After wildfires, the loss of the humus layer leads to increased runoff and pollutants entering rivers. This study examined the long-term effects of wildfires on water quality. We statistically analyzed the changes in the water quality of streams surrounding the wildfire area. We used eight water quality parameters provided by the National Institute of Environmental Research for the analysis. To assess the impact of the wildfires, we employed t-tests and point-biserial correlation analysis to compare the changes in water quality indicators before and after the wildfires. Additionally, an analysis of variance was conducted to evaluate the impact of three wildfires, each occurring in different periods, on the water quality in a single river basin. The results showed increasing trends in hydrogen ion concentration (pH), electrical conductivity, and dissolved oxygen after the wildfire, whereas biochemical oxygen demand, total phosphorus, and total nitrogen exhibited decreasing trends. The impact of wildfires on changes in suspended solids was relatively minimal. It is expected that the results of this study provide valuable insights into developing water quality management and restoration plans following wildfires.

  • Wildfires have high effects on water quality.

  • Three statistical methods are used to investigate the impact of wildfires on water quality variability.

  • Post-wildfire influences the increase of pH, EC, and DO and the decrease of COD, BOD, TN, and TP.

Despite the advancement of numerous strategies aimed at mitigating wildfire damage, the scale of wildfire damage is increasing due to climate change (Flannigan et al. 2016; Abatzoglou et al. 2021). According to the Wildfire Statistical Yearbook (2022), the number of wildfires and damaged areas in South Korea has increased significantly since the 2010s. A characteristic of wildfires in South Korea is their tendency to occur frequently during the spring and winter months, with many of them happening between December and May. Around 60% of all wildfires take place in spring, and 26% are reported in winter. The effects of wildfires are not limited to bare land but also affect the hydrological and water quality of rivers (Cha & Shim 2015).

Fallen leaves and humus layers on the forest floor temporarily store rainfall and reduce surface runoff, preventing sudden flooding and soil erosion. When vegetation is burned due to a wildfire, the soil is heated, and organic matter is oxidized to fill voids, forming an impermeable layer on the soil surface. This reduces infiltration capacity, leading to increased runoff, which can cause secondary damage, such as flash floods and debris flows during monsoon season (Lee 2022). Although various studies have been conducted on the hydrological aspects of wildfire occurrence in South Korea, relatively few have examined the impact of wildfires on river water quality changes (Lee 2006; Lee et al. 2022).

In terms of water quality, transformations are observed in key indicators, such as total nitrogen (TN), total phosphorus (TP), and suspended solids (SSs), owing to the influence of organic substances, nutrients, and metals leached from wildfire-affected areas into rivers (Bitner et al. 2001). In wildfire-affected areas, the pH of the soil increases as vegetation undergoes combustion (Agbeshie et al. 2022), leading to soil exposure and subsequent eutrophication as phosphorus and nitrogen from the soil are washed into rivers by rainfall, resulting in additional damage (Emmerton et al. 2020; De Palma-Dow et al. 2022). The impact of wildfires is increasing in Korea, where forests cover 63.7% of the land (Korea Forest Service 2020), highlighting the significance of monitoring changes in river water quality due to wildfire effects.

In South Korea, water quality monitoring and regulation are implemented through a comprehensive system led by the Ministry of Environment and the National Institute of Environmental Research (NIER). The Automatic Water Quality Monitoring Network enables real-time tracking of key indicators such as hydrogen ion concentration (pH), TP, TN, dissolved oxygen (DO), biochemical oxygen demand (BOD), chemical oxygen demand (COD), SS, and electrical conductivity (EC), all of which are critical for assessing the impact of environmental events, including wildfires, on water bodies (Ministry of Environment 2024). This system plays a crucial role in identifying trends in water quality changes, which is essential for understanding the long-term impacts of wildfires on water bodies. The framework also includes controls on pollution, which are for managing water quality fluctuations caused by post-wildfire events. This monitoring system provides the foundation for evaluating how wildfires may influence water quality trends over time, supporting both immediate and long-term management strategies.

In South Korea, although monitoring and modeling methods have been developed to evaluate changes in various water quality factors, such as pH, TN, TP, and EC, before and after wildfires, there is still a lack of studies analyzing the specific effects of wildfires on river water quality (Jeong et al. 2004; Emmerton et al. 2020; Hampton et al. 2022; Paul et al. 2022). For example, Emelko et al. (2016) confirmed that phosphorus directly affects the occurrence of eutrophication, algae, and blue-green algae as secondary damage to wildfires. After 6–7 years in areas affected by wildfires, the phosphorus concentration in the surrounding rivers increased by monitoring the phosphorus in the micro-deposits flowing into the rivers by rainfall. Additionally, Hohner et al. (2019) examined the impact of the size and severity of wildfires on water quality by observing changes in TN, TP, turbidity, and total organic carbon. The results indicated a difference of approximately seven times in the concentration of nitrate-form nitrogen depending on the severity. They also confirmed that the increased concentration of organic carbon following a wildfire affected water quality in the long term (Hohner et al. 2019). In addition, De Palma-Dow et al. (2022) analyzed various water quality factors, including TP, DO, and water temperature, through 50-year data monitoring to evaluate the impact on the lake according to the scale and severity of wildfire damage. In this study, we confirmed that in most lakes, there was an increase in cyanobacteria owing to excessive phosphorus in the form of phosphates leached from areas damaged by wildfires.

Although research on the impact of wildfires on the water quality of nearby rivers is being actively pursued globally, there is a lack of such research in South Korea. Therefore, this study aimed to statistically examine the impact of wildfires on water quality by comparing data before and after wildfires, focusing on eight major water quality indicators: pH, TP, TN, DO, BOD, COD, SS, and EC.

Study area

South Korea, the study area, is geographically located between 38° 36′ 49″ N and 33° 06′ 40″ N latitude, and between 124° 36′ 35″ E and 131° 52′ 22″ E longitude. Approximately 63% of the total land area (62,980 km2) is forested (Figure 1). On average, approximately 460 wildfires occur annually, causing approximately 21 km2 of damage, with an average annual rainfall of 1,306 mm, of which approximately 54% or 710.9 mm is concentrated between July and August. According to the Wildfire Statistical Yearbook (2022), wildfires frequently occurred between December and May of the following year. In this study, an additional analysis was conducted focusing on the Gangwon-do area. Gangwon-do is a suitable region for observing the effects of wildfires on the water quality of rivers from various perspectives, as rivers flowing directly into the sea and others flowing inland coexist owing to their geographical characteristics.
Figure 1

Study area of this study. The red heat map indicates the locations of wildfire and the blue point denotes the water quality observatory used in this study.

Figure 1

Study area of this study. The red heat map indicates the locations of wildfire and the blue point denotes the water quality observatory used in this study.

Close modal

Datasets

Wildfire information

This study constructed a dataset based on the wildfire occurrence statistics provided by the Korea Forest Service (https://www.forest.go.kr). For dataset construction, data from 14,300 wildfire incidents recorded over 42 years from 1978 to 2019, including the date and time of occurrence, location, and affected area, were used. Among the datasets, 128 instances of large wildfires with an affected area of over 0.3 km2 were selected. Ultimately, data from 27 wildfire incidents were selected based on the availability of water quality monitoring station data covering the entire study period. The addresses of the wildfire sites were converted into coordinates, visualized using Quantum Geographic Information System (QGIS), and utilized in the study.

Water quality datasets

To construct the water quality dataset, the monthly average results from the water quality monitoring network provided by the Ministry of Environment (https://www.me.go.kr) were used. Water quality observational data from January 1989 to August 2023 were used in this dataset. To examine the impact of wildfires on river water quality, water quality monitoring stations encompassing areas affected by wildfires were selected, and a dataset spanning 15 years centered on wildfire occurrence dates was constructed. We selected 59 water quality monitoring stations that met the aforementioned conditions and configured 76 datasets. The inclusion criteria for the water quality monitoring stations were: (1) proximity to wildfire-affected areas, (2) availability of continuous water quality data for the entire study period, and (3) coverage of both upstream and downstream sections of the rivers within the wildfire impacted basins. Stations that did not meet these criteria, such as those with incomplete data or located far from the wildfire-affected regions, were excluded. The water quality datasets used in this study included physicochemical indicators, such as pH, DO, EC, and SS, and organic pollutant indicators, such as BOD, COD, TN, and TP. This study was conducted using datasets from water quality monitoring stations located upstream and downstream of the target rivers, including those within basins and boundaries encompassing wildfire-damaged areas. The datasets were categorized based on their expected degree of impact from the wildfires: 36 datasets from areas expected to be directly affected, including the wildfire-damaged locations; 34 datasets from points expected to be less affected because of their distance from the wildfire-damaged areas; and six datasets from locations anticipated to have minimal or no impact.

Statistical analysis

The impact of wildfires on river water quality was statistically analyzed using the SciPy library in Python. Welch's t-tests and point-biserial correlation analyses were performed to compare the differences between the indicators before and after the wildfires. The dataset used for the analysis was based on the assumption that normality is achieved under the Central Limit Theorem, given a sufficiently large sample size. Additionally, Welch's t-test was applied because the assumption of homogeneity of variance between river segments might not have been satisfied. This ensured reliable results, even when the variances between the two groups were different. Additionally, to examine the correlation and differences between the upstream and downstream sections of the rivers affected by wildfires, correlation analysis and analysis of variance (ANOVA) were conducted. For the ANOVA, a post hoc test was performed using Tukey's honest significant difference (HSD) test. Based on the 95% confidence intervals of each index, we determined whether the wildfires had a lasting impact on river water quality (Wagenbrenner et al. 2021).

t-test

A t-test was used to compare the average difference between the two groups, and changes in river water quality were analyzed by comparing the water quality data before and after the wildfire.

The basic formula for a t-test is as follows:
(1)
where , represent the means of each group; , represent the sample variances of each group; and , represent the sample sizes of the groups (Equation (1); Welch (1947)).

Point-biserial correlation analysis

A point-biserial correlation analysis was used to analyze the correlation between continuous and binary variables. Through this analysis, the relationship between the occurrence of wildfires (binary variable) and water quality indicators (continuous variables) was examined to identify significant changes in water quality indicators after wildfires.

Point-biserial correlation coefficient () is calculated as follows:
(2)
where , represent the means of the continuous variables when the dichotomous variables are one and zero, respectively; represents the standard deviation of all metric observations; , are the numbers of observations where the dichotomous variables are 1 and 0, respectively; and N is the total number of observations (Equation (2); Kornbrot (2014)).

ANOVA

ANOVA is a method used to determine the statistical significance of differences among three or more groups. In this study, ANOVA was used to analyze the water quality data. We focused on monitoring stations located upstream and downstream of wildfire locations and compared the data before and after the occurrence of wildfires. This study verified whether there were statistically significant changes in river water quality at upstream and downstream points before and after the wildfires. This approach allowed us to confirm the specific impact of wildfires on river water quality by identifying statistically significant changes.

The basic formula for ANOVA is as follows:
(3)
where (intergroup mean square) represents the variance between groups, and (within-group mean square) represents the variance within groups (Equation (3) McDonald model (2009)).

In this study, after conducting ANOVA, Tukey HSD, a type of post hoc test, was used. Tukey's HSD is particularly useful when the number of groups is large because it effectively controls the total number of type I errors that can occur when comparing each pair of groups individually (Nanda et al. 2021).

Variability of water quality after wildfire

Statistical analysis of the eight water quality indicators after the wildfires showed increases in pH, EC, and DO, whereas TN, TP, COD, BOD, and SS tended to decrease (Table 1 and Figure 2). Based on these results and the characteristics of the water quality indicators, they were categorized. The categories included physicochemical indicators such as pH, EC, and DO; indicators of organic pollution such as COD and BOD; and nutrient indicators such as TN, TP, and SS.
Table 1

Mean, maximum, minimum, and standard deviation by indicator before and after wildfires

IndexBefore wildfire
After wildfire
Variation mean
MeanStandardMeanStandard
(Min/Max)deviation(Min/Max)deviation
pH 7.77 0.53 7.89 0.55 4.45% 
7.20 8.47  7.26 8.63   
EC (μS/cm) 604.59 506.43 1,038.75 519.24 36.65% 
42.45 11,127  42.59 28,936   
DO (mg/L) 9.81 2.33 10.15 2.50 17.63% 
4.33 13.30  4.83 12.77   
TN (mg/L) 4.34 2.10 3.44 1.56 −29.70% 
0.56 20.76  0.52 15.79   
TP (mg/L) 0.22 0.17 0.17 0.12 −50.68% 
0.01 1.66  0.01 1.82   
COD (mg/L) 6.71 2.60 5.74 2.26 −25.58% 
1.13 29.50  1.26 29.80   
BOD (mg/L) 5.66 3.28 3.79 2.36 −33.42% 
0.40 49.70  0.41 37.42   
SS (mg/L) 13.28 12.59 12.41 14.06 −44.05% 
0.40 52.20  0.43 42.82   
IndexBefore wildfire
After wildfire
Variation mean
MeanStandardMeanStandard
(Min/Max)deviation(Min/Max)deviation
pH 7.77 0.53 7.89 0.55 4.45% 
7.20 8.47  7.26 8.63   
EC (μS/cm) 604.59 506.43 1,038.75 519.24 36.65% 
42.45 11,127  42.59 28,936   
DO (mg/L) 9.81 2.33 10.15 2.50 17.63% 
4.33 13.30  4.83 12.77   
TN (mg/L) 4.34 2.10 3.44 1.56 −29.70% 
0.56 20.76  0.52 15.79   
TP (mg/L) 0.22 0.17 0.17 0.12 −50.68% 
0.01 1.66  0.01 1.82   
COD (mg/L) 6.71 2.60 5.74 2.26 −25.58% 
1.13 29.50  1.26 29.80   
BOD (mg/L) 5.66 3.28 3.79 2.36 −33.42% 
0.40 49.70  0.41 37.42   
SS (mg/L) 13.28 12.59 12.41 14.06 −44.05% 
0.40 52.20  0.43 42.82   
Figure 2

t-test, point-biserial correlation results by index.

Figure 2

t-test, point-biserial correlation results by index.

Close modal

Impact of wildfires on physicochemical indicators

Analysis of water quality changes in the river pH, EC, and DO after the wildfires confirmed a similar pattern. This pattern was also observed in observatories located near or downstream, excluding those in wildfire-affected areas. Among these, the pH was the most consistently affected index. Of the 76 datasets, it increased in 53 observatories after the wildfires and 29 of these showed significant increases. Of the 29 datasets showing a significant increase, more than 20 included wildfire occurrence sites (Table 2 and Figure 3(a)). We concluded that wildfires have a long-term effect on river pH.
Table 2

Distribution of water quality indicator increase points based on the inclusion of wildfire-damaged areas

IndexDamaged area
Around damaged area
naIncreasedSignificantnIncreasedSignificant
pH 36 28 20 34 20 
EC 36 22 13 34 22 14 
DO 36 22 15 34 24 
IndexDamaged area
Around damaged area
naIncreasedSignificantnIncreasedSignificant
pH 36 28 20 34 20 
EC 36 22 13 34 22 14 
DO 36 22 15 34 24 

an = number of stations.

Figure 3

Average by index before and after wildfires: (a) pH, (b) EC, and (c) DO.

Figure 3

Average by index before and after wildfires: (a) pH, (b) EC, and (c) DO.

Close modal

Next, consistent changes in EC were confirmed. Fifty datasets show consistent increases, with 31 reaching statistical significance. The numbers of observation stations, including the wildfire site and surrounding stations, were similar at 13 and 14 stations, respectively (Table 2 and Figure 3(b)). Unlike pH, EC affects water quality over a wide area, extending downstream from the wildfire-affected sites.

Finally, consistent changes were observed in DO. Of the 76 datasets, 49 showed increases in the DO levels, of which 24 were significant. Significant differences were identified in 15 datasets, including wildfire-affected areas (Table 2 and Figure 3(c)). In the case of DO, an indicator significantly affected by water temperature, most stations showed improvement in water quality after the wildfires.

Impact of wildfires on nutritional indicators

The analysis indicated a trend of decreasing levels of TN and TP after the wildfires compared to the levels observed before the wildfires. This trend was most pronounced for TN. Among the 55 datasets analyzed post-wildfire, 45 showed a statistically significant reduction in TN levels. Statistically significant reductions were noted at 24 locations among the datasets from the areas affected by wildfires (Table 3 and Figure 4(a)).
Table 3

Distribution of water quality indicator decrease points based on the inclusion of wildfire-damaged areas

IndexDamaged area
Around damaged area
naIncreasedSignificantnIncreasedSignificant
TN 36 29 24 34 22 18 
TP 36 23 14 34 20 
COD 36 22 17 34 15 11 
BOD 36 28 19 34 28 20 
SS 36 22 16 34 13 
IndexDamaged area
Around damaged area
naIncreasedSignificantnIncreasedSignificant
TN 36 29 24 34 22 18 
TP 36 23 14 34 20 
COD 36 22 17 34 15 11 
BOD 36 28 19 34 28 20 
SS 36 22 16 34 13 

an = number of stations.

Figure 4

Average by index before and after wildfires: (a) TN and (b) TP.

Figure 4

Average by index before and after wildfires: (a) TN and (b) TP.

Close modal

After the wildfires, an initial deterioration in water quality was observed, marked by increases in TN and TP as organic pollutants and sediment were washed into rivers, causing pollution. Nitrogen and phosphorus, primarily found in vegetation, bacteria, and soil, with the majority residing in the soil (Palviainen et al. 2017), contributed to this pollution. Short-term water quality impacts were detected at monitoring stations within and downstream of the wildfire-affected areas. However, during long-term monitoring, water quality improvements were noted, particularly through reductions in TP levels. Of the 49 datasets showing a decrease in TP, 25 exhibited significant reductions, with 14 significant decreases recorded at stations within wildfire-damaged areas (Table 3 and Figure 4(b)).

Impact of wildfires on organic matter indicators

Statistical analysis revealed long-term improvements in the water quality indicators COD and BOD. A statistically significant decrease in COD was observed at 40 locations, with 29 showing a statistically significant decrease. Among the datasets that included wildfire-affected areas, 17 showed significant decreases (Table 3 and Figure 5(a)). Similarly, a decrease in BOD was observed at 60 locations, with 42 locations showing a significant reduction. A significant decrease was observed in 19 datasets that included areas affected by wildfires (Table 3 and Figure 5(b)). After the wildfires, the levels of COD and BOD correlated with the flow rate of rivers (Ha et al. 2022), and there was a tendency for the levels to return to pre-wildfire levels within 3–6 months after the event.
Figure 5

Average by index before and after wildfires: (a) COD and (b) BOD.

Figure 5

Average by index before and after wildfires: (a) COD and (b) BOD.

Close modal
Figure 6

Average by index of SS before and after wildfires.

Figure 6

Average by index of SS before and after wildfires.

Close modal
Figure 7

Analysis of pH level correlations at monitoring stations before and after wildfire events.

Figure 7

Analysis of pH level correlations at monitoring stations before and after wildfire events.

Close modal
Figure 8

Analysis of EC level correlations at monitoring stations before and after wildfire events.

Figure 8

Analysis of EC level correlations at monitoring stations before and after wildfire events.

Close modal

Impact of wildfires on SSs

Statistical analysis indicated that the SS indicator was not significantly affected by wildfires. According to the analysis, 39 datasets showed a decrease, and 37 exhibited an increase. Among these, 13 showed a significant increase, and 22 showed a significant decrease Figure 6 and Table 3. Based on these results, we concluded that the SS indicator was not significantly influenced by wildfires. This conclusion aligns with previous research findings that reported that SS values may increase immediately after a wildfire but tend to return to pre-wildfire levels within 3–6 months (Uzun et al. 2020).

Role of post-wildfire on water quality

ANOVA was conducted on wildfires in Gangneung city, Gangwon province, among the locations affected by wildfires. In Gangneung city, a total of three major wildfires occurred in the years 2000, 2004, and 2017, all of which occurred within the Gangneung Namdaecheon watershed. The Ssangchon water quality monitoring station is located upstream and the Gangneung and Ponam water quality monitoring stations are located downstream. Examination of the sub-basin boundaries revealed that the Ssangchon water quality monitoring station experienced two major wildfires, whereas the Gangneung water quality monitoring station was affected by one. The analysis focused on indicators of water quality deterioration, such as pH and EC, reflecting previous analysis results.

ANOVA and correlation analysis were conducted on the pH levels at three water quality monitoring stations located alongside Gangneung Daenamcheon that were affected by wildfires in 2000, 2004, and 2017. The first analysis focused on wildfires in the watershed in 2000, including at the Gangneung water quality monitoring station. According to the ANOVA and correlation analysis results, significant differences were observed at the Ponam water quality monitoring station compared with the other two stations (Tables 4 and 5). Additionally, correlation analysis indicated that the correlation coefficient between stations increased after the wildfire, indicating that although the impact of the wildfire extended downstream, the extent of its effects varied (Figure 7).

Table 4

Tukey HSD post hoc test results for pH analysis before wildfire, following ANOVA

GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 0.028 0.9678 −0.2985 0.2425 False 
Middle_2000 Lower_2000 −0.0803 0.7641 −0.3508 0.1903 False 
Upper_2000 Lower_2000 −0.0523 0.8908 −0.3212 0.2167 False 
Upper_2004 Middle_2004 −0.0696 0.7549 −0.1599 0.2991 False 
Middle_2004 Lower_2004 −0.0202 0.9766 −0.2497 0.2093 False 
Upper_2004 Lower_2004 −0.0898 0.6252 −0.3186 0.1391 False 
Upper_2017 Middle_2017 0.2273 0.0341 −0.4411 −0.0135 True 
Middle_2017 Lower_2017 −0.4158 −0.629 −0.2026 True 
Upper_2017 Lower_2017 −0.1885 0.0938 −0.4011 0.0241 False 
GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 0.028 0.9678 −0.2985 0.2425 False 
Middle_2000 Lower_2000 −0.0803 0.7641 −0.3508 0.1903 False 
Upper_2000 Lower_2000 −0.0523 0.8908 −0.3212 0.2167 False 
Upper_2004 Middle_2004 −0.0696 0.7549 −0.1599 0.2991 False 
Middle_2004 Lower_2004 −0.0202 0.9766 −0.2497 0.2093 False 
Upper_2004 Lower_2004 −0.0898 0.6252 −0.3186 0.1391 False 
Upper_2017 Middle_2017 0.2273 0.0341 −0.4411 −0.0135 True 
Middle_2017 Lower_2017 −0.4158 −0.629 −0.2026 True 
Upper_2017 Lower_2017 −0.1885 0.0938 −0.4011 0.0241 False 
Table 5

Tukey HSD post hoc test results for pH analysis after wildfire, following ANOVA

GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 −0.0361 0.9147 −0.1753 0.2474 False 
Middle_2000 Lower_2000 −0.177 0.1208 −0.3883 0.0344 False 
Upper_2000 Lower_2000 −0.213 0.0469 −0.4238 −0.0023 True 
Upper_2004 Middle_2004 0.1663 0.1652 −0.3817 0.0491 False 
Middle_2004 Lower_2004 −0.5152 −0.7306 −0.2998 True 
Upper_2004 Lower_2004 −0.3489 0.0005 −0.5643 −0.1335 True 
Upper_2017 Middle_2017 0.2293 0.0628 −0.4682 0.0095 False 
Middle_2017 Lower_2017 −0.2 0.1206 −0.4388 0.0388 False 
Upper_2017 Lower_2017 0.0293 0.9548 −0.2095 0.2682 False 
GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 −0.0361 0.9147 −0.1753 0.2474 False 
Middle_2000 Lower_2000 −0.177 0.1208 −0.3883 0.0344 False 
Upper_2000 Lower_2000 −0.213 0.0469 −0.4238 −0.0023 True 
Upper_2004 Middle_2004 0.1663 0.1652 −0.3817 0.0491 False 
Middle_2004 Lower_2004 −0.5152 −0.7306 −0.2998 True 
Upper_2004 Lower_2004 −0.3489 0.0005 −0.5643 −0.1335 True 
Upper_2017 Middle_2017 0.2293 0.0628 −0.4682 0.0095 False 
Middle_2017 Lower_2017 −0.2 0.1206 −0.4388 0.0388 False 
Upper_2017 Lower_2017 0.0293 0.9548 −0.2095 0.2682 False 
Table 6

Tukey HSD post hoc test results for EC analysis before wildfire, following ANOVA

GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 5.5212 0.968 −59.1234 48.0811 False 
Middle_2000 Lower_2000 86.5163 0.0005 32.9141 140.1186 True 
Upper_2000 Lower_2000 92.0375 0.0002 38.7756 145.2994 True 
Upper_2004 Middle_2004 −2.2746 0.9838 −28.9121 33.4612 False 
Middle_2004 Lower_2004 34.5473 0.0258 3.3606 65.7339 True 
Upper_2004 Lower_2004 32.2727 0.0399 1.1753 63.3702 True 
Upper_2017 Middle_2017 24.422 0.3024 −63.3303 14.4863 False 
Middle_2017 Lower_2017 45.2441 0.0175 6.4446 84.0436 True 
Upper_2017 Lower_2017 69.6661 0.0001 30.9783 108.3539 True 
GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 5.5212 0.968 −59.1234 48.0811 False 
Middle_2000 Lower_2000 86.5163 0.0005 32.9141 140.1186 True 
Upper_2000 Lower_2000 92.0375 0.0002 38.7756 145.2994 True 
Upper_2004 Middle_2004 −2.2746 0.9838 −28.9121 33.4612 False 
Middle_2004 Lower_2004 34.5473 0.0258 3.3606 65.7339 True 
Upper_2004 Lower_2004 32.2727 0.0399 1.1753 63.3702 True 
Upper_2017 Middle_2017 24.422 0.3024 −63.3303 14.4863 False 
Middle_2017 Lower_2017 45.2441 0.0175 6.4446 84.0436 True 
Upper_2017 Lower_2017 69.6661 0.0001 30.9783 108.3539 True 
Table 7

Tukey HSD post hoc test results for EC analysis after wildfire, following ANOVA

GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 −2.2746 0.9849 −29.9961 34.5453 False 
Middle_2000 Lower_2000 47.079 0.002 14.8083 79.3497 True 
Upper_2000 Lower_2000 44.8043 0.0033 12.6219 76.9868 True 
Upper_2004 Middle_2004 14.7826 0.3705 −40.6449 11.0797 False 
Middle_2004 Lower_2004 51.1087 25.2464 76.971 True 
Upper_2004 Lower_2004 65.8913 40.029 91.7536 True 
Upper_2017 Middle_2017 9.1467 0.9769 −113.8848 95.5915 False 
Middle_2017 Lower_2017 22.4133 0.869 −82.3248 127.1515 False 
Upper_2017 Lower_2017 31.56 0.7573 −73.1781 136.2981 FALSE 
GroupMeandiffp-valueLowerUpperReject
Upper_2000 Middle_2000 −2.2746 0.9849 −29.9961 34.5453 False 
Middle_2000 Lower_2000 47.079 0.002 14.8083 79.3497 True 
Upper_2000 Lower_2000 44.8043 0.0033 12.6219 76.9868 True 
Upper_2004 Middle_2004 14.7826 0.3705 −40.6449 11.0797 False 
Middle_2004 Lower_2004 51.1087 25.2464 76.971 True 
Upper_2004 Lower_2004 65.8913 40.029 91.7536 True 
Upper_2017 Middle_2017 9.1467 0.9769 −113.8848 95.5915 False 
Middle_2017 Lower_2017 22.4133 0.869 −82.3248 127.1515 False 
Upper_2017 Lower_2017 31.56 0.7573 −73.1781 136.2981 FALSE 

Subsequent analysis was conducted on the wildfire in April 2004 in the basin, including that at the Ssangchon water quality monitoring station. Before the wildfire, the dataset showed no significant differences; however, the correlation analysis revealed a high correlation among the stations. After the wildfire, there was a significant difference between the Ponam stations and the other stations; overall, there was a difference between each station. This suggests that the analysis was influenced by the relatively short interval following the 2000 wildfire, which had a significant short-term impact.

The analysis of the wildfires in 2017 in the basin, including those at the Ssangchon water quality monitoring station, showed significant differences in combinations, excluding Ssangchon Ponam, before the wildfire. Correlation analysis prior to the wildfire indicated a low level of correlation. After the wildfire, no significant differences were observed, and the correlation analysis revealed a high level of correlation. This indicates that the wildfires in the Ssangchon water quality observatory basin affected the entire river.

Changes in EC after the wildfires were analyzed using ANOVA and correlation analysis for water quality observatories located upstream, middle, and downstream of the Gangneung Daenamcheon stream, including the three wildfires in 2000, 2004, and 2017. It has been confirmed that the correlation coefficient between stations has increased since the wildfires in 2000 and 2004 and that the impact of wildfires has affected the EC of downstream water stations. No strong correlations were observed after the 2017 wildfires (Figure 8). In addition, ANOVA revealed a significant difference in all pairings, except for Ssangchon-Gangneung, before the 2000 wildfire (Table 6). The 2004 wildfire showed no significant difference, with a correlation coefficient of 0.98 between Ssangchon and Gangneung before the wildfire event. However, significant differences were observed in all the pairings after the wildfire. This is believed to have decreased as the water pollution level decreased downstream after the wildfire and was diluted (Table 7). In 2017, significant differences were noted in all pairings before the wildfire. However, no differences were observed in any post-wildfire pairings.

This study investigated data over a 7-year period following major wildfires. The results indicated that changes in river water quality exhibited similar trends across most of the study areas. These findings are presented in Figure 2 and in Tables 2 and 3. These changes were consistent even in the downstream areas far from the wildfire. This was confirmed by the table of results for each indicator. This study focused on wildfires larger than 0.3 km2 and analyzed datasets divided into sub-basin units based on the occurrence of wildfires.

This study confirms the long-term impact of wildfires on river water quality. While previous research has predominantly focused on short-term changes, this study analyzed long-term trends across multiple rivers in Korea, underscoring the sustained effects of wildfires on water quality through statistical analysis. Following wildfires, water pollution in rivers becomes evident, particularly after the first rainfall, when rapid changes in water quality occur. Pollutants such as ash, soil, and metals from burned vegetation are washed into rivers, contributing to this pollution (Doerr et al. 2000; Larsen et al. 2009; Nam et al. 2022). The exposed, water-repellent soil layers exacerbate the inflow of pollutants into rivers, leading to increased runoff, accelerated soil erosion, and quicker transportation of contaminants into water systems (Jumps et al. 2022).

Wildfires also alter the physicochemical properties of soil organic matter, clay, and metallic minerals under high temperatures (Samburova et al. 2023). These changes increase the amount of water-soluble organic matter in the soil and facilitate the leaching of heavy metals into groundwater (Sazawa et al. 2020; Fajković et al. 2022). Nevertheless, long-term monitoring provides statistically significant evidence of gradual water quality improvement over time.

Among the analyzed indicators, SS shows a rapid increase immediately after a wildfire, typically returning to pre-wildfire levels within 3–4 months (Jeong et al. 2004). However, some studies have observed that SS levels may continue to rise for 5–7 years following a wildfire (Yu et al. 2019). SS levels generally recover as vegetation regrows and direct erosion from rainfall decreases, gradually bringing the concentration of SSs in rivers back to pre-wildfire levels (Lee et al. 2022). In this study, no long-term effects on SS levels in rivers were observed after wildfires.

Changes in pH and EC are indicators of long-term water pollution caused by the outflow of ash and metal components into the soil after a wildfire. Ash consists primarily of basic substances such as potassium carbonate and is composed of metal components containing sodium (Na) and magnesium (Mg) (Ulery et al. 1993; Pereira et al. 2014). When these components enter rivers, they increase the pH and EC of water (Romero-Matos et al. 2023).

Improvements in water quality were observed after the wildfires in terms of DO, BOD, TN, and TP. Following the short-term water pollution observed immediately after the wildfire, the water quality improved from a long-term perspective. In particular, DO is influenced by water temperature and seasonality and is inversely correlated with the level of pollution in rivers. It is also used as an indicator of the river's self-purification capabilities (Sanders et al. 2022). It is believed that the recovery of the river's self-purification capabilities and the reduction in pollutant discharge contributed to the decreases in BOD, TN, and TP.

This study provides a detailed analysis of the long-term effects of wildfires on river water quality by examining 7 years of post-wildfire data. The analysis of eight water quality indicators demonstrated that wildfires have impacts on river environments. Specifically, wildfires were associated with increases in pH, EC, and DO levels, while reductions were observed in BOD, TN, and TP concentrations. These findings suggest that wildfires induce notable changes in river water chemistry, potentially affecting water ecosystems.

Additionally, an ANOVA analysis conducted on three wildfire events in a single river showed an increase in the correlation between water quality indicators after the wildfires, with similar patterns of change observed at downstream monitoring stations. This suggests that the effects of wildfires may spread downstream over time, highlighting their potential for long-term and widespread impacts on river water quality.

The study does have limitations, as it was unable to fully account for all water quality indicators due to data collection constraints. However, by securing long-term data and analyzing water quality changes based on the scale and severity of wildfires, this research provides a more detailed understanding of post-wildfire water quality dynamics. Such findings offer valuable insights for developing water quality management and restoration plans following wildfires.

Future research should aim to achieve a more comprehensive understanding of the long-term effects of wildfires on water quality by analyzing a broader range of variables, including wildfire frequency, scale, and the characteristics of the affected areas, over an extended period.

This study was supported by a research fund from Chosun University, in 2023.

All relevant data are available from an online repository or repositories.

The authors declare there is no conflict.

Abatzoglou
J. T.
,
Battisti
D. S.
,
Williams
A. P.
,
Hansen
W. D.
,
Harvey
B. J.
&
Kolden
C. A.
(
2021
)
Projected increases in western US forest fire despite growing fuel constraints
,
Communications Earth & Environment
,
2
(
1
),
1
8
.
https://doi.org/10.1038/s43247-021-00299-0
.
Agbeshie
A. A.
,
Abugre
S.
,
Atta-Darkwa
T.
&
Awuah
R.
(
2022
)
A review of the effects of forest fire on soil properties
,
Journal of Forestry Research
,
33
(
5
),
1419
1441
.
https://doi.org/10.1007/s11676-022-01475-4
.
Bitner
K.
,
Gallaher
B.
&
Mullen
K.
(
2001
)
Review of Wildfire Effects on Chemical Water Quality (No. LA-13826-MS)
.
Los Alamos, NM, USA
:
Los Alamos National Lab.(LANL)
.
https://doi.org/10.2172/781455
.
Cha
S.
&
Shim
J. K.
(
2015
)
Effects of forest fire on soil nutrients dynamics
,
Korean Journal of Nature Conservation
,
13
(
3_4
),
97
110
.
https://doi.org/10.30960/kjnc.2015.13.3_4.97
.
De Palma-Dow
A.
,
McCullough
I. M.
&
Brentrup
J. A.
(
2022
)
Turning up the heat: Long-term water quality responses to wildfires and climate change in a hypereutrophic lake
,
Ecosphere
,
13
(
12
),
e4271
.
https://doi.org/10.1002/ecs2.4271
.
Doerr
S. H.
,
Shakesby
R. A.
&
Walsh
R.
(
2000
)
Soil water repellency: Its causes, characteristics and hydro-geomorphological significance
,
Earth-Science Reviews
,
51
(
1–4
),
33
65
.
https://doi.org/10.1016/S0012-8252(00)00011-8
.
Emelko
M. B.
,
Stone
M.
,
Silins
U.
,
Allin
D.
,
Collins
A. L.
,
Williams
C. H.
&
… Bladon
K. D.
(
2016
)
Sediment-phosphorus dynamics can shift aquatic ecology and cause downstream legacy effects after wildfire in large river systems
,
Global Change Biology
,
22
(
3
),
1168
1184
.
https://doi.org/10.1111/gcb.13073
.
Emmerton
C. A.
,
Cooke
C. A.
,
Hustins
S.
,
Silins
U.
,
Emelko
M. B.
,
Lewis
T.
&
… Orwin
J. F.
(
2020
)
Severe western Canadian wildfire affects water quality even at large basin scales
,
Water Research
,
183
,
116071
.
https://doi.org/10.1016/j.watres.2020.116071
.
Fajković
H.
,
Ivanić
M.
,
Nemet
I.
,
Rončević
S.
,
Kampić
Š.
&
Vazdar
L. D.
(
2022
)
Heat-induced changes in soil properties: Fires as cause for remobilization of chemical elements
. Journal of Hydrology and Hydromechanics, 70 (4), 421–431.
https://doi.org/10.2478/johh-2022-0024
.
Flannigan
M. D.
,
Wotton
B. M.
,
Marshall
G. A.
,
De Groot
W. J.
,
Johnston
J.
,
Jurko
N.
&
Cantin
A. S.
(
2016
)
Fuel moisture sensitivity to temperature and precipitation: Climate change implications
,
Climatic Change
,
134
,
59
71
.
https://doi.org/10.1007/s10584-015-1521-0
.
Ha
D. W.
,
Jung
K. Y.
,
Baek
J.
,
Lee
G. S.
,
Lee
Y.
,
Shin
D. S.
&
Na
E. H.
(
2022
)
Trend analysis using long-Term monitoring data of water quality at Churyeongcheon and Yocheon basins
,
Sustainability
,
14
(
15
),
9770
.
https://doi.org/10.3390/su14159770
.
Hampton
T. B.
,
Lin
S.
&
Basu
N. B.
(
2022
)
Forest fire effects on stream water quality at continental scales: A meta-analysis
,
Environmental Research Letters
,
17
(
6
),
064003
.
https://doi.org/10.1088/1748-9326/ac6a6c
.
Hohner
A. K.
,
Rhoades
C. C.
,
Wilkerson
P.
&
Rosario-Ortiz
F. L.
(
2019
)
Wildfires alter forest watersheds and threaten drinking water quality
,
Accounts of Chemical Research
,
52
(
5
),
1234
1244
.
https://doi.org/10.1021/acs.accounts.8b00670
.
Jeong
Y. H.
,
Kim
K. H.
,
Chung
C. G.
,
Jun
J. H.
&
Yoo
J. Y.
(
2004
)
Effects of forest fire on the quantities and qualities of the streamwater in small forest catchments
,
Journal of Korean Forestry Society
,
93
(
7
), 446–452.
Jumps
N.
,
Gray
A. B.
,
Guilinger
J. J.
&
Cowger
W. C.
(
2022
)
Wildfire impacts on the persistent suspended sediment dynamics of the Ventura River
,
California
.
Journal of Hydrology: Regional Studies
,
41
,
101096
.
https://doi.org/10.1016/j.ejrh.2022.101096
.
Korea Forest Research Institute
(
2020
) 2020 Korea's Forest Resources. Available at: https://kfss.forest.go.kr.
Kornbrot
D.
(
2014
)
Point biserial correlation. Wiley StatsRef: Statistics Reference Online. https://doi.org/10.1002/9781118445112.stat06227
.
Larsen
I. J.
,
MacDonald
L. H.
,
Brown
E.
,
Rough
D.
,
Welsh
M. J.
,
Pietraszek
J. H.
&
… Schaffrath
K.
(
2009
)
Causes of post-fire runoff and erosion: Water repellency, cover, or soil sealing?
,
Soil Science Society of America Journal
,
73
(
4
),
1393
1407
.
https://doi.org/10.2136/sssaj2007.0432
.
Lee
H. H.
(
2006
)
Characteristics of surface flow on the forest fire sites by using rainfall simulator
,
Journal of Korean Society of Forest Science
,
95
(
3
),
350
357
.
Lee
K. M.
(
2022
)
Changes in Soil Hydraulic Properties by Different Types of Forest Fire Damage and its Influence on Hydrologic Responses
.
Seoul: Graduate School of Seoul National University
.
https://doi.org/10.23170/snu.000000171133.11032.0001060
.
Lee
K. W.
,
Lee
C. W.
&
Eu
S.
(
2022
)
Temporal trend in landslide occurrences in post-fire areas over the past Two decades
,
Journal of the Korean Society of Hazard Mitigation
,
22
(
4
),
47
55
.
https://doi.org/10.9798/KOSHAM.2022.22.4.47
.
Mcdonald
J. H.
(
2009
)
Handbook of Biological Statistics
.
Baltimore, MD: Sparky House Publishing.
Ministry of Environment
(
2024
)
2024 Water Environment Monitoring Network Installation and Operation Plan
. https://www.water.or.kr.
Nam
S.
,
Yang
H.
,
Lim
H.
,
Kim
J.
,
Li
Q.
,
Moon
H.
&
Choi
H. T.
(
2022
)
Short-term effects of forest fire on water quality along a headwater stream in the immediate post-Fire period
,
Water
,
15
(
1
),
131
.
https://doi.org/10.3390/w15010131
.
Nanda
A.
,
Mohapatra
B. B.
,
Mahapatra
A. P. K.
,
Mahapatra
A. P. K.
&
Mahapatra
A. P. K.
(
2021
)
Multiple comparison test by tukey's honestly significant difference (HSD): Do the confident level control type I error
,
International Journal of Statistics and Applied Mathematics
,
6
(
1
),
59
65
.
https://doi.org/10.22271/maths.2021.v6.i1a.636
.
Palviainen
M.
,
Pumpanen
J.
,
Berninger
F.
,
Ritala
K.
,
Duan
B.
,
Heinonsalo
J.
, Sun, H., Köster, E. &
Köster
K.
(
2017
)
Nitrogen balance along a northern boreal forest fire chronosequence
,
PLoS One
,
12
(
3
),
e0174720
.
https://doi.org/10.1371/journal.pone.0174720
.
Paul
M. J.
,
LeDuc
S. D.
,
Lassiter
M. G.
,
Moorhead
L. C.
,
Noyes
P. D.
&
Leibowitz
S. G.
(
2022
)
Wildfire induces changes in receiving waters: A review with considerations for water quality management
,
Water Resources Research
,
58
(
9
),
e2021WR030699
.
https://doi.org/10.1029/2021WR030699
.
Pereira
P.
,
Úbeda
X.
,
Martin
D.
,
Mataix-Solera
J.
,
Cerdà
A.
&
Burguet
M.
(
2014
)
Wildfire effects on extractable elements in ash from a Pinus pinaster forest in Portugal
,
Hydrological Processes
,
28
(
11
),
3681
3690
.
https://doi.org/10.1002/hyp.9907
.
Romero-Matos
J.
,
Cánovas
C. R.
,
Macías
F.
,
Pérez-López
R.
,
León
R.
,
Millán-Becerro
R.
&
Nieto
J. M.
(
2023
)
Wildfire effects on the hydrogeochemistry of a river severely polluted by acid mine drainage
,
Water Research
,
233
,
119791
.
https://doi.org/10.1016/j.watres.2023.119791
.
Samburova
V.
,
Schneider
E.
,
Rüger
C. P.
,
Inouye
S.
,
Sion
B.
,
Axelrod
K.
&
… Moosmüller
H.
(
2023
)
Modification of soil hydroscopic and chemical properties caused by four recent California, USA megafires
.
Fire
,
6
(
5
),
186
.
https://doi.org/10.3390/fire6050186
.
Sanders
A. M.
,
Coble
A. A.
,
Swartz
A. G.
,
River
M.
,
James
P.
&
Warren
D. R.
(
2022
)
Heat and smoke from wildfires influence water temperature and dissolved oxygen levels in headwater streams
,
Freshwater Science
,
41
(
4
),
665
679
.
https://doi.org/10.1086/722632
.
Ulery
A. L.
,
Graham
R. C.
&
Amrhein
C.
(
1993
)
Wood-ash composition and soil pH following intense burning
,
Soil Science
,
156
(
5
),
358
364
.
Uzun
H.
,
Dahlgren
R. A.
,
Olivares
C.
,
Erdem
C. U.
,
Karanfil
T.
&
Chow
A. T.
(
2020
)
Two years of post-wildfire impacts on dissolved organic matter, nitrogen, and precursors of disinfection by-products in California stream waters
,
Water Research
,
181
,
115891
.
https://doi.org/10.1016/j.watres.2020.115891
.
Wagenbrenner
J. W.
,
Ebel
B. A.
,
Bladon
K. D.
&
Kinoshita
A. M.
(
2021
)
Post-wildfire hydrologic recovery in Mediterranean climates: A systematic review and case study to identify current knowledge and opportunities
,
Journal of Hydrology
,
602
,
126772
.
https://doi.org/10.1016/j.jhydrol.2021.126772
.
Welch
B. L.
(
1947
)
The generalization of ‘STUDENT'S'problem when several different population varlances are involved
,
Biometrika
,
34
(
1–2
),
28
35
.
https://doi.org/10.1093/biomet/34.1-2.28
.
Wildfire Statistical Yearbook
(
2022
)
Korea Forest Service
. https://www.forest.go.kr.
Yu
M.
,
Bishop
T. F.
&
Van Ogtrop
F. F.
(
2019
)
Assessment of the decadal impact of wildfire on water quality in forested catchments
,
Water
,
11
(
3
),
533
.
https://doi.org/10.3390/w11030533
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).