218
© IWA Publishing 2014 Water Science & Technology
|
70.2
|
2014
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea M. A. Paule, S. A. Memon, B.-Y. Lee, S. R. Umer and C.-H. Lee
ABSTRACT Stormwater runoff quality is sensitive to land use and land cover (LULC) change. It is difficult to understand their relationship in predicting the pollution potential and developing watershed management practices to eliminate or reduce the pollution risk. In this study, the relationship between LULC change and stormwater runoff quality in two separate monitoring sites comprising a construction area (Site 1) and mixed land use (Site 2) was analyzed using geographic information system (GIS), event mean concentration (EMC), and correlation analysis. It was detected that bare land area increased, while other land use areas such as agriculture, commercial, forest, grassland, parking lot, residential, and road reduced. Based on the analyses performed, high maximum range
M. A. Paule S. A. Memon B.-Y. Lee S. R. Umer C.-H. Lee (corresponding author) Department of Environmental Engineering and Energy, Myongji University, 116 Myongji-ro, Cheoin-gu, Yongin-si, Gyeonggi-do 449-728, South Korea E-mail: [email protected]
and average EMCs were found in Site 2 for most of the water pollutants. Also, urban areas and increased conversion of LULC into bare land corresponded to degradation of stormwater quality. Correlation analysis between LULC and stormwater quality showed the influence of different factors such as farming practices, geographical location, and amount of precipitation, vegetation loss, and anthropogenic activities in monitoring sites. This research found that GIS application was an efficient tool for monthly monitoring, validation and statistical analysis of LULC change in the study area. Key words
| event mean concentration, geographic information system, land use/land cover, stormwater
INTRODUCTION Stormwater runoff pollution is considered as a leading source for water quality impairment and degradation in Korea (Memon et al. ). Land development typically involves the removal of vegetation, soil grading and compaction, construction of impervious surfaces, and hydraulic conveyance systems (Emerson & Traver ). This process alters the watershed hydrology, reducing the vegetative interception of rainfall, infiltration, and groundwater recharge, with contaminant increases in surface water runoff (Li et al. ). As stormwater flows in on-going land development areas, it also picks up natural and manmade contaminants that are accumulated on the surface during dry days and transports them to receiving water bodies. The significance of the relationship between land development and stormwater runoff quality needs a deeper understanding due to increased recognition that non-point source (NPS) pollution has become a major environmental concern in Korea. Numerous studies have been performed to find the relationships between the water quality and doi: 10.2166/wst.2014.207
stormwater runoff (Zampella et al. ; Li et al. ; Lee et al. ; Chu et al. ). However, these studies have focused on suspended solids (SS) and nutrients. NPS pollution monitoring presents a great challenge because of the dispersed origin and loss of pollutants in response to hydrological processes and land use patterns. Also, a large set of data is required for the NPS pollution characterization contrary to point source pollution (Mishra et al. ). Consequently, geographic information system (GIS) and water quality models are used for the assessment and management of NPS pollution. In recent years, when Korea initiated its economic reform and open-door policy, rapid urbanization and economic expansion have resulted in massive land alteration and environmental degradation. Urbanization is considered as the major source of NPS pollution in the form of land use change impact on catchment morphological features and hydrological behavior, which completely disturbs the mechanisms of source, process and sink of urban NPS pollution.
219
M. A. Paule et al.
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Land use modifications associated with urbanization result in changes in stormwater runoff characteristics such as increase in runoff volume and peak flow rate. In this research, therefore, land use and land cover (LULC) change GIS data and stormwater runoff quality data from 2011 to 2012 were collected and analyzed: (1) to monitor the LULC change and its impact on runoff quality; (2) to determine the correlation factor for event mean concentration (EMC) of stormwater pollutants from different land use sites; and (3) to obtain baseline data for stormwater management and water quality modeling.
MATERIAL AND METHODS Site description The study was conducted within Geum-Hak watershed in Yongin City, Gyeonggi Province, which is located at the north-western part of South Korea (Figure 1). Two representative monitoring sites were selected on the basis of sitespecific and hydrological characteristics to investigate the impacts of LULC development on stormwater runoff quality. Site 1 (area 0.634 km2) is an outlet of a sedimentation pond and it was considered as a groundwork activities area due to on-going conversion of LULC for developing a residential complex. The LULC in this site includes land alteration (conversion of forest, agricultural, or commercial
Figure 1
|
Monitoring site locations.
Water Science & Technology
|
70.2
|
2014
to bare land), forest, and agricultural land. Site 2 (area 1.398 km2), which includes Site 1 and the additional downstream catchment area, is the outlet of the Geum-Hak stream that eventually flows into Paldang reservoir, the major source of drinking water for the Seoul Metropolitan area and nearby provinces. This site is categorized as a mixed catchment because it covers the land alteration, forest, agriculture, and urban area.
Methods The overall procedure used in this study is shown in Figure 2, which includes: (1) stormwater runoff monitoring, water quality and water quantity data analysis; (2) monthly monitoring of LULC change; (3) calculation of EMC; and (4) analysis of the relationship between LULC and water quality data using Pearson’s correlation.
Monitoring and analysis Fifteen storm events from June 2011 to December 2012 were monitored and a total of 398 grab samples (n ¼ 9–20 samples in each sites) were collected. Samples were analyzed for typical water quality pollutants including SS, chemical oxygen demand (COD), 5-day biological oxygen demand (BOD5), total nitrogen (TN), and total phosphorus (TP) using the standard methods (APHA ). Hydrologic
220
Figure 2
M. A. Paule et al.
|
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Water Science & Technology
|
70.2
|
2014
The overall procedure used in this study.
data gathered from each event included antecedent dry days (ADD), event rainfall, runoff duration, average rainfall intensity, and runoff rate. ADD is the number of dry days before a rain event, whereas rainfall intensity is a measure of the amount of rain that falls over time. LULC change pattern Land use maps from the Korean Ministry of Environment Republic of Korea (MOE) and Ministry of Agriculture and Forestry Republic of Korea (MAF) in raster format with 4 m resolution were used to measure the LULC composition in the study area. The land use types in the MAF maps were classified from Landsat TM images (30 m resolution). Monthly field visits were done to update and validate the LULC alteration within the study area. The original MOE land use map was classified into 23 land use types. However in this study, the LULC was reclassified into nine categories: agriculture, bare land (soil digging, soil filling, gravels, bare ground, and bare rocks), commercial, ditch/water (drainage system and settling pond), forest, grassland, parking lot, residential, and road. GIS was used to calculate the area of each
land use and update and validate the LULC change within the monitoring areas. Event mean concentration Typically, the concentration of a pollutant in stormwater runoff is described based on the EMC, due to the large fluctuation during rainfall events. The EMC can be defined as the total mass pollutant load yielded from a site during a storm event divided by the total runoff water volume discharged during the storm (Lang et al. ). It can be calculated as: EMC ¼
M ¼ V
P C Q Δt Pt t Qt Δt
where the EMC is the event mean concentration (mg/L); M is the total mass of pollutant over the entire event duration (g); V is total volume flow over entire storm event duration (m3); t is the time (min); Ct is the time variable concentration (mg/L); Qt is the time variable flow (m3/min); and Δt is the discrete time interval (min).
M. A. Paule et al.
221
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Correlation analysis Pearson’s (parametric) correlation coefficient matrix was adopted to perform bivariate analysis of storm water quality parameters and landscape pattern indices in the multivariate statistical analysis software SPSS 12.0. Correlations between EMC and rainfall variables and between EMC and LULC change pattern were determined. The two-sided test method was chosen for significant level of P-value at 0.05.
RESULTS AND DISCUSSION Rainfall and runoff characteristics All events above a minimum rainfall (generally 1.0 mm rainfall) were monitored. A summary of event details are shown in Table 1, which includes event date, ADD, rainfall,
Table 1
|
Hydrological description of stormwater runoff sampled
Runoff duration (min)
Event measured
ADD
Total rainfall
Average rainfall intensity
(days)
(mm)
(mm/hr)
Site 1
Site 2
22 June 2011
9
24.5
4.08
360
360
7 July 2011
3
49.5
9.90
300
300
26 July 2011
1
57.0
10.06
340
340
3 August 2011
0.8
1.0
0.33
180
180
12 August 2011
1
55.5
9.79
340
340
19 September 2011
9
16.5
1.42
700
700
14 October 2011
14
7
1.40
300
300
2 December 2011
1
4
0.40
600
600
29 June 2012
31
63.7
5.10
850
960
18 July 2012
3
33.5
2.64
385
860
12 August 2012 4 September 2012 13 September 2012 22 October 2012
23 3.8
28.5
2.37
280
705
74.0
5.69
700
775
4
11.0
2.64
–
530
11
47.5
5.68
–
495
–
480
16 November 2012
3
7.5
1.07
Minimum
0.8
1.0
0.33
Maximum
180.0 180.0
31.0
74.0
10.06
850.0 960.0
Median
3.8
28.5
2.64
350.0 495.0
Mean
7.8
32.0
4.17
444.6 528.3
± SD
8.9
24.2
3.45
211.2 231.7
SD: standard deviation.
Water Science & Technology
|
70.2
|
2014
average rainfall intensity, and runoff duration. The event rainfall varies from 1.0 to 74 mm; ADD varies from 0.8 to 31 days and; average rainfall intensity varies from 0.33 to 10.6 mm/hr. The runoff duration the during monitoring period ranged from 180 to 850 min in Site 1 and 180 to 960 min in Site 2. Hydrological variables were analyzed by using rainfall gauge data at monitoring sites, local meteorological data, and flow rate data. LULC change pattern LULC change in this study was classified into several distinct patterns, such as agriculture, bare land, commercial, ditch/water, forest, grassland, parking lot, residential, and road using the GIS technique. Figure 3 shows the LULC composition before groundwork (Phase I) and during groundwork activities between 2011 (Phase II) and 2012 (Phase III) in the construction area. Under Phase I, the LULC of agriculture, bare land, commercial, ditch/water, forest, grassland, parking lot, residential, and road accounted for 17.16, 4.79, 2.07, 0.98, 41.86, 12.45, 8.99, 5.47, and 6.23% of the total study area, respectively, but it changed rapidly due to groundwork activities during Phases II and III period, as presented in Table 2. Results revealed that bare land increased exponentially about 615.72% and ditch/water increased about 35.61% from Phases I to III period. However, other land uses such as agriculture (78.82%), commercial (44.16%), forest (15.44%), grassland (15.61%), parking lot (50.12%), residential (25.61%), and road (17.51%) reduced in spatial extent from Phases I to III. Event mean concentration Table 3 presents the statistics of EMC for some constituents measured during Phases II and III in both monitoring sites. The range of and average EMC values were found to be higher in Site 2 than in Site 1 for most of the pollutants. This could be because Site 2 covers a larger catchment area and collects discharge from Site 1 and surrounding urban areas, whereas the sedimentation pond at Site 1 caused removal of SS. Also, it was observed that Phase III generally had higher range and mean EMC values for all parameters possibly because construction of a settling pond, soil alteration (e.g. soil digging and soil filling), rapid increase of bare land, and vegetation loss were the major groundwork activities, contributing to the change in pollutant concentration in the study area. In similar studies, it was found that such conditions may affect soil erosion and
M. A. Paule et al.
222
Figure 3
|
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Water Science & Technology
|
|
70.2
2014
LULC patterns in construction area: (a) Phase 1, (b) Phase II (July), (c) Phase II (September), (d) Phase II (December), (e) Phase III (July), (f) Phase III (September), and (g) Phase III (December).
Table 2
|
Percentage increase of land use/land cover change (minus values indicate decrease in spatial extent)
Phase I Land use
Agriculture
m2
Phase II m2
%
239,993
Phase III m2
%
17.16
54,349
%
77.35
50,839
78.82
Bare land
66,999
4.79
369,584
451.63
479,526
615.72
Commercial
28,911
2.07
19,101
33.93
16,145
44.16
Ditch/water
13,648
0.98
22,157
62.34
18,509
35.61
Forest
585,381
41.86
516,529
11.76
494,976
15.44
Grassland
174,111
12.45
162,274
6.80
146,927
15.61
Parking lot
125,775
8.99
98,728
21.50
62,731
50.12
Residential
76,480
5.47
71,047
7.10
56,896
25.61
Road
87,084
6.23
84,613
2.84
71,832
17.51
1,398,382
100.00
1,398,382
352.68
1,398,382
404.06
TOTAL
sediment delivery, which caused increase of SS concentration in stormwater runoff (Chu et al. ). To determine the potential impact of rainfall variables in estimating the EMC of pollutant concerns, correlations were investigated for Phases II and III in both monitoring sites
(Table 4). From the correlation analysis, the ADD was weakly and negatively correlated to all EMC pollutants (the highest was TN in Phase III, r ¼ 0.456). Average rainfall intensity appears to be negatively correlated with all pollutants; r values ranged from 0.501 (TP in Phase III) to
Table 3
|
M. A. Paule et al.
223
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Water Science & Technology
|
70.2
|
2014
Summary of EMC at each monitoring site in Phases II and III (unit : mg/L)
Phase II
Phase III
Parameter
Site
Range
Mean
SD
Range
Mean
SD
SS
Site 1 Site 2
68.2–187.63 43.73–426.58
109 250.8
57.81 151.63
428.25–1927.55 52.48–3,652.53
915.12 1021.5
684.11 1218.11
BOD5
Site 1 Site 2
1.08–15.85 1.80–68.24
6.05 16.38
5.49 21.85
15.21–37.25 8.03–58.35
22.30 28.20
10.28 15.42
COD
Site 1 Site 2
13.28–58.42 19.94–303.67
29.40 88.06
15.82 107.02
29.52–70.16 14.13–202.75
49.03 73.26
16.68 62.72
TN
Site 1 Site 2
1.46–4.16 1.61–12.87
2.59 4.44
1.23 3.62
4.87–12.20 3.77–11.54
7.31 7.12
3.32 2.48
TP
Site 1 Site 2
0.33–0.87 0.54–2.20
0.52 1.19
0.21 0.61
6.20–31.52 0.84–45.49
17.85 15.36
10.97 15.50
SD: standard deviation.
Table 4
|
Correlation analysis results between EMCs and rainfall variables for Phase II
Table 5
(first row) and Phase III (second row)
BOD5
|
Pearson correlation coefficients for LULC change pattern and EMC for Phase II (first row) and Phase III (second row)
Parameter
SS
ADD
0.051 0.187
0.385 0.406
0.336 0.415
0.352 0.456
0.251 0.334
Average rainfall intensity
0.351 0.497
0.387 0.455
0.264 0.381
0.214 0.362
0.487 0.501
Runoff length
0.501 0.566
0.351 0.484
0.415 0.479
0.456 0.485
0.413 0.462
Rainfall depth
0.562 0.579
0.321 0.401
0.291 0.304
0.247 0.284
0.376 0.401
COD
TN
Water quality variables
TP
0.214 (TN in Phase II). Runoff length and runoff depth showed positive correlation to pollutants but some coefficient were low. BOD, COD, TN, and TP were positively correlated to runoff length than to rainfall depth; however, SS was more positively correlated with rainfall depth than with runoff length. Due to weak correlation between EMC and rainfall variables during both phases, it was hypothesized that not only the rainfall variables contributed to the EMCs; other factors should be considered. Relationship between land use and land cover change and water quality Table 5 shows the Pearson correlation coefficients between individual LULC and stormwater quality variables. Correlations significant at P < 0.05 are expressed in bold text. Agricultural land use showed positive correlation for TN and TP. Although it was less than 15% of total area (Figure 3), it may have contributed to an increase of nutrient
Land use types
SS
BOD
COD
TN
TP
Agriculture
0.201 0.130
0.284 0.252
0.095 0.076
0.601 0.707
0.653 0.698
Bare land
0.912 0.978
0.680 0.752
0.658 0.684
0.639 0.788
0.675 0.508
Commercial
0.714 0.638
0.757 0.736
0.603 0.652
0.680 0.540
0.634 0.526
Forest
0.350 0.249
0.618 0.574
0.342 0.276
0.282 0.261
0.276 0.246
Grassland
0.025 0.076
0.588 0.504
0.349 0.304
0.148 0.388
0.047 0.271
Parking lot
0.567 0.531
0.597 0.558
0.638 0.602
0.547 0.558
0.533 0.625
Residential
0.638 0.569
0.591 0.585
0.657 0.497
0.523 0.582
0.576 0.565
Road
0.729 0.570
0.608 0.597
0.656 0.501
0.692 0.584
0.590 0.507
Pollutants are in mg/L; all P-values are less than 0.05.
concentration in stormwater runoff due to possibly the effect of fertilizer used in rice paddy fields within the study area. In this study, it was hypothesized that the farming practices and geographical location were factors that affect the correlation between agricultural land and nutrients. A similar case in a previous study (Lee et al. ) found that the rice paddies in Korea received intensive application of fertilizers during spring and fall, in the seasons during which the water quality parameters were monitored. Also, the farmers keep paddies flooded after the application of fertilizer to
224
M. A. Paule et al.
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
allow increase of nutrient uptake by the rice plants. However, large amounts of precipitation can cause fertilized paddies to overflow through drainage weirs and more negatively affect water quality (Lee et al. ). Forest and grassland were negatively correlated with SS, COD, TN, and TP and were positively correlated with BOD. This result is similar to other research in which it was observed that increases in forest and grassland area will reduce the concentration of SS, COD, TP, and TN, increase the concentration of BOD, and consequently improve the water quality (Fu et al. ; Amiri & Nakane ). Urban land use corresponding to commercial, residential, parking lot, and road had positive correlation with all water quality parameters. This was obvious due to the high impervious cover within the catchment area, and a direct relation between imperviousness and surface runoff. The result is consistent with the findings from other research, which indicate that higher percentages of urban land use types had a profound influence on the quality of stormwater runoff owing to the introduction of pollutants of physical, chemical, and biological origin resulting from various anthropogenic activities common to urban areas (Goonetilleke et al. ). Typically, human land uses in watersheds result in fragmented natural areas with numerous negative ecological and environmental effects (Lee et al. ). Bare land and SS concentration have a positive relationship. A high degree of soil digging and transferring of soil from forest to elevated area was observed during Phase III. Therefore, bare land use was a source of soil erosion and sediment delivery to downstream receiving waters. It was found in previous studies that vegetation loss, soil disturbance, increased soil exposure, and heavy rainstorms will lead to an increase in the potential for soil erosion (Chu et al. ) and subsequently degradation of water quality. The positive relationship between bare land use and TN could be explained in the same manner as mentioned in a previous study whereby weathering of bare rocks and gravel during urban development activities could lead to increase in TN concentration (Li et al. ). The concentrations of BOD and COD were positively correlated with bare land, indicating that the increase of bare land area has resulted in discharge of a large amount of organic pollutants into the river, severely affecting water quality. Overall, the water quality variables were degraded from 2011 to 2012 where the bare land areas were dramatically increased and their impact was observed. Also, the builtup areas (commercial, parking lot, residential, and road) have a positive relationship with the water quality variables. So, it was clear that the main cause of deteriorating water
Water Science & Technology
|
70.2
|
2014
quality in the study area was bare land and built-up areas, which have the potential to generate a large amount of NPS pollution from stormwater runoff.
CONCLUSION Various studies have conducted investigations of the relationship between the LULC change and water quality in stormwater runoff. However, in explaining the relationship between LULC change and water quality, many studies have emphasized specific land uses such as urban and agricultural land. In this study, an integrated approach, involving stormwater runoff data, GIS, and EMC analysis, was used to conduct a study on the impact of the on-going construction activities and mixed land use on stormwater runoff. The result suggest that the concentration of most of the pollutants was found to be higher in Site 2, which consisted of the discharge from Site 1 (the construction zone) and surrounding urban areas. Rapid groundwork activities led to an increase in bare land use percentage and a decrease in built-up areas, which resulted in degradation of water quality in the study area specifically during Phase III. The EMCs for pollutants were mostly negatively or marginally correlated with all rainfall variables, which suggests that the runoff EMC is highly variable due to site and event characteristics. The agricultural land had positive correlation with nutrients. Farming practices, geographical location and amount of precipitation were the suggested factors for the nutrient correlation. The forest and grassland were negatively correlated to SS, COD, TN, and TP and were positively correlated to BOD. Bare land and SS had a positive relationship due to vegetation loss, soil disturbance, and increased soil exposure. The urban land use types showed positive correlation to all water quality parameters owing to their direct relation with imperviousness. The results indicate that application of GIS (monthly monitoring, updating, and validating the LULC change) to integrate EMC and correlation analysis can develop a decision-making support system to manage the land development and control the NPS impact at the watershed scale. This study provides scientific reference for local land use optimization and water pollution control and assists the formulation of policies for coordinating water resource exploitation and protection. In the future work, this study may refine the method and indicators to deeply reveal the reasons causing water quality change within the study area.
225
M. A. Paule et al.
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
ACKNOWLEDGEMENT This research was supported by the Korea Environmental Technology and Industrial Institute, Next Generation Eco Innovation Project (No. 413-111-003).
REFERENCES Amiri, B. & Nakane, K. Modeling the linkage between river water quality and landscape metrics in the Chugoku district of Japan. Water Resources Management 23 (1), 931–956. APHA Standard Methods for the Examination of Water and Wastewater 20th edn, American Public Health Association/ American Water Works Association/Water Environment Federation, Washington, DC, USA. Chu, H., Liu, C. & Wang, C. Identifying the relationships between water quality and land cover changes in the TsengWen Reservoir Watershed of Taiwan. International Journal of Environmental Research and Public Health 10 (2), 478–489. Emerson, C. H. & Traver, R. G. Multiyear and seasonal variation of infiltration from storm-water best management practices. Journal of Irrigation and Drainage Engineering. 134, 598–605. Fu, B., Ma, K., Zhou, H. & Chen, L. The impact of land use change on soil nutrition distribution of losses plateau. Science Bulletin 43 (22), 2444–2448. Goonetilleke, A., Thomas, E., Ginn, S. & Gilbert, D. Understanding the role of land use in urban stormwater quality management. Journal of Environmental Management. 74 (1), 31–42.
Water Science & Technology
|
70.2
|
2014
Lang, M., Li, P. & Yan, X. Runoff concentration and load of nitrogen and phosphorus from a residential area in an intensive agricultural watershed. Science of the Total Environment 458–460, 238–245. Lee, S., Hwang, S., Lee, S., Hwang, H. & Sung, H. Landscape ecological approach to the relationships of land use patterns in watersheds to water quality characteristics. Landscape and Urban Planning 92 (2), 80–89. Lee, J. Y., Yang, J. S., Kim, D. K. & Han, M. Y. Relationship between land use and water quality in a small watershed in South Korea. Water Science and Technology 62 (11), 2607–2615. Li, S., Gu, S., Liu, W., Han, H. & Zhang, Q. Water quality in relation to land use and land cover in the upper Han River Basin, China. CATENA 75 (2), 216–222. Li, H., Sharkey, L., Hunt, W. & Davis, A. Mitigation of impervious surface hydrology using bioretention in North Carolina and Maryland. Journal of Hydrologic Engineering 14 (Special Issue: Impervious Surfaces in Hydrologic Modeling and Monitoring), 407–415. Memon, S., Paule, M. C., Park, S., Lee, B., Kang, S., Umer, R. & Lee, C. Monitoring of land use change impact on stormwater runoff and pollutant loading estimation in Yongin watershed Korea. Desalination and Water Treatment 51 (19–21), 4088–4096. Mishra, A., Singh, R. & Singh, V. P. Evaluation of non-point source N and P loads in a small mixed land use land cover watershed. Journal Earth and Environmental Sciences 2 (4), 362–372. Zampella, R. A., Procopio, N. A., Lathrop, R. G. & Dow, C. L. Relationship of land-use-land-cover patterns and surface-water quality in the Mullica River Basin. Journal of the American Water Resources Association 43 (3), 594–604.
First received 5 December 2013; accepted in revised form 15 April 2014. Available online 3 May 2014
Copyright of Water Science & Technology is the property of IWA Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
© IWA Publishing 2014 Water Science & Technology
|
70.2
|
2014
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea M. A. Paule, S. A. Memon, B.-Y. Lee, S. R. Umer and C.-H. Lee
ABSTRACT Stormwater runoff quality is sensitive to land use and land cover (LULC) change. It is difficult to understand their relationship in predicting the pollution potential and developing watershed management practices to eliminate or reduce the pollution risk. In this study, the relationship between LULC change and stormwater runoff quality in two separate monitoring sites comprising a construction area (Site 1) and mixed land use (Site 2) was analyzed using geographic information system (GIS), event mean concentration (EMC), and correlation analysis. It was detected that bare land area increased, while other land use areas such as agriculture, commercial, forest, grassland, parking lot, residential, and road reduced. Based on the analyses performed, high maximum range
M. A. Paule S. A. Memon B.-Y. Lee S. R. Umer C.-H. Lee (corresponding author) Department of Environmental Engineering and Energy, Myongji University, 116 Myongji-ro, Cheoin-gu, Yongin-si, Gyeonggi-do 449-728, South Korea E-mail: [email protected]
and average EMCs were found in Site 2 for most of the water pollutants. Also, urban areas and increased conversion of LULC into bare land corresponded to degradation of stormwater quality. Correlation analysis between LULC and stormwater quality showed the influence of different factors such as farming practices, geographical location, and amount of precipitation, vegetation loss, and anthropogenic activities in monitoring sites. This research found that GIS application was an efficient tool for monthly monitoring, validation and statistical analysis of LULC change in the study area. Key words
| event mean concentration, geographic information system, land use/land cover, stormwater
INTRODUCTION Stormwater runoff pollution is considered as a leading source for water quality impairment and degradation in Korea (Memon et al. ). Land development typically involves the removal of vegetation, soil grading and compaction, construction of impervious surfaces, and hydraulic conveyance systems (Emerson & Traver ). This process alters the watershed hydrology, reducing the vegetative interception of rainfall, infiltration, and groundwater recharge, with contaminant increases in surface water runoff (Li et al. ). As stormwater flows in on-going land development areas, it also picks up natural and manmade contaminants that are accumulated on the surface during dry days and transports them to receiving water bodies. The significance of the relationship between land development and stormwater runoff quality needs a deeper understanding due to increased recognition that non-point source (NPS) pollution has become a major environmental concern in Korea. Numerous studies have been performed to find the relationships between the water quality and doi: 10.2166/wst.2014.207
stormwater runoff (Zampella et al. ; Li et al. ; Lee et al. ; Chu et al. ). However, these studies have focused on suspended solids (SS) and nutrients. NPS pollution monitoring presents a great challenge because of the dispersed origin and loss of pollutants in response to hydrological processes and land use patterns. Also, a large set of data is required for the NPS pollution characterization contrary to point source pollution (Mishra et al. ). Consequently, geographic information system (GIS) and water quality models are used for the assessment and management of NPS pollution. In recent years, when Korea initiated its economic reform and open-door policy, rapid urbanization and economic expansion have resulted in massive land alteration and environmental degradation. Urbanization is considered as the major source of NPS pollution in the form of land use change impact on catchment morphological features and hydrological behavior, which completely disturbs the mechanisms of source, process and sink of urban NPS pollution.
219
M. A. Paule et al.
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Land use modifications associated with urbanization result in changes in stormwater runoff characteristics such as increase in runoff volume and peak flow rate. In this research, therefore, land use and land cover (LULC) change GIS data and stormwater runoff quality data from 2011 to 2012 were collected and analyzed: (1) to monitor the LULC change and its impact on runoff quality; (2) to determine the correlation factor for event mean concentration (EMC) of stormwater pollutants from different land use sites; and (3) to obtain baseline data for stormwater management and water quality modeling.
MATERIAL AND METHODS Site description The study was conducted within Geum-Hak watershed in Yongin City, Gyeonggi Province, which is located at the north-western part of South Korea (Figure 1). Two representative monitoring sites were selected on the basis of sitespecific and hydrological characteristics to investigate the impacts of LULC development on stormwater runoff quality. Site 1 (area 0.634 km2) is an outlet of a sedimentation pond and it was considered as a groundwork activities area due to on-going conversion of LULC for developing a residential complex. The LULC in this site includes land alteration (conversion of forest, agricultural, or commercial
Figure 1
|
Monitoring site locations.
Water Science & Technology
|
70.2
|
2014
to bare land), forest, and agricultural land. Site 2 (area 1.398 km2), which includes Site 1 and the additional downstream catchment area, is the outlet of the Geum-Hak stream that eventually flows into Paldang reservoir, the major source of drinking water for the Seoul Metropolitan area and nearby provinces. This site is categorized as a mixed catchment because it covers the land alteration, forest, agriculture, and urban area.
Methods The overall procedure used in this study is shown in Figure 2, which includes: (1) stormwater runoff monitoring, water quality and water quantity data analysis; (2) monthly monitoring of LULC change; (3) calculation of EMC; and (4) analysis of the relationship between LULC and water quality data using Pearson’s correlation.
Monitoring and analysis Fifteen storm events from June 2011 to December 2012 were monitored and a total of 398 grab samples (n ¼ 9–20 samples in each sites) were collected. Samples were analyzed for typical water quality pollutants including SS, chemical oxygen demand (COD), 5-day biological oxygen demand (BOD5), total nitrogen (TN), and total phosphorus (TP) using the standard methods (APHA ). Hydrologic
220
Figure 2
M. A. Paule et al.
|
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Water Science & Technology
|
70.2
|
2014
The overall procedure used in this study.
data gathered from each event included antecedent dry days (ADD), event rainfall, runoff duration, average rainfall intensity, and runoff rate. ADD is the number of dry days before a rain event, whereas rainfall intensity is a measure of the amount of rain that falls over time. LULC change pattern Land use maps from the Korean Ministry of Environment Republic of Korea (MOE) and Ministry of Agriculture and Forestry Republic of Korea (MAF) in raster format with 4 m resolution were used to measure the LULC composition in the study area. The land use types in the MAF maps were classified from Landsat TM images (30 m resolution). Monthly field visits were done to update and validate the LULC alteration within the study area. The original MOE land use map was classified into 23 land use types. However in this study, the LULC was reclassified into nine categories: agriculture, bare land (soil digging, soil filling, gravels, bare ground, and bare rocks), commercial, ditch/water (drainage system and settling pond), forest, grassland, parking lot, residential, and road. GIS was used to calculate the area of each
land use and update and validate the LULC change within the monitoring areas. Event mean concentration Typically, the concentration of a pollutant in stormwater runoff is described based on the EMC, due to the large fluctuation during rainfall events. The EMC can be defined as the total mass pollutant load yielded from a site during a storm event divided by the total runoff water volume discharged during the storm (Lang et al. ). It can be calculated as: EMC ¼
M ¼ V
P C Q Δt Pt t Qt Δt
where the EMC is the event mean concentration (mg/L); M is the total mass of pollutant over the entire event duration (g); V is total volume flow over entire storm event duration (m3); t is the time (min); Ct is the time variable concentration (mg/L); Qt is the time variable flow (m3/min); and Δt is the discrete time interval (min).
M. A. Paule et al.
221
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Correlation analysis Pearson’s (parametric) correlation coefficient matrix was adopted to perform bivariate analysis of storm water quality parameters and landscape pattern indices in the multivariate statistical analysis software SPSS 12.0. Correlations between EMC and rainfall variables and between EMC and LULC change pattern were determined. The two-sided test method was chosen for significant level of P-value at 0.05.
RESULTS AND DISCUSSION Rainfall and runoff characteristics All events above a minimum rainfall (generally 1.0 mm rainfall) were monitored. A summary of event details are shown in Table 1, which includes event date, ADD, rainfall,
Table 1
|
Hydrological description of stormwater runoff sampled
Runoff duration (min)
Event measured
ADD
Total rainfall
Average rainfall intensity
(days)
(mm)
(mm/hr)
Site 1
Site 2
22 June 2011
9
24.5
4.08
360
360
7 July 2011
3
49.5
9.90
300
300
26 July 2011
1
57.0
10.06
340
340
3 August 2011
0.8
1.0
0.33
180
180
12 August 2011
1
55.5
9.79
340
340
19 September 2011
9
16.5
1.42
700
700
14 October 2011
14
7
1.40
300
300
2 December 2011
1
4
0.40
600
600
29 June 2012
31
63.7
5.10
850
960
18 July 2012
3
33.5
2.64
385
860
12 August 2012 4 September 2012 13 September 2012 22 October 2012
23 3.8
28.5
2.37
280
705
74.0
5.69
700
775
4
11.0
2.64
–
530
11
47.5
5.68
–
495
–
480
16 November 2012
3
7.5
1.07
Minimum
0.8
1.0
0.33
Maximum
180.0 180.0
31.0
74.0
10.06
850.0 960.0
Median
3.8
28.5
2.64
350.0 495.0
Mean
7.8
32.0
4.17
444.6 528.3
± SD
8.9
24.2
3.45
211.2 231.7
SD: standard deviation.
Water Science & Technology
|
70.2
|
2014
average rainfall intensity, and runoff duration. The event rainfall varies from 1.0 to 74 mm; ADD varies from 0.8 to 31 days and; average rainfall intensity varies from 0.33 to 10.6 mm/hr. The runoff duration the during monitoring period ranged from 180 to 850 min in Site 1 and 180 to 960 min in Site 2. Hydrological variables were analyzed by using rainfall gauge data at monitoring sites, local meteorological data, and flow rate data. LULC change pattern LULC change in this study was classified into several distinct patterns, such as agriculture, bare land, commercial, ditch/water, forest, grassland, parking lot, residential, and road using the GIS technique. Figure 3 shows the LULC composition before groundwork (Phase I) and during groundwork activities between 2011 (Phase II) and 2012 (Phase III) in the construction area. Under Phase I, the LULC of agriculture, bare land, commercial, ditch/water, forest, grassland, parking lot, residential, and road accounted for 17.16, 4.79, 2.07, 0.98, 41.86, 12.45, 8.99, 5.47, and 6.23% of the total study area, respectively, but it changed rapidly due to groundwork activities during Phases II and III period, as presented in Table 2. Results revealed that bare land increased exponentially about 615.72% and ditch/water increased about 35.61% from Phases I to III period. However, other land uses such as agriculture (78.82%), commercial (44.16%), forest (15.44%), grassland (15.61%), parking lot (50.12%), residential (25.61%), and road (17.51%) reduced in spatial extent from Phases I to III. Event mean concentration Table 3 presents the statistics of EMC for some constituents measured during Phases II and III in both monitoring sites. The range of and average EMC values were found to be higher in Site 2 than in Site 1 for most of the pollutants. This could be because Site 2 covers a larger catchment area and collects discharge from Site 1 and surrounding urban areas, whereas the sedimentation pond at Site 1 caused removal of SS. Also, it was observed that Phase III generally had higher range and mean EMC values for all parameters possibly because construction of a settling pond, soil alteration (e.g. soil digging and soil filling), rapid increase of bare land, and vegetation loss were the major groundwork activities, contributing to the change in pollutant concentration in the study area. In similar studies, it was found that such conditions may affect soil erosion and
M. A. Paule et al.
222
Figure 3
|
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Water Science & Technology
|
|
70.2
2014
LULC patterns in construction area: (a) Phase 1, (b) Phase II (July), (c) Phase II (September), (d) Phase II (December), (e) Phase III (July), (f) Phase III (September), and (g) Phase III (December).
Table 2
|
Percentage increase of land use/land cover change (minus values indicate decrease in spatial extent)
Phase I Land use
Agriculture
m2
Phase II m2
%
239,993
Phase III m2
%
17.16
54,349
%
77.35
50,839
78.82
Bare land
66,999
4.79
369,584
451.63
479,526
615.72
Commercial
28,911
2.07
19,101
33.93
16,145
44.16
Ditch/water
13,648
0.98
22,157
62.34
18,509
35.61
Forest
585,381
41.86
516,529
11.76
494,976
15.44
Grassland
174,111
12.45
162,274
6.80
146,927
15.61
Parking lot
125,775
8.99
98,728
21.50
62,731
50.12
Residential
76,480
5.47
71,047
7.10
56,896
25.61
Road
87,084
6.23
84,613
2.84
71,832
17.51
1,398,382
100.00
1,398,382
352.68
1,398,382
404.06
TOTAL
sediment delivery, which caused increase of SS concentration in stormwater runoff (Chu et al. ). To determine the potential impact of rainfall variables in estimating the EMC of pollutant concerns, correlations were investigated for Phases II and III in both monitoring sites
(Table 4). From the correlation analysis, the ADD was weakly and negatively correlated to all EMC pollutants (the highest was TN in Phase III, r ¼ 0.456). Average rainfall intensity appears to be negatively correlated with all pollutants; r values ranged from 0.501 (TP in Phase III) to
Table 3
|
M. A. Paule et al.
223
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
Water Science & Technology
|
70.2
|
2014
Summary of EMC at each monitoring site in Phases II and III (unit : mg/L)
Phase II
Phase III
Parameter
Site
Range
Mean
SD
Range
Mean
SD
SS
Site 1 Site 2
68.2–187.63 43.73–426.58
109 250.8
57.81 151.63
428.25–1927.55 52.48–3,652.53
915.12 1021.5
684.11 1218.11
BOD5
Site 1 Site 2
1.08–15.85 1.80–68.24
6.05 16.38
5.49 21.85
15.21–37.25 8.03–58.35
22.30 28.20
10.28 15.42
COD
Site 1 Site 2
13.28–58.42 19.94–303.67
29.40 88.06
15.82 107.02
29.52–70.16 14.13–202.75
49.03 73.26
16.68 62.72
TN
Site 1 Site 2
1.46–4.16 1.61–12.87
2.59 4.44
1.23 3.62
4.87–12.20 3.77–11.54
7.31 7.12
3.32 2.48
TP
Site 1 Site 2
0.33–0.87 0.54–2.20
0.52 1.19
0.21 0.61
6.20–31.52 0.84–45.49
17.85 15.36
10.97 15.50
SD: standard deviation.
Table 4
|
Correlation analysis results between EMCs and rainfall variables for Phase II
Table 5
(first row) and Phase III (second row)
BOD5
|
Pearson correlation coefficients for LULC change pattern and EMC for Phase II (first row) and Phase III (second row)
Parameter
SS
ADD
0.051 0.187
0.385 0.406
0.336 0.415
0.352 0.456
0.251 0.334
Average rainfall intensity
0.351 0.497
0.387 0.455
0.264 0.381
0.214 0.362
0.487 0.501
Runoff length
0.501 0.566
0.351 0.484
0.415 0.479
0.456 0.485
0.413 0.462
Rainfall depth
0.562 0.579
0.321 0.401
0.291 0.304
0.247 0.284
0.376 0.401
COD
TN
Water quality variables
TP
0.214 (TN in Phase II). Runoff length and runoff depth showed positive correlation to pollutants but some coefficient were low. BOD, COD, TN, and TP were positively correlated to runoff length than to rainfall depth; however, SS was more positively correlated with rainfall depth than with runoff length. Due to weak correlation between EMC and rainfall variables during both phases, it was hypothesized that not only the rainfall variables contributed to the EMCs; other factors should be considered. Relationship between land use and land cover change and water quality Table 5 shows the Pearson correlation coefficients between individual LULC and stormwater quality variables. Correlations significant at P < 0.05 are expressed in bold text. Agricultural land use showed positive correlation for TN and TP. Although it was less than 15% of total area (Figure 3), it may have contributed to an increase of nutrient
Land use types
SS
BOD
COD
TN
TP
Agriculture
0.201 0.130
0.284 0.252
0.095 0.076
0.601 0.707
0.653 0.698
Bare land
0.912 0.978
0.680 0.752
0.658 0.684
0.639 0.788
0.675 0.508
Commercial
0.714 0.638
0.757 0.736
0.603 0.652
0.680 0.540
0.634 0.526
Forest
0.350 0.249
0.618 0.574
0.342 0.276
0.282 0.261
0.276 0.246
Grassland
0.025 0.076
0.588 0.504
0.349 0.304
0.148 0.388
0.047 0.271
Parking lot
0.567 0.531
0.597 0.558
0.638 0.602
0.547 0.558
0.533 0.625
Residential
0.638 0.569
0.591 0.585
0.657 0.497
0.523 0.582
0.576 0.565
Road
0.729 0.570
0.608 0.597
0.656 0.501
0.692 0.584
0.590 0.507
Pollutants are in mg/L; all P-values are less than 0.05.
concentration in stormwater runoff due to possibly the effect of fertilizer used in rice paddy fields within the study area. In this study, it was hypothesized that the farming practices and geographical location were factors that affect the correlation between agricultural land and nutrients. A similar case in a previous study (Lee et al. ) found that the rice paddies in Korea received intensive application of fertilizers during spring and fall, in the seasons during which the water quality parameters were monitored. Also, the farmers keep paddies flooded after the application of fertilizer to
224
M. A. Paule et al.
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
allow increase of nutrient uptake by the rice plants. However, large amounts of precipitation can cause fertilized paddies to overflow through drainage weirs and more negatively affect water quality (Lee et al. ). Forest and grassland were negatively correlated with SS, COD, TN, and TP and were positively correlated with BOD. This result is similar to other research in which it was observed that increases in forest and grassland area will reduce the concentration of SS, COD, TP, and TN, increase the concentration of BOD, and consequently improve the water quality (Fu et al. ; Amiri & Nakane ). Urban land use corresponding to commercial, residential, parking lot, and road had positive correlation with all water quality parameters. This was obvious due to the high impervious cover within the catchment area, and a direct relation between imperviousness and surface runoff. The result is consistent with the findings from other research, which indicate that higher percentages of urban land use types had a profound influence on the quality of stormwater runoff owing to the introduction of pollutants of physical, chemical, and biological origin resulting from various anthropogenic activities common to urban areas (Goonetilleke et al. ). Typically, human land uses in watersheds result in fragmented natural areas with numerous negative ecological and environmental effects (Lee et al. ). Bare land and SS concentration have a positive relationship. A high degree of soil digging and transferring of soil from forest to elevated area was observed during Phase III. Therefore, bare land use was a source of soil erosion and sediment delivery to downstream receiving waters. It was found in previous studies that vegetation loss, soil disturbance, increased soil exposure, and heavy rainstorms will lead to an increase in the potential for soil erosion (Chu et al. ) and subsequently degradation of water quality. The positive relationship between bare land use and TN could be explained in the same manner as mentioned in a previous study whereby weathering of bare rocks and gravel during urban development activities could lead to increase in TN concentration (Li et al. ). The concentrations of BOD and COD were positively correlated with bare land, indicating that the increase of bare land area has resulted in discharge of a large amount of organic pollutants into the river, severely affecting water quality. Overall, the water quality variables were degraded from 2011 to 2012 where the bare land areas were dramatically increased and their impact was observed. Also, the builtup areas (commercial, parking lot, residential, and road) have a positive relationship with the water quality variables. So, it was clear that the main cause of deteriorating water
Water Science & Technology
|
70.2
|
2014
quality in the study area was bare land and built-up areas, which have the potential to generate a large amount of NPS pollution from stormwater runoff.
CONCLUSION Various studies have conducted investigations of the relationship between the LULC change and water quality in stormwater runoff. However, in explaining the relationship between LULC change and water quality, many studies have emphasized specific land uses such as urban and agricultural land. In this study, an integrated approach, involving stormwater runoff data, GIS, and EMC analysis, was used to conduct a study on the impact of the on-going construction activities and mixed land use on stormwater runoff. The result suggest that the concentration of most of the pollutants was found to be higher in Site 2, which consisted of the discharge from Site 1 (the construction zone) and surrounding urban areas. Rapid groundwork activities led to an increase in bare land use percentage and a decrease in built-up areas, which resulted in degradation of water quality in the study area specifically during Phase III. The EMCs for pollutants were mostly negatively or marginally correlated with all rainfall variables, which suggests that the runoff EMC is highly variable due to site and event characteristics. The agricultural land had positive correlation with nutrients. Farming practices, geographical location and amount of precipitation were the suggested factors for the nutrient correlation. The forest and grassland were negatively correlated to SS, COD, TN, and TP and were positively correlated to BOD. Bare land and SS had a positive relationship due to vegetation loss, soil disturbance, and increased soil exposure. The urban land use types showed positive correlation to all water quality parameters owing to their direct relation with imperviousness. The results indicate that application of GIS (monthly monitoring, updating, and validating the LULC change) to integrate EMC and correlation analysis can develop a decision-making support system to manage the land development and control the NPS impact at the watershed scale. This study provides scientific reference for local land use optimization and water pollution control and assists the formulation of policies for coordinating water resource exploitation and protection. In the future work, this study may refine the method and indicators to deeply reveal the reasons causing water quality change within the study area.
225
M. A. Paule et al.
|
Stormwater runoff quality in correlation to land use and land cover development in Yongin, South Korea
ACKNOWLEDGEMENT This research was supported by the Korea Environmental Technology and Industrial Institute, Next Generation Eco Innovation Project (No. 413-111-003).
REFERENCES Amiri, B. & Nakane, K. Modeling the linkage between river water quality and landscape metrics in the Chugoku district of Japan. Water Resources Management 23 (1), 931–956. APHA Standard Methods for the Examination of Water and Wastewater 20th edn, American Public Health Association/ American Water Works Association/Water Environment Federation, Washington, DC, USA. Chu, H., Liu, C. & Wang, C. Identifying the relationships between water quality and land cover changes in the TsengWen Reservoir Watershed of Taiwan. International Journal of Environmental Research and Public Health 10 (2), 478–489. Emerson, C. H. & Traver, R. G. Multiyear and seasonal variation of infiltration from storm-water best management practices. Journal of Irrigation and Drainage Engineering. 134, 598–605. Fu, B., Ma, K., Zhou, H. & Chen, L. The impact of land use change on soil nutrition distribution of losses plateau. Science Bulletin 43 (22), 2444–2448. Goonetilleke, A., Thomas, E., Ginn, S. & Gilbert, D. Understanding the role of land use in urban stormwater quality management. Journal of Environmental Management. 74 (1), 31–42.
Water Science & Technology
|
70.2
|
2014
Lang, M., Li, P. & Yan, X. Runoff concentration and load of nitrogen and phosphorus from a residential area in an intensive agricultural watershed. Science of the Total Environment 458–460, 238–245. Lee, S., Hwang, S., Lee, S., Hwang, H. & Sung, H. Landscape ecological approach to the relationships of land use patterns in watersheds to water quality characteristics. Landscape and Urban Planning 92 (2), 80–89. Lee, J. Y., Yang, J. S., Kim, D. K. & Han, M. Y. Relationship between land use and water quality in a small watershed in South Korea. Water Science and Technology 62 (11), 2607–2615. Li, S., Gu, S., Liu, W., Han, H. & Zhang, Q. Water quality in relation to land use and land cover in the upper Han River Basin, China. CATENA 75 (2), 216–222. Li, H., Sharkey, L., Hunt, W. & Davis, A. Mitigation of impervious surface hydrology using bioretention in North Carolina and Maryland. Journal of Hydrologic Engineering 14 (Special Issue: Impervious Surfaces in Hydrologic Modeling and Monitoring), 407–415. Memon, S., Paule, M. C., Park, S., Lee, B., Kang, S., Umer, R. & Lee, C. Monitoring of land use change impact on stormwater runoff and pollutant loading estimation in Yongin watershed Korea. Desalination and Water Treatment 51 (19–21), 4088–4096. Mishra, A., Singh, R. & Singh, V. P. Evaluation of non-point source N and P loads in a small mixed land use land cover watershed. Journal Earth and Environmental Sciences 2 (4), 362–372. Zampella, R. A., Procopio, N. A., Lathrop, R. G. & Dow, C. L. Relationship of land-use-land-cover patterns and surface-water quality in the Mullica River Basin. Journal of the American Water Resources Association 43 (3), 594–604.
First received 5 December 2013; accepted in revised form 15 April 2014. Available online 3 May 2014
Copyright of Water Science & Technology is the property of IWA Publishing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
