may lead to an underestimation of post-wildfire water contamination magnitude, potentially compromising the long-term reliability of safe drinking water supply and increasing public health concerns [21]. Ensuring that mitigation strategies remain climate-adaptive can enable decision-makers to implement proactive measures that protect DWIs and reinforce DWTP long-term resilience against intensifying wildfire-climate dynamics. Current drinking water vulnerability assessments do not explicitly consider wildfire risk, leaving a gap in proactive source-water protection under changing fire regimes.
The aim of this study is to develop a comprehensive and practical framework that enables stakeholders to assess the vulnerability of DWIs to post-wildfire water contamination by identifying areas that generate higher potential impacts at DWIs. This research proposes an approach targeted by four key objectives: (1) develop a simple and relevant methodology for broad applicability and rapid assessment of the vulnerability of DWIs to wildfire threats, (2) include readily available data to facilitate implementation across diverse geographic and environmental contexts, (3) select parameters that directly influence the post-wildfire runoff and erosion response to ensure representativeness, and (4) integrate climate change to account for increasing threats from wildfires. The framework is structured around the three post-wildfire response domains described by Moody et al. [4] (i.e., fire regime, precipitation regime and hydro-geomorphic regime), while also incorporating forest cover and proximity to DWIs as additional variables to enhance comparative assessments between forested watersheds. By simplifying the complex interplay driving post-wildfire water contamination processes, the framework provides a structured approach for evaluating these threats to DWI vulnerability. It integrates two indices that consider key variables, including historical and projected fire regimes, historical and projected rainfall data, runoff potential, forest cover, and proximity of forested watersheds to DWIs. This design enables stakeholders to systematically identify high-potential-impact areas, supporting proactive wildfire risk reduction and drinking water protection strategies.
2. Materials and methods
2.1. Vulnerability assessment framework
The vulnerability assessment framework developed in this study builds upon the post-wildfire response organizational framework proposed by Moody et al. [4], grouping fire regime, precipitation regime, and hydro-geomorphic regime into three respective post-wildfire response domains aiming to better predict the post-wildfire runoff and erosion response. Each domain is represented by quantifiable metrics that most directly affect the post-wildfire runoff and erosion response, contributing to a robust and adaptable vulnerability assessment index. Although the methodology integrates elements from established approaches [4,22], it has been adapted to capture regional variations across diverse geographic and environmental contexts, while also accounting for the potential effects of climate change—thus enhancing its broad applicability and relevance. The methodology minimizes the need for extensive data processing while maximizing accessibility by using publicly available or easily obtainable datasets. This approach enables rapid vulnerability assessment for drinking water protection planning without relying on long-term hydrological data or complex modeling, which can be time-consuming. Thus, the framework provides a qualitative analysis based on realistic worst-case scenarios rather than a quantitative analysis. However, the framework does not consider fire risks at the wildland-urban interface, where fires can spread through non-forested watersheds and still threaten DWIs.
Geospatial data processing was conducted using the geographic software QGIS 3.34 to produce a set of five variables within the WGS84 coordinate system. The five key variables are structured within the three post-wildfire response domains defined by Moody et al. [4]: fire regime zone under climate change (fire regime), precipitation under climate change (rainfall regime), and runoff potential (hydro-geomorphic regime). Additionally, two complementary variables—forest cover and distance to DWIs—are incorporated to further refine the assessment (Fig 1). These variables were selected for their implication in post-wildfire hydrologic and geomorphic responses, potentially impacting DWI vulnerability through surface water contamination. At the same time, they remain rapid and simple to assess, ensuring a comprehensive and practical vulnerability assessment approach (see Sections 3.1.1 to 3.1.4). An overview of the five key variables is provided in Table 1 (see Table A in S1 Appendix for detailed data sources).
To effectively determine the potential impact of the post-wildfire runoff and erosion response on DWI vulnerability, several assumptions were made to ensure a structured and comprehensive evaluation. The framework considers wildfire severity ranging from moderate to extreme to capture a broad spectrum of landscape impacts, from vegetation loss to disturbed soil properties, which can significantly influence runoff and erosion response. It is assumed that entire sub-watersheds contribute to the post-wildfire runoff and erosion response during the entire post-wildfire rainfall events, representing a worst-case scenario to capture the most severe disruption in soil-hydraulic properties which can amplify contamination transport. Finally, to simplify contaminant transport dynamics, hydrological connectivity is assumed to be continuous and uniform, allowing for a more systematic assessment of downstream impacts on DWIs.
The first vulnerability assessment index considers each sub-watershed as a point source of contamination at its outlet. This index, denoted as WF1, is developed to evaluate the potential contribution of each sub-watershed to the post-wildfire runoff and erosion, serving as an indicator of its potential impact on DWI vulnerability. The WF1 index is calculated as shown in Equation (1):
where FC represents the percentage of forest cover within a sub-watershed (Section 3.1.1), FRV represents the fire regime vulnerability value, defined as the likelihood of a wildfire occurring in a forested watershed (Section 3.1.2), RP represents the runoff potential value (Section 3.1.3), and CC accounts for rainfall conditions under climate change during the fire season (Section 3.1.4).
A second index (WF2) is introduced to assess the exposure of downstream DWIs by incorporating the distance between the sub-watershed outlet and downstream DWIs. This index explicitly incorporates the dilution effect associated with increasing distance, reflecting the assumption that contaminants are progressively attenuated as they are transported through the river network toward the DWIs [31]. The WF2 index is calculated as shown in Equation (2):
where x is the in-stream length between the sub-watershed outlet and downstream DWIs, in kilometers. The exponential decay relationship, derived from a literature review of contaminant decay curves and expert consultations [31,32], provides a generalized representation of contaminant decay in rivers and streams. Their approach applies broadly to various drinking water contaminants, including sediment. This equation was previously applied in wildfire vulnerability assessments for surface water supplies [22], reinforcing its applicability for evaluating the exposure of downstream DWIs. WF1 identifies the post-wildfire contamination potential at the outlet of each sub-watershed; WF2 adjusts this based on spatial proximity to DWIs. WF2 does not introduce an independent vulnerability assessment index with a unique scale, but rather modulates the WF1 scores. This second index has the potential to aid managers in identifying priority areas where potential adaptation strategies may be the most needed to mitigate threats to DWTPs. This refinement enables more targeted prioritization of forest management actions and supports the implementation of potential adaptation strategies aimed at proactively protecting DWIs by mitigating threats from post-wildfire water contamination.
Five levels of potential impacts were used to describe the vulnerability of DWIs from the post-wildfire water contamination, as provided in Table 2: (1) very low potential impact: sub-watershed area is not vulnerable to wildfire; (2) low potential impact: treatment is at risk but the distribution of drinking water is not affected; (3) moderate potential impact: distribution of drinking water is still possible with the implementation of mitigation measures or if the contamination is brief; (4) high potential impact: distribution of drinking water over an extended period (weeks) is not sustainable, requiring significant mitigation efforts to maintain minimal supply levels; and (5) very high potential impact: distribution of drinking water over an extended period (weeks) is not sustainable, necessitating the urgent implementation of alternative water sources to ensure continued and safe water supply.
2.1.1. Forest cover.
Robichaud et al. [25] assign a score based on land use type, with forest interpreted as high fuel hazard. In contrast, Robinne et al. [22] directly integrate the proportion of forest cover, with higher values indicating a greater importance to drinking water supply. This framework expands on that perspective by recognizing forest cover as both a driver of wildfire behaviors and a contributor to drinking water quality protection, allowing for a more nuanced assessment. Based on this dual role, forest cover qualify the potential occurrence and magnitude of post-wildfire water contamination that may compromise downstream DWIs. Higher forest cover value indicating greater vulnerability, as highly-forested watersheds are more susceptible to post-wildfire water contamination threats. Moreover, DWTPs supplied by those watersheds rely on treatment processes optimized for naturally high-quality water, making them more vulnerable. Forest cover was quantified as the proportion of forested area within each selected sub-watershed. For more details on the data processing, see S2 Appendix.
2.1.2. Fire regime domain: Fire regime zone under climate change.
Various methods exist to delineate fire regime (Table A in S2 Appendix)—typically considering fire frequency, periodicity, intensity, size, landscape pattern, season, and burn depth [33,34]. A fire regime zone (FRZ) score under climate change is assigned to each selected sub-watershed, based on selected factors—herein, the area burned and the number of wildfires. The FRZ scores remains adaptable, either to the integration of additional factors or to refinement with more detailed datasets. However, it is essential that FRZ scores incorporate both historical and projected fire regime data to capture evolving wildfire dynamics. The FRZ score under climate change allows for a comparative assessment across sub-watersheds, giving greater weight to those located within zones with higher wildfire susceptibility. By incorporating both historical and projected fire regime data, this framework enhances the identification of sub-watersheds with the greatest potential impact on DWI vulnerability through runoff and erosion, enabling a more proactive assessment of future contamination threats. The fire regime projections in this study are based on RCP8.5, as it is considered the most realistic in the long term [35].
The selected factors are combined using a mixed aggregation method to assess the fire regime vulnerability of each sub-watershed [36]. For a more comprehensive understanding of this method and its advantages for water resource assessment, additional information can be found in Kammoun et al. [37]. The FRZ score under climate change is calculated as shown in Equation (3):
where wi represents the weight assigned to each selected factor (Ii), herein the area burned (I1) and the number of wildfires (I2). These two factors are considered complementary and of equal importance in delineating fire regime [33,34,38], and are thus each assigned a weight of 0.5. The sum of all weights must equal 1. Cadd and Cmax are weighting factors determined through iteration, set to 0.4 and 0.6, respectively. For necessary information required to apply the FRZ score under climate change, see S2 Appendix.
2.1.3. Hydro-geomorphic regime: Runoff potential.
To simplify this process and avoid complex modeling, the runoff potential score is based on the pre-wildfire condition of soils, and is determined by the drainage capacity and the slope percentage of selected sub-watersheds. In this framework, the sub-watershed’s drainage capacity refers to its ability to convey surface runoff to a common outlet, such as river or lake. Sub-watersheds with steeper slopes and lower drainage capacity exhibit greater runoff potential, thereby increasing the vulnerability of DWIs to post-wildfire water contamination through runoff and erosion. This framework considers surface runoff and erosion as the dominant mechanisms driving contaminant transport to surface water. Subsurface transport processes, such as shallow interflow or groundwater flow, are not considered in this assessment.
For each sub-watershed, a weight is assigned according to the classification shown in Table 3, based on drainage capacity and percentage of slope, following the approach outlined by Bolinder et al. [39]. The classification of drainage capacity is based on the Canadian National Soil Database (NSDB), which categorizes soils according to texture and water storage capacity. The scoring approach allows for a simplified evaluation of runoff potential and its potential impact on downstream DWI vulnerability. Within a given sub-watershed, drainage capacity and slope can vary spatially, leading to areas with different runoff potential scores. To reflect the realistic worst-case scenario, the score from the most extensive area is used in the vulnerability assessment.
2.1.4. Rainfall regime: Rainfall under climate change.
To determine rainfall depth, Intensity-Duration-Frequency curves should be used [4]. These curves described the relationship between rainfall intensity, the duration of a storm, and the statistical likelihood of that event occurring. However, since the projection of these curves under climate change is not available across regions, an integrated rainfall variable (CC) was developed. Climate change effects on rainfall are represented through relative change ratios calculated for two rainfall metrics—the total summer precipitation and the highest cumulative precipitation over five days—as shown in Equations 4 and 5. The first rainfall metric accounts for the seasonal shifts, while the second captures the short-term intensity relevant to first-flush contamination. Rainfall projection data were retrieved from CMIP6 climate simulations under a single high-emissions scenario (SSP3-7.0), to reflect a high-impact environmental context marked by altered rainfall regimes.
where refer to the 50th percentile (median) value of the rainfall metric from historical data and
refer to the 50th percentile value of the rainfall metric from projection data over the 2041–2070 horizon. These percentile values are calculated for each rainfall metric (Ps and P5) to assess relative changes in rainfall under climate change. Ps represents the relative changes in total summer precipitation (June to August). P5 represents the relative change of rainfall for the highest cumulative precipitation over any 5 consecutive days between April and September. These 50th percentile values are available on climate data products, and do not require users to perform calculations. The 50th percentile was selected to represent a typical yet substantial rainfall event while limiting the influence of extreme outliers. The source details for both metrics are provided in Table A in S1 Appendix.
2.2. Study area
In this study, three municipalities in Quebec, Canada, were selected to apply the vulnerability assessment framework (Fig 2). Quebec experienced an exceptionally active fire season in 2023, with 4.32 Mha burned—an area greater than the combined area burned over the previous two decades [40,41]. For confidentiality purposes, the three municipalities studied will be referred to as ‘Municipality 1’, ‘Municipality 2’, and ‘Municipality 3’. Municipality 1, located within the Southern Laurentians ecoregion, has four DWIs, while the other two municipalities, located within the Central Laurentians ecoregion, have two and one DWI, respectively. Site selection was based on differences in fire regimes, as each municipality exhibits distinct wildfire dynamics that influence both the extent of landscape exposure to wildfire and the vulnerability of DWIs to contamination.
The sources of mapped data are as follows: (1) Quebec border shape source: https://open.canada.ca/data/en/dataset/ef70dc3b-1069-4037-9bce-61f47e628a1d [42], available under an open license, (2) ecoregion border shape source: https://sis.agr.gc.ca/cansis/nsdb/ecostrat/gis_data.html [43], available under an open license, (3) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license.
2.2.1. Southern Laurentians ecoregion.
The first municipality (Municipality 1) lies within the Southern Laurentians ecoregion (Boreal Shield ecozone) [45], which is characterized by a low to mid-boreal ecoclimate with an average annual temperature of 1.5°C. Winters are long and cold, with an average temperature of -10°C, while summers are short, averaging 14.5°C. Annual precipitation varies greatly within the region, ranging from 800 to 1600 mm, depending on the location. Surficial deposits are mainly composed of till and fluvioglacial sediments of varying thicknesses. The area is dominated by mixed wood stands of white spruce (Picea glauca), paper birch (Betula papyrifera), balsam poplar (Populus balsamifera), and trembling aspen (Populus tremuloides). Black spruce (Pecea mariana), balsam fir (Abies balsamea), and tamarack (Larix laricina) are found on poorly drained sites. The mean annual area burned by wildfires is 0.06%. The principal land use activities are forestry, hunting and trapping, recreation, tourism, and some farming (<2% of the ecoregion). The Municipality 1 is divided into four sectors, each with its own DWI, three of these are supplied by a large river, while the fourth is supplied by one of its tributaries. This latter may be more vulnerable due to its location, as the contributing upstream area includes a larger proportion of forest cover compared to other sites.
2.2.2. Central Laurentians ecoregion.
The second and third municipalities (Municipality 2 and 3) lie within the Central Laurentians ecoregion of eastern Canada (Boreal Shield ecozone) [45], which is characterized by a high to mid-boreal ecoclimate with an average annual temperature of 0°C. Winters are cold, with an average temperature of -12.5°C, while summers are cool, averaging 12°C. Mean annual precipitation ranges from 800 to 1000 mm from north to south. Closed stands of black spruce (Picea mariana (Mill.) B.S.P.) and balsam fir (Abies balsamifera L. Mill.) are dominant along lower slopes, whereas upper slopes are dominated by more open stands of black spruce with some white spruce (Picea glauca (Moench) Voss) and paper birch (Betula papyrifera Marsh.), usually associated with lichens and feathermosses. Eastern white cedar (Thuja occidentalis L.) and black spruce are associated with wetlands. In the drier, northern parts of the region, white pine (Pinus strobus L.) and jack pine (Pinus banksiana Lamb.), along with spruce and balsam fir, are more common. Red pine (Pinus resinosa Ait.) occupies rocky outcrops and glacial fluvial sand deposits, but its presence is limited. The second municipality is supplied by two DWIs. In contrast, the third municipality relies on a single DWI. Both municipalities are located in densely forested landscape.
2.2.3. Protection zones.
In Quebec, provincial water regulations define three protection zones (immediate, intermediate, and outer) as buffer distances around DWIs where contaminants from anthropogenic activities or potential events may affect water quality or quantity. These zones encompass both surface water and the adjacent riparian lands along shorelines and upstream tributaries, and are used for land use restrictions and vulnerability assessments. The immediate and intermediate protection zones, as defined in the Water Withdrawal and Protection Regulation [46], are of interest for this framework to identify forested watersheds influencing DWI vulnerability. The immediate protection zone—located within 50 meters upstream and 500 meters downstream of the DWI, and within 10 meters of the riparian zones—is an area where the risk of contamination is highest due to the short transport time of contaminants and minimal dilution [46]. The intermediate protection zone—extending up to 10 km upstream and 50 meters downstream of the DWI, and within 120 meters of the riparian zones—is an area where the travel time to DWIs is too short to allow for intervention in the event of an accidental spill or a spike in contaminant concentration [46]. However, because protection zones are based on fixed distances around DWIs, they do not necessarily account for upstream watershed dynamics. As a result, diffuse pollution from sources beyond these zones (e.g., wildfire ash, sediments, nutrients) can still reach the DWIs. To address this limitation, this study delineates protection zones using an enhanced version of the approach proposed by Kammoun et al. [37], which details the regulatory method and sub-watershed selection criteria. This method follows the regulatory method while incorporating level 1 and 2 sub-watersheds whose outlets fall within the boundaries of the protection zones. when the entire drainage area contributing to the DWI is contained within a single large sub-watershed, this study improves management efficiency by integrating level 3 and 4 sub-watersheds.
The literature indicates that the impacts of post-wildfire runoff and erosion can be observed many kilometers downstream from the burned watershed [2,12]. Consequently, the approach was adapted to include an additional protection zone extending up to 30 km from the DWIs. This distance was selected based on findings from Smith et al. [2], who documented water quality degradation affecting downstream areas located over 30 km from the fire-impacted watershed. A 30 km buffer zone thus provides a conservative margin to account for the potential downstream transport of contaminants that can still pose a threat to DWTPs, potentially exceeding treatment design thresholds. These three zones (i.e., immediate, intermediate, and buffer) enable regional-scale application of the framework, guiding stakeholders to focus management efforts within this critical area to protect drinking water quality.
3. Results
The application of the vulnerability assessment framework to the three municipalities reveals marked spatial differences in the exposure of DWIs to post-wildfire runoff and erosion contamination, particularly under climate change influences and when accounting for proximity effects. In total, 40, 24, and 16 sub-watersheds were evaluated for Municipality 1, 2 and 3, respectively. WF1 was applied to estimate the potential contribution of each sub-watershed to the post-wildfire runoff and erosion, while WF2 explicitly incorporates the dilution effect associated with increasing distance. All the results can be found in S3 Appendix.
3.1. Vulnerability assessment
Municipality 1 has four DWIs, each surrounded by designated protection zones, resulting in four distinct zones (DWI-1A, DWI-1B, DWI-1C, and DWI-1D). The framework was applied independently to each DWI to evaluate their specific vulnerability to sub-watershed potential impact in the event of a wildfire. A first score is calculated for each selected sub-watershed, which excludes the rainfall regime variable (Fig A in S3 Appendix). By removing external influences on the post-wildfire runoff and erosion response (i.e., rainfall regime), this first score, hereafter referred to as intrinsic WF1, provides insight into the intrinsic potential impact of each sub-watershed from the post-wildfire runoff and erosion response. In other words, this is a vulnerability score based only on internal watershed characteristics, like slope, forest cover, and runoff potential. This intrinsic WF1 score serves as a reference point to understand how changing rainfall patterns under climate change can influence potential impacts on DWI vulnerability, as captured by WF1 and WF2 indices.
Across Municipality 1, intrinsic WF1 scores primarily indicate a ‘Moderate’ potential impact level on DWI vulnerability, with overall scores ranging from 1.5 to 24.4. Sub-watersheds 5 and 7 (Fig 3A and 3B), exhibit a ‘Low’ potential impact level (1.5 and 2.7) due to lower forest cover percentages (27% and 49%), which reduce their post-wildfire runoff and erosion potential—likely due to low fuel availability and potentially more fragmented landscape within the sub-watershed.
Zone C highlights high scores around DWI-1C, clustered within the intermediate and buffer zones, whereas Zones A, B, and D show no high-score. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) sub-watersheds data source: https://www.donneesquebec.ca/recherche/fr/dataset/bassins-hydrographiques-multi-echelles-du-quebec [47], available under open license.
By comparing these results with the complete WF1 scores (Fig 3), the analysis highlights the importance of incorporating the rainfall domain into the vulnerability assessment indices. A noticeable increase in potential impact levels is observed, particularly around DWI-1C. Four sub-watersheds (10, 11, 20, and 21), previously classified as having a ‘Moderate’ potential impact, are reclassified as ‘High’ potential impact (orange to light red, Fig A in S3 Appendix vs Fig 3) due to increased runoff and erosion potentially driven by higher rainfall under climate change. This comparison underscores how changing rainfall patterns under climate change can potentially influence water quality through the post-wildfire runoff and erosion response by potentially increasing the hydrological connectivity to surface water.
The WF2 index highlights the critical role of sub-watershed proximity to downstream DWIs in refining the vulnerability assessment (Fig 4). In general, sub-watersheds with outlets located closer to DWIs retain their initial potential impact level assigned by WF1 index due to minimal dilution capacity. In contrast, sub-watersheds farther away exhibit reduced potential impact levels, as seen for sub-watersheds 1, 4, and 6 around DWI-1B, where WF2 scores decrease by 15–25% compared to WF1 scores (Table A in S3 Appendix). Notably, a decrease is observed for sub-watersheds 10, and 11 around DWI-1C, shifting from ‘High’ to ‘Moderate’ due to distance from the DWI (light red to orange zones, Fig 3 vs Fig 4).
Zone C highlight high scores around DWI-1C, which are clustered within the buffer zone, whereas zones A, B, and D show no high-score. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) sub-watersheds data source: https://www.donneesquebec.ca/recherche/fr/dataset/bassins-hydrographiques-multi-echelles-du-quebec [47], available under an open license.
However, distance alone does not ensure a lower potential impact level. Sub-watersheds with extensive forest cover and high runoff potential may retain their same high classification level despite the increased distance from DWIs. This is particularly relevant around DWI-1D, where sub-watersheds with over 80% forest cover and steep slopes maintain a ‘High’ potential impact level even with dilution effects. Consequently, sub-watersheds that retain the same classification level from the WF1 index to the WF2 index should be prioritized for protective measures to mitigate potential contamination at DWIs in the event of a wildfire, as these areas have the highest potential impact on DWI vulnerability.
Municipality 2 has two DWIs within their respective designated protection zones referred to as DWI-2A and DWI-2B. The intrinsic WF1 scores (i.e., excluding rainfall regime variable; Fig B in S3 Appendix) for all sub-watersheds around both DWIs show considerable variability, driven by differences in key variables such as forest cover (17%-98%) and runoff potential (0.3-1.0) (Table B in S3 Appendix). All sub-watersheds belong to the same fire regime, ensuring that fire regime variable is not a factor in the score variation. Around DWI-2A, sub-watershed 3 exhibits the lowest intrinsic WF1 score (3.6), corresponding to a ‘Low’ intrinsic potential impact level, due to limited forest cover (17%) and moderate runoff potential (0.3). In comparison, sub-watersheds 2, 4, and 5 have higher intrinsic WF1 scores due to greater forest cover (50%, 84%, and 62%, respectively), classifying them in the ‘Moderate’ intrinsic potential impact level. This increase in scores and shift in levels reflect the role of forest density in amplifying the potential consequences of the post-wildfire runoff and erosion response on drinking water resources, as greater vegetation coverage can contribute to wildfire spread and also impact the natural filtration systems provided by the forests. In contrast, sub-watershed 8 has the highest intrinsic WF1 score of 70.5, indicating a ‘Very high’ intrinsic potential impact level, characterized by extensive forest cover (98%) and high runoff potential (1.0). Around DWI-2B, sub-watershed 19 has the lowest intrinsic WF1 score (9.0), corresponding to a ‘Low’ potential impact level. This classification is attributed to its moderate forest cover (41%), which reduces the availability of combustible material for wildfire to spread, and its moderate runoff potential (0.3), which limits the transportation of contaminants to drinking water resources via runoff and erosion. In contrast, sub-watershed 18 exhibits the highest intrinsic WF1 score (71.0), classified as ‘Very High’ potential impact level. This area is characterized by a nearly complete forest coverage (99%) and a high runoff potential (1.0), both of which contribute to an increased potential impact on DWI-2B vulnerability in the event of a wildfire.
The integration of the rainfall regime variable under climate change amplifies the WF1 index scores of all sub-watersheds by an average of 14% (Fig 5; Table B in S3 Appendix). This adjustment led to a shift in potential impact level for only two sub-watersheds around DWI-2A—sub-watersheds 7 and 12—which transitioned from a ‘High’ to a ‘Very high’ potential impact level, increasing WF1 scores (49.5 to 56.4 and 49.6 to 56.6, respectively).
Very high scores are observed across Zones A and B, clustered within immediate, intermediate, and buffer zones. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) sub-watersheds data source: https://www.donneesquebec.ca/recherche/fr/dataset/bassins-hydrographiques-multi-echelles-du-quebec [47], available under an open license.
Since the increase in WF1 scores results from applying a fixed multiplier—corresponding to the rainfall regime variable under climate change—the magnitude of change is directly proportional to the initial intrinsic WF1 scores. As a result, sub-watersheds with higher intrinsic WF1 scores (i.e., those already classified as ‘High’ or ‘Very high’) experienced the most substantial increases. For example, around DWI-2A, sub-watersheds 6, 8, 10, 13, and 14 all saw an increase of approximately 9.7 points, whereas sub-watersheds 2–5, classified as ‘Low’ or ‘Moderate’, increased by a maximum of 3 points (Table B in S3 Appendix). A similar trend is observed around DWI-2B, where sub-watersheds 18, 22, and 23 experienced significant increases of approximately 9 points, though they were already classified as ‘Very high’ potential impact level. This pattern confirms that sub-watersheds with the highest intrinsic WF1 scores are the most affected by the integration of the rainfall regime variable under climate change. This highlights the importance of prioritizing these areas for wildfire mitigation efforts, as their elevated runoff and erosion potential can increase the impact of post-wildfire water contamination at DWIs. The mobilization of sediments and nutrients may also negatively impact drinking water treatment efficiency by altering the quality of drinking water supplies.
As observed in Municipality 1, WF2 index scores in Municipality 2 show that sub-watersheds located closer to DWIs tend to retain their WF1 potential impact levels due to limited dilution capacity, while those farther away experience reductions (Fig 6). However, the extent of this reduction is not uniform and depends on both distance and landscape characteristics such as runoff potential and forest cover. Around DWI-2A (Fig 6A), sub-watersheds 8, 9, 13, and 14 exhibit the largest reductions applying the proximity function (between 24 and 26 points). These sub-watersheds initially had high WF1 scores, driven by extensive forest cover (88–98%) and high runoff potential, but their potential impact levels on DWI vulnerability were moderated by lower proximity, allowing for greater dilution. Similarly, around DWI-2B, sub-watersheds 23 and 24 had a significant decline (between 21 and 23 points), further reinforcing the dominant role of distance in reducing potential impact levels. However, sub-watersheds 18 (WF2: 72.8), maintain a ‘High’ potential impact level despite dilution effects. Its nearly complete forest cover (≥99%) and high runoff potential (1.0) continue to drive a strong post-wildfire runoff and erosion response, indicating that in some cases, source characteristics can outweigh the benefits of transport-related attenuation. These results emphasize the need to prioritize sub-watersheds where the combination of high runoff potential and extensive forest cover can sustain higher post-wildfire water contamination at DWIs. The persistence of high WF2 scores in sub-watersheds located within the immediate protection zone also highlights the need for targeted mitigation strategies, as shorter transport distances reduce the dilution capacity and the response time for intervention, thereby increasing the likelihood of post-wildfire water contamination reaching DWIs.
Very high and high scores are observed across Zones A and B, clustered within immediate, intermediate, and buffer zones. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) sub-watersheds data source: https://www.donneesquebec.ca/recherche/fr/dataset/bassins-hydrographiques-multi-echelles-du-quebec [47], available under an open license.
Municipality 3, which includes one DWI referred to as DWI-3A, exhibits a uniformly high potential impact profile, with intrinsic WF1 scores ranging from 26.5 to 30.0 across all sub-watersheds (Fig C in S3 Appendix). In contrast to Municipalities 1 and 2—where variability in forest cover and runoff potential contributed to a wider range of intrinsic WF1 scores, Municipality 3 is characterized by consistently high forest cover (88–100%; Table C in S3 Appendix). This extensive coverage contributes to providing high water quality by supporting natural filtration that benefits DWTPs. However, forest cover can also contribute to the spread of wildfires. This creates a management trade-off, where efforts to preserve high water quality must be balanced against the risk of wildfire-driven contamination. In this context, integrated approaches that account for both forest resilience and fire risk reduction are essential for protecting DWI-3A.
The integration of the rainfall regime variable under climate change (1.09) proportionally increases intrinsic WF1 scores across all sub-watersheds in Municipality 3 (Fig 7), yet the overall classification remains within the ‘High’ potential impact level. This outcome contrasts with Municipalities 1 and 2, where some sub-watersheds shifted into a higher potential impact level. These upward shifts were likely due to intrinsic WF1 scores already positioned near the upper threshold of their respective potential impact levels, making them more sensitive to even slight increases in hydrological connectivity—captured through the integration of the rainfall regime variable. In Municipality 3, however, intrinsic WF1 scores were clustered near the lower threshold of the ‘High’ potential impact level, limiting the potential for such transitions despite a similar increase (Municipality 1: 1.13, Municipality 2: 1.14, Municipality 3: 1.09). Even so, this pattern underscores that even modest changes in hydrological connectivity can potentially amplify the post-wildfire runoff and erosion response, potentially increasing DWI vulnerability to post-wildfire water contamination. It also highlights the importance of understanding where sub-watersheds lie within their classification spectrum, as this positioning influences their sensitivity to future climate-related changes.
High scores are observed across all selected sub-watershed. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) sub-watersheds data source: https://www.donneesquebec.ca/recherche/fr/dataset/bassins-hydrographiques-multi-echelles-du-quebec [47], available under an open license.
WF2 index scores remain within the same ‘High’ potential impact level as those established by WF1, even for sub-watersheds located further from DWI-3A, where dilution effects would typically lower contamination potential (Fig 8). These observations suggest that while WF2 index introduces spatial refinement through proximity-based adjustments, it does not shift classification levels in this municipality. Given this uniform impact level profile on DWI-3A vulnerability, prioritizing mitigation efforts may require going beyond classification impact levels—such as analyzing absolute WF2 scores or engaging with stakeholders to identify priority areas based on local conditions and management objectives. Additionally, further refinement of the framework through the integration of advanced landscape parameters (see Discussion) could support more targeted interventions to mitigate contamination risks to DWIs.
High scores are observed across all selected sub-watershed. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) sub-watersheds data source: https://www.donneesquebec.ca/recherche/fr/dataset/bassins-hydrographiques-multi-echelles-du-quebec [47], available under an open license.
3.2. Forest cover considerations
As illustrated in Fig 9, all three municipalities exhibit extensive forest cover, which plays a dual role in influencing both wildfire behavior and drinking water quality protection. The WF1 and WF2 indices capture how this interaction manifests: higher forest cover increases DWI vulnerability to post-wildfire water contamination due to greater fuel availability, while simultaneously supporting drinking water treatment by providing ecosystem services that enhance source water quality. This dual influence highlights the inherent trade-off between wildfire risk and water quality protection in forest catchments.
A) Municipality 1, B) Municipality 2, and C) Municipality 3. The sources of mapped data are as follows: (1) drinking water intakes and protection zones were acquired from municipal partners, (2) hydrology and watershed data source: https://open.canada.ca/data/en/dataset/a4b190fe-e090-4e6d-881e-b87956c07977 [44], available under an open license, and (3) forest data source: https://ouvert.canada.ca/data/fr/dataset/97126362-5a85-4fe0-9dc2-915464cfdbb7 [48], available under an open license.
The influence of forest cover on WF1 scores varies across municipalities, reflecting differences in landscape composition. Municipality 1 exhibits the highest heterogeneity in forest distribution (17%–98%), resulting in greater fluctuations in WF1 scores. Sub-watersheds with lower forest cover (e.g., sub-watershed 3 with 17%) correspond to lower WF1 scores, while those with extensive forest cover (e.g., sub-watershed 8 with 98%) exhibit higher scores. In contrast, Municipality 2, despite a more uniform forest cover (41%–99%), demonstrates similar trends: sub-watersheds combining high forest cover and high runoff potential show the highest potential impact on DWI vulnerability. However, sub-watersheds with high forest cover but lower runoff potential have lower WF1 scores, emphasizing that forested landscapes alone do not determine wildfire-driven water contamination potential. Municipality 3, characterized by consistently high forest cover (88%–100%) and high runoff potential, displays a narrower range of WF1 scores (26.5–30.0), all within the ‘High’ potential impact level. While forests provide fuel for wildfires, they also act as a natural filtration system that can help maintain source water quality under non-fire conditions. Consequently, wildfires in highly forested watersheds can amplify the extent of post-wildfire water contamination while simultaneously compromising the multiple-barrier benefits that forests provide, posing challenges for DWTP resilience to wildfire.
4. Discussion
4.1. Key drivers of the post-wildfire vulnerability of drinking water intakes
The application of the framework to three municipalities revealed key differences in DWI vulnerability to post-wildfire runoff and erosion, highlighting the influence of landscape heterogeneity in shaping post-wildfire water contamination risks. The narrower range of WF1 scores observed for Municipality 3 suggests that homogeneous landscapes may yield more predictable post-wildfire vulnerability patterns, which could simplify management strategies compared to municipalities with highly heterogeneous landscape. Beyond landscape structure, climate change effects on rainfall also play a defining role. Municipality 2 shows the highest sensitivity to increased rainfall, as reflected by its rainfall adjustment variable (1.14), which leads to the most significant shifts in potential impact level. This suggests that sub-watersheds in Municipality 2 are particularly susceptible to climate-induced increases in hydrological connectivity contributing to the post-wildfire runoff and erosion response. These results underscore the importance of accounting for both intrinsic landscape characteristics and climate-driven rainfall changes in vulnerability assessments. They also reinforce the need for region-specific mitigation strategies that reflect the geographic and environmental context of each municipality.
In addition to these factors, differences between historical and projected fire regimes under climate change further shape post-wildfire contamination risks. In this study, each municipality falls within a distinct fire regime zone, meaning that all selected sub-watersheds within a given municipality share the same fire regime score. While this uniformity eliminates intra-municipality variability in WF1 scores due to fire regime, its inclusion remains essential for inter-municipality differences in vulnerability. This is particularly evident when comparing sub-watersheds that share similar forest cover and runoff potential across municipalities but differ in their fire regime scores, revealing how fire regimes alone can modulate potential impact on DWI vulnerability. For instance, in Municipality 2 (fire regime score: 0.59), WF1 scores are systematically higher than those observed in Municipality 1 (fire regime score: 0.49) and Municipality 3 (fire regime score: 0.35), despite comparable landscape characteristics. This pattern reflects a greater projected increase in wildfire activity under climate change in Municipality 2, which in turn results in higher DWI vulnerability to post-wildfire runoff and erosion. These findings support the idea that—even when uniform across sub-watersheds—the fire regime score remains a critical component of the vulnerability assessment, as it modulates the overall intensity of the post-wildfire runoff and erosion response and can help define priority areas for mitigation under changing climate conditions.
This multi-site application shows that the framework not only can identify DWI vulnerability at the sub-watershed scale but also captures broader spatial trends, offering insight into how geographic and environmental contexts drive potential post-wildfire water contamination. Applying the framework across multiple ecoregions could help determine whether specific ecological zones are inherently more susceptible to potential wildfire-induced impacts on water quality at DWIs. Identifying such patterns could support broader deployment of mitigation strategies at the ecoregion level, ensuring that areas with higher exposure to wildfire-related water contamination are addressed beyond municipal boundaries. This approach can contribute to establishing an initial layer of protection for surface drinking water sources by identifying upstream threats at larger spatial scales. Municipalities can further implement targeted potential adaptation strategies at a finer scale [8], enhancing DWI resilience to post-wildfire water contamination and strengthening long-term drinking water quality protection. This framework may also hold broader international relevance for rapidly developing regions, where land-use change and wildfire risk may co-occur. In such contexts, proactive planning for drinking water protection becomes especially important.
4.2. Implications for forest management and drinking water protection
Although the results indicate that higher forest cover is associated with higher WF1 index scores, it is essential to avoid misinterpreting this as evidence that reducing forest cover would mitigate surface water contamination risks by limiting wildfire behavior. While dense forests do provide fuel for wildfires to spread, they are also a fundamental component of the multiple-barrier approach to drinking water protection. Forested landscapes act as a natural filtration system, reducing sediment, nutrients, and contaminants before they reach DWTPs [17]. The post-wildfire runoff and erosion response may increase contamination risks in forested watersheds, but maintaining forest integrity remains critical for sustaining long-term water quality and supporting DWTP resilience. This dual role of forests highlights the need for land management strategies that do not treat forests solely as a vulnerability factor, but rather as a valuable asset to be managed in ways that balance wildfire mitigation with drinking water protection. To achieve this balance, selective fuel management strategies—such as prescribed burns and strategic thinning—can help reduce wildfire intensity while preserving essential ecosystem services [49,50]. These strategies can be maximized by identifying and prioritizing watersheds for implementation at a larger scale. Additionally, targeted erosion control measures in sub-watersheds with both high forest cover and elevated runoff potential can limit post-wildfire sediment transport, thereby minimizing impacts on downstream DWIs. Reinforcing protective infrastructure and operational capacity at DWTPs (e.g., intake redesign, flexible offtake points, early warning systems, or securing alternative water sources) is also vital to ensure drinking water safety during and after wildfire events [51]. Effective coordination between drinking water managers and other involved agencies (e.g., environmental resource agencies, forest management agencies) is equally important, yet often occurs reactively in response to wildfire impacts rather than as part of proactive planning.
4.3. Strengths and novelty of the framework
The post-wildfire runoff and erosion indices developed in this study offer a practical scoring approach for assessing the vulnerability of DWIs to post-wildfire water contamination through runoff and erosion. The framework supports systematic sub-watershed prioritization using accessible data, and its streamlined design facilitates faster decision-making and efficient resource allocation. This make it particularly suitable for planning contexts and urban development where modeling capacity and data availability are limited. The framework can be integrated into both emergency response efforts and long-terms source water protection planning, which strengthens its practical relevance for municipalities. In the short term, it supports the development of targeted land use regulations to mitigate post-wildfire impacts on DWIs, adapt infrastructure, and guide urban development in the context of land use change. Over the long term, it helps stakeholders proactively safeguard DWI integrity and strengthen resilience of drinking water systems. Its dual utility in both emergency response and strategic planning is especially valuable as wildfire activity intensifies under climate change [18,52–54].
This approach complements existing tools by jointly integrating climate change and hydrological connectivity, an approach not combined in previous indices. It provides a climate-integrated, data-accessible tool for prioritizing sub-watersheds based on their potential to contribute to post-wildfire water contamination under current and future conditions, as future rainfall extremes may amplify erosion and sediment delivery to DWIs [29,30]. Incorporating climate change provides a temporal dimension that strengthens long-term water source protection planning, enabling municipalities to anticipate and prepare for climate-driven post-wildfire water contamination potential.
The framework relies on spatial datasets with inherent uncertainties, along with assumptions and simplifications that may influence the index score interpretation. The chosen high-emissions scenario (SSP3-7.0) is considered “unlikely” under current policies [55], but remains relevant for assessing how intensified rainfall could influence post-wildfire runoff and erosion response. Using this scenario is not intended to predict exact future climate conditions, but rather to explore how sub-watershed response may evolve under precipitation extremes. Forest cover data do not capture dynamic changes related to logging, regeneration, or disturbance. Similarly, drainage capacity and slope may overlook site-specific heterogeneity—particularly soil characteristics and land use types—that can influence the post-wildfire runoff and erosion response at finer spatial scales. Fire regime scores were also assigned uniformly within FRZ, limiting sensitivity to small-scale variations in topography or vegetation structure that can affect fire susceptibility. Nonetheless, the qualitative structure of the framework mitigates the influence of these uncertainties by supporting relative comparisons rather than predictions. Its design enables prioritization of sub-watersheds despite data variability while remaining adaptable to more refined datasets when available. The framework can be adapted to diverse geographic and environmental contexts by incorporating locally relevant variables, based on available data or expert advice. Its flexible structure of the framework facilitates interdisciplinary collaboration with experts in hydrology, ecology, and water quality, allowing further refinement of post-wildfire hydrologic and geomorphic components.
The framework is also consistent with existing drinking water protection regulations—such as Quebec’s Water Withdrawal and Protection Regulation (WWPR) [56]—and complements climate-resilient drinking water protection planning strategies, supporting broader national adaptation strategies. Ensuring DWTP resilience to growing wildfire threats will require both adaptable assessment tools and scalable protection planning.
The framework also demonstrates coherence between the intrinsic WF1 scores and the historical wildfire activity observed in the studied municipalities. In Municipality 1, the four zones experienced numerous wildfires from 1972 to 2024, mostly small events (<2 km2) with two larger wildfires (70 and 77 km2), resulting in a cumulative burned area of less than 1% of the total watershed areas (Table D and Fig D in S3 Appendix). This aligns with the predominance of ‘Low’ and ‘Moderate’ intrinsic WF1 potential impact levels. Municipality 2 experienced some large wildfires, with cumulative burned area reaching 22% in zone A and 16% in zone B. This historical wildfire activity is consistent with the predominance of ‘High’ and ‘Very high’ intrinsic WF1 potential impact levels. Although Municipality 3 experienced only one wildfire in 2023 (7.5km2; 1.4% of the basin), it lies within the boreal forest, in a zone dominated by coniferous species such as black spruce and jack pine. These forest types are known to sustain more intense and harder-to-control wildfires [57]. In this context, a ‘High’ intrinsic WF1 scores remains consistent with the wildfire behaviour characteristic of this region.
4.4. Future refinements and limitation
Future refinements could expand the set of environmental variables considered. Case studies across varying landscapes—categorized by landscape attributes (e.g., tree type, slope, and soil characteristics), and catchment type (e.g., lakes, rivers, and reservoirs)—could clarify which variables most strongly influence contamination risks. Such insights could then identify priorities for framework improvement while maintaining simplicity of application. Similarly, morphometric variables—such as watershed geometry, drainage networks, texture, and relief aspects—should also be considered in future refinements. These variables are widely used in resource management [58] because they influence runoff, peak discharge, and soil erodibility [59]. Their integration could enhance sub-watershed prioritizations by better distinguishing areas with higher susceptibility to erosion, thereby guiding upstream erosion control to protect DWIs. However, their calculation requires significant data processing, which limits their practicality for rapid assessments [58].
Road density was not included in this framework, as ignition risk and suppression capacity are outside its scope. However, it could complement broader vulnerability assessments that address these aspects. Roads may act as firebreaks by disrupting fuel continuity and facilitating firefighting operations, yet they can also increase wildfire ignition risks, particularly in fire regimes dominated by human-caused wildfires, where roads enable both accidental and intentional ignitions [60].
Beyond environmental drivers, this framework does not account for socio-economic dimensions. Unlike some indices that weigh population size in vulnerability assessments [22,25], this approach focuses on environmental features. As a result, it prioritizes sub-watersheds with the highest potential impact, regardless of population served by the DWTP.
While additional environmental and hydrological variables could refine sub-watersheds prioritization, their integration must be consciously balanced against the framework’s goal of accessibility and simplicity.
This framework does not quantify runoff volume and sediment loads, which are critical for determining contaminant mobilization and its impact on DWI vulnerability [2,61]. Consequently, the magnitude and duration of post-wildfire impacts on surface water quality—and their operational consequences for DWTPs—cannot be assessed quantitatively. It also does not track the fate and transport of specific harmful contaminants, limiting its capacity to guide treatment adaptation strategies. This is important, as contaminants pose distinct operational and regulatory challenges. For instance, post-wildfire fluctuations in dissolved organic carbon levels can influence disinfection by-products formation [12], whereas increased heavy metal levels are often more easily managed through most conventional and advanced treatment technologies [6]. Identifying contaminants of concern under site-specific conditions is therefore essential for guiding treatment responses.
The exclusion of hydrological modeling limits the capacity to capture the complexity of post-wildfire runoff and erosion processes. The framework is designed not to predict impacts quantitatively, but to provide a qualitative assessment of DWI vulnerability and support sub-watershed prioritization based on hydrogeomorphic factors. This focus makes it a practical tool for guiding upstream mitigation measures to reduce potential downstream impacts on DWIs.
The use of indicators and indices is common in environmental management, leading to the need for their validation. Such validation is challenging given that environmental indicators and indices frequently lack straightforward objectives, and their outcomes are instead inherently subjective [62]. According to Rykiel [63], validation, in its broad use, simply means showing that a model (or index) meets its performance criteria. In this study, the main objective was to develop a qualitative framework to rapidly assess the vulnerability of drinking water sources to wildfire threats under a changing climate using readily available data. The parameters included in the indices are based directly on these data sources and are intended to support the prioritization and targeting of sub-watersheds for management actions. These actions may include the water quality data collection to confirm the impact of occurring wildfire threats, or initiating discussions with DWTP operators in areas where the vulnerability is deemed high.
5. Conclusions
The proposed framework builds upon past frameworks that linked wildfire hazard to DWI vulnerability, while considering post-wildfire runoff and erosion responses. These responses directly govern contamination threats to water supplies, and their inclusion in the framework ensures connectivity between the burned watershed and downstream DWIs. As in past framework, climate change has been integrated into wildfire hazard. In addition, the framework incorporates projected changes in precipitation patterns that may alter post-wildfire hydrologic and geomorphic processes. It provides an integrated approach to assess the regional-scale vulnerability of DWIs to the post-wildfire runoff and erosion response by incorporating key environmental variables, while also accounting for dilution effect associated with increasing distance. Although this framework is simple and draws on readily available data, it does not quantify runoff volumes, erosion yields, or contaminant loads. Still, it reveals significant trends that can inform proactive wildfire risk management and source water protection strategies. The following conclusions can be drawn:
- A regional-scale vulnerability assessment conducted in three Quebec municipalities with distinct environmental characteristics identified forest cover and runoff potential as critical factors influencing post-wildfire water contamination.
- Incorporating rainfall data under climate change amplifies potential impact levels in some sub-watersheds and WF2 refined prioritization by highlighting the importance of proximity to DWIs.
- The framework is designed as a qualitative and accessible tool, so it does not quantify runoff volumes or specific contaminant loads, excludes detailed hydrological modeling, and focusses on environmental variables. This simplicity supports broad applicability while also pointing to opportunities for future refinement, such as integrating morphometric parameters.
- Despite its limitations, the simplicity of the framework and its reliance on readily available data make it a practical tool for an initial vulnerability assessment of DWIs to the post-wildfire runoff and erosion response, supporting land-use planning and source water protection by identifying upstream areas where wildfire mitigation could effectively reduce risks to drinking water systems.
- Future work could extend the framework to include fire risks at the wildland-urban interface, where fires can spread through non-forested areas and still threaten DWIs
- As wildfire activity intensifies under climate change, this framework can guide proactive, landscape-based approaches to protect public health and enhance DWTP resilience.
Supporting information
S1 Text. Recent record-breaking wildfire events.
https://doi.org/10.1371/journal.pwat.0000491.s001
(DOCX)
S1 Appendix. List of data sources and references.
https://doi.org/10.1371/journal.pwat.0000491.s002
(DOCX)
S2 Appendix. Variable definitions and data processing.
https://doi.org/10.1371/journal.pwat.0000491.s003
(DOCX)
Acknowledgments
We thank Quebec’s Ministry of the Environment, the partner municipalities, and the watershed organization team responsible for the study areas for contributing to the data collection and vulnerability assessments.
References
- 1. Poon PK, Kinoshita AM. Spatial and temporal evapotranspiration trends after wildfire in semi-arid landscapes. J Hydrol. 2018;559:71–83.
- 2. Smith HG, Sheridan GJ, Lane PNJ, Nyman P, Haydon S. Wildfire effects on water quality in forest catchments: A review with implications for water supply. J Hydrol. 2011;396(1–2):170–92.
- 3. Shakesby RA, Doerr SH. Wildfire as a hydrological and geomorphological agent. Earth Sci Rev. 2006;74(3–4):269–307.
- 4. Moody JA, Shakesby RA, Robichaud PR, Cannon SH, Martin DA. Current research issues related to post-wildfire runoff and erosion processes. Earth Sci Rev. 2013;122:10–37.
- 5. Abraham J, Dowling K, Florentine S. Risk of post-fire metal mobilization into surface water resources: A review. Sci Total Environ. 2017;599–600:1740–55. pmid:28535601
- 6. Emelko MB, Silins U, Bladon KD, Stone M. Implications of land disturbance on drinking water treatability in a changing climate: demonstrating the need for “source water supply and protection” strategies. Water Res. 2011;45(2):461–72. pmid:20951401
- 7. Hohner AK, Rhoades CC, Wilkerson P, Rosario-Ortiz FL. Wildfires Alter Forest Watersheds and Threaten Drinking Water Quality. Acc Chem Res. 2019;52(5):1234–44. pmid:31059225
- 8. Murphy SF, Writer JH, McCleskey RB, Martin DA. The role of precipitation type, intensity, and spatial distribution in source water quality after wildfire. Environ Res Lett. 2015;10(8):084007.
- 9. Writer JH, Hohner A, Oropeza J, Schmidt A, Cawley KM, Rosario‐Ortiz FL. Water treatment implications after the High Park Wildfire, Colorado. J AWWA. 2014;106(4).
- 10. Robinne F-N, Hallema DW, Bladon KD, Buttle JM. Wildfire impacts on hydrologic ecosystem services in North American high-latitude forests: A scoping review. J Hydrol. 2020;581:124360.
- 11. Rust AJ, Hogue TS, Saxe S, McCray J. Post-fire water-quality response in the western United States. Int J Wildland Fire. 2018;27(3):203–16.
- 12. Emmerton CA, Cooke CA, Hustins S, Silins U, Emelko MB, Lewis T, et al. Severe western Canadian wildfire affects water quality even at large basin scales. Water Res. 2020;183:116071. pmid:32717650
- 13. Robinne F-N, Hallema DW, Bladon KD, Flannigan MD, Boisramé G, Bréthaut CM, et al. Scientists’ warning on extreme wildfire risks to water supply. Hydrol Process. 2021;35(5):e14086. pmid:34248273
- 14. Hohner AK, Terry LG, Townsend EB, Summers RS, Rosario-Ortiz FL. Water treatment process evaluation of wildfire-affected sediment leachates. Environ Sci Water Res Technol. 2017;3(2):352–65.
- 15.
Dudley N, Stolton S. Running pure: the importance of forest protected areas to drinking water, a research report for the Worls Bank/WWF Alliance for Forest Conservation and Sustainable Use. Gland, Switzerland: World Bank; 2003.
- 16. McLauchlan KK, Higuera PE, Miesel J, Rogers BM, Schweitzer J, Shuman JK, et al. Fire as a fundamental ecological process: Research advances and frontiers. J Ecol. 2020;108(5):2047–69.
- 17. Ernst C, Gullick R, Nixon K. Conserving forests to protect water. Opflow. 2004;30(5):1–7.
- 18. Jones MW, Abatzoglou JT, Veraverbeke S, Andela N, Lasslop G, Forkel M, et al. Global and Regional Trends and Drivers of Fire Under Climate Change. Rev Geophys. 2022;60(3).
- 19. Sil A, Azevedo JC, Fernandes PM, Regos A, Vaz AS, Honrado JP. (Wild)fire is not an ecosystem service. Front Ecol Environ. 2019;17(8):429–30.
- 20. Bisson PA, Rieman BE, Luce C, Hessburg PF, Lee DC, Kershner JL, et al. Fire and aquatic ecosystems of the western USA: current knowledge and key questions. Forest Ecol Manag. 2003;178(1–2):213–29.
- 21. Bladon KD, Emelko MB, Silins U, Stone M. Wildfire and the future of water supply. Environ Sci Technol. 2014;48(16):8936–43. pmid:25007310
- 22. Robinne FN, Bladon KD, Silins U, Emelko MB, Flannigan MD, Parisien MA, et al. A Regional-Scale Index for Assessing the Exposure of Drinking-Water Sources to Wildfires. Forests. 2019;10(5).
- 23. Robinne F-N, Bladon KD, Miller C, Parisien M-A, Mathieu J, Flannigan MD. A spatial evaluation of global wildfire-water risks to human and natural systems. Sci Total Environ. 2018;610–611:1193–206. pmid:28851140
- 24. Eea E. Environmental indicators: typology and overview. Europe Environ. 1999.
- 25. Robichaud PJL, Padowski JC. Drinking water under fire: Water utilities’ vulnerability to wildfires in the Pacific Northwest. J Am Water Resour Assoc. 2023;60(2):590–602.
- 26. Neris J, Santin C, Lew R, Robichaud PR, Elliot WJ, Lewis SA, et al. Designing tools to predict and mitigate impacts on water quality following the Australian 2019/2020 wildfires: Insights from Sydney’s largest water supply catchment. Integr Environ Assess Manag. 2021;17(6):1151–61. pmid:33751793
- 27. Flannigan MD, Krawchuk MA, de Groot WJ, Wotton BM, Gowman LM. Implications of changing climate for global wildland fire. Int J Wildland Fire. 2009;18(5):483–507.
- 28. Westerling AL, Hidalgo HG, Cayan DR, Swetnam TW. Warming and earlier spring increase western U.S. forest wildfire activity. Science. 2006;313(5789):940–3. pmid:16825536
- 29. Loiselle D, Du X, Alessi DS, Bladon KD, Faramarzi M. Projecting impacts of wildfire and climate change on streamflow, sediment, and organic carbon yields in a forested watershed. J Hydrol. 2020;590:125403.
- 30. Moody JA, Martin DA. Synthesis of sediment yields after wildland fire in different rainfall regimes in the western United States. Int J Wildland Fire. 2009;18(1):96–115.
- 31.
Weidner E, Todd A. From the Forest to the Faucet: Drinking Water and Forests in the US. Method papers. Washington, DC, USA: USDA Forest Service; 2011.
- 32.
Mack E, Lilja R, Claggett S, Sun G, Caldwell P. Forests to Faucets 2.0: connecting forests, water, and communities. Washington, DC: U.S. Department of Agriculture, Forest Service, Washington Office; 2022. Report No.: Gen. Tech. Rep. WO-99.
- 33. Hanes CC, Wang X, Jain P, Parisien M-A, Little JM, Flannigan MD. Fire-regime changes in Canada over the last half century. Can J For Res. 2019;49(3):256–69.
- 34. Whitman E, Batllori E, Parisien M, Miller C, Coop JD, Krawchuk MA, et al. The climate space of fire regimes in north‐western North America. J Biogeograph. 2015;42(9):1736–49.
- 35. Huard D, Fyke J, Capellán‐Pérez I, Matthews HD, Partanen A. Estimating the Likelihood of GHG Concentration Scenarios From Probabilistic Integrated Assessment Model Simulations. Earth Future. 2022;10(10).
- 36. Langhans SD, Reichert P, Schuwirth N. The method matters: A guide for indicator aggregation in ecological assessments. Ecol Ind. 2014;45:494–507.
- 37. Kammoun R, McQuaid N, Lessard V, Goitom EA, Prévost M, Bichai F, et al. Risk assessment of drinking water intake contamination from agricultural activities using a Bayesian network. PLOS Water. 2023;2(7):e0000073.
- 38. Erni S, Wang X, Taylor S, Boulanger Y, Swystun T, Flannigan M, et al. Developing a two-level fire regime zonation system for Canada. Can J For Res. 2020;50(3):259–73.
- 39. Bolinder MA, Simard RR, Beauchemin S, MacDonald KB. Indicator of risk of water contamination by P for Soil Landscape of Canada polygons. Can J Soil Sci. 2000;80(1):153–63.
- 40.
CIFFC. Canada report: 2023 fire season. 2023.
- 41. Jain P, Barber QE, Taylor SW, Whitman E, Castellanos Acuna D, Boulanger Y, et al. Drivers and Impacts of the Record-Breaking 2023 Wildfire Season in Canada. Nat Commun. 2024;15(1):6764. pmid:39164286
- 42.
Statistics Canada. 2021 Census - Boundary files; 2024 [cited 2024 Sep 4]. Database: Government of Canada Open Data Portal [Internet]. Available from: https://open.canada.ca/data/en/dataset/ef70dc3b-1069-4037-9bce-61f47e628a1d
- 43.
Environment Canada & Agriculture and Agri-Food Canada. The National Ecological Framework for Canada; 1999 [cited 2024 Sep 4]. Database: National Soil Database (NSDB) [Internet]. Available from: https://sis.agr.gc.ca/cansis/nsdb/ecostrat/gis_data.html
- 44.
Natural Resources Canada. National Hydro Network (NHN); 2024 [cited 2024 Sep 4]. Database: Government of Canada Open Data Portal [Internet]. Available from: https://osdp-psdo.canada.ca/dp/en/search/metadata/NRCAN-FGP-1-a4b190fe-e090-4e6d-881e-b87956c07977
- 45.
Ecological Stratification Working Group. A National Ecological Framework for Canada. Ottawa/Hull: Agriculture and Agri-Food Canada, Research Branch, Centre for Land and Biological Resources Research and Environment Canada, State of the Environment Directorate, Ecozone Analysis Branch; 1995.
- 46.
Water Withdrawal and Protection Regulation- Q-2, r. 35.2. 2014.
- 47.
Ministère du Développement durable, de l`Environnement et de la Lutte contre les changements climatiques (MDELCC). Bassins hydrographiques multiéchelles du Québec; 2018 [cited 2024 Sep 4]. Database: Données ouvertes du Québec [Internet]. Available from: https://www.donneesquebec.ca/recherche/dataset/bassins-hydrographiques-multi-echelles-du-quebec/resource/183c65a7-acb7-4e51-b3c5-5d34e5014771
- 48.
Gouvernement du Canada. Couverture du sol du Canada, circa 2000 (vectoriel) - Série GéoBase, 1996-2005; 2005 [cited 2024 Sep 4]. Database: Government of Canada Open Data Portal [Internet]. Available from: https://ouvert.canada.ca/data/fr/dataset/97126362-5a85-4fe0-9dc2-915464cfdbb7
- 49. Fernandes PM, Botelho HS. A review of prescribed burning effectiveness in fire hazard reduction. Int J Wildland Fire. 2003;12(2):117–28.
- 50. Brodie EG, Knapp EE, Brooks WR, Drury SA, Ritchie MW. Forest thinning and prescribed burning treatments reduce wildfire severity and buffer the impacts of severe fire weather. Fire Ecol. 2024;20(1).
- 51. Shah NW, Baillie BR, Bishop K, Ferraz S, Högbom L, Nettles J. The effects of forest management on water quality. For Ecolo Manag. 2022;522:120397.
- 52. Abatzoglou JT, Williams AP. Impact of anthropogenic climate change on wildfire across western US forests. Proc Natl Acad Sci U S A. 2016;113(42):11770–5. pmid:27791053
- 53. Flannigan M, Cantin AS, de Groot WJ, Wotton M, Newbery A, Gowman LM. Global wildland fire season severity in the 21st century. For Ecol Manag. 2013;294:54–61.
- 54. Wasserman TN, Mueller SE. Climate influences on future fire severity: a synthesis of climate-fire interactions and impacts on fire regimes, high-severity fire, and forests in the western United States. Fire Ecol. 2023;19(1).
- 55. Hausfather Z, Peters GP. Emissions - the “business as usual” story is misleading. Nature. 2020;577(7792):618–20. pmid:31996825
- 56.
MELCCFP. The Water Withdrawal and Protection Regulation (Regulation amending the WWPR). 2025. Available from: https://www.environnement.gouv.qc.ca/eau/prelevements/reglement-prelevement-protection/index-en.htm#:~:text=The%20Water%20Withdrawal%20and%20Protection,(Regulation%20amending%20the%20WWPR
- 57. Rowe JS, Scotter GW. Fire in the Boreal Forest. Quat Res. 1973;3(3):444–64.
- 58. Singh MC, Satpute S, Prasad V. Remote sensing and GIS-based watershed prioritization for land and water conservation planning and management. Water Sci Technol. 2023;88(1):233–65. pmid:37452545
- 59. Tripathi MP, Panda RK, Raghuwanshi NS. Development of effective management plan for critical subwatersheds using SWAT model. Hydrol Process. 2005;19(3):809–26.
- 60.
Morrison PH. Roads and Wildfires. Winthrop, Washington: Pacific Biodiversity Institute; 2007.
- 61. Langhans C, Smith HG, Chong DMO, Nyman P, Lane PNJ, Sheridan GJ. A model for assessing water quality risk in catchments prone to wildfire. J Hydrol. 2016;534:407–26.
- 62. Moriarty PE, Hodgson EE, Froehlich HE, Hennessey SM, Marshall KN, Oken KL, et al. The need for validation of ecological indices. Ecol Ind. 2018;84:546–52.
- 63. Rykiel EJ Jr. Testing ecological models: the meaning of validation. Ecol Model. 1996;90(3):229–44.