Large barrages have been constructed on the main rivers in South Korea to store water and mitigate fluvial flooding damage. However, the increase in water levels behind the barrages can potentially lead to a rise in groundwater levels in the riversides. The purpose of this study was to describe the effect of a barrage on groundwater levels and to test the applicability of a numerical model to groundwater inundation in this context. The Shincheon–Baekcheon catchment is characterised mainly by agricultural land use and includes significant greenhouse cultivation. Its two zones, which are lower A and upper B basins, mainly yield fine- and coarse-grained deposits, respectively. Trend and distribution analyses of manual and automatic measurements of groundwater levels indicated that: (1) the groundwater levels generally increased as the river water levels rose after the river was dammed; (2) the significant correlation between groundwater and river water levels could lead to reductions in the groundwater levels if the barrage gates were opened as a control measure; and (3) the lowering of high groundwater levels during dry seasons is important for preventing soil wetting in the riversides.

INTRODUCTION

In 2008, the South Korean Government launched the Four Major Rivers Restoration Project (4MRRP), a large project whose vision is part of the ‘Reviving the Rivers: A New Korea’. The objectives of the 4MRRP included: (1) mitigating water-related problems caused by climate change; (2) achieving a balance between the needs of humans and the environment; (3) re-creation of natural land; and (4) increasing the harmony between local development and environmental sustainability. The five core tasks of the 4MRRP were: (1) securing the water supply; (2) mitigating flood damage; (3) improving water quality and restoring the ecosystem; (4) creating public spaces for residents; and (5) river-oriented community development (MLTM 2010). The project commenced in 2009 and was completed in the summer of 2012, after an investment of 20 billion US dollars. The main elements of the project were the dredging of stream sediments, construction of barrages, reinforcement of river banks, small-scale hydropower production, and ecological streams. A total of 0.57 billion m3 of sediment was dredged along the four main rivers. Sixteen large barrages and three large dams were planned and constructed to store water and control flooding.

A number of researchers have demonstrated the connection between surface water and groundwater with diverse approaches, including water level measurement, geochemical data, water temperature differences, tracers, isotope studies, and numerical models (Sophocleous et al. 1988; Winter et al. 1988; Harvey et al. 1997; Constantz et al. 2001; Paulsen et al. 2001; LaBolle et al. 2003; Anderson 2005; Arumi et al. 2009; Guggenmos et al. 2011). Many researchers have also shown that increases in the surface water levels caused by canal excavation or barrage construction can alter the depth to groundwater (Dreher 1991; Hill 1996; Dreher & Gunatilaka 1998; Harvey & Sibray 2001; Renken et al. 2005; Kim et al. 2013, 2014). The Freudenau power plant in Vienna, Austria, was constructed in 1992 on the Danube River. The elevated water level resulting from the impoundment of the river had the potential to drastically alter the groundwater flow and to submerge some areas of the city. To prevent this inundation, a system of sealing walls was constructed along the western river bank, and a groundwater management system using pumping/injection wells and monitoring wells was proposed. This fully automated groundwater management system maintains the groundwater flow in a nearly natural and predefined state (Dreher & Gunatilaka 1996; Gunatilaka & Dreher 1996). The construction of Aswan Dam in Egypt resulted in continuously rising levels of saline groundwater. The groundwater infiltrated the porous sandstone and limestone foundations, and the dissolved salts were absorbed by the historical structures. A groundwater pumping system was introduced to help reduce the damage caused by the saline groundwater (Hassan 2006).

The infiltration of surface water into a river bank and the surrounding alluvium and the consequent rise in groundwater levels can potentially increase the risk of surface flooding. Shallow groundwater levels are generally found in low-lying areas and in close proximity to surface water bodies. Shallow groundwater can result in marshy conditions, which encourage the growth of willows and aquatic plants, thus limiting the usefulness of the land. Heavy rainfall over the land can also increase the extent of shallow groundwater levels and increase the potential for groundwater flooding. In many shallow groundwater zones, such as wetlands and riparian areas, groundwater fluctuations can have a direct influence on root zone soil moisture; consequently, capillary action affects the extent of wet or swampy areas. Furthermore, surface evaporation and infiltration are influenced by shallow groundwater, with reductions in infiltration leading to increased surface runoff and flooding (Chen & Hu 2004).

New approaches to flood risk assessment based on groundwater hydrology have been studied in a number of countries, including the United Kingdom, France, Germany, and Ireland (Macdonald et al. 2007, 2008; Cobby et al. 2009; Kreibich et al. 2009; Sommer et al. 2009; Pennequin 2010; Hughes et al. 2012; Naughton et al. 2012). Two types of groundwater flooding commonly occur. First, a rising water level in an unconfined aquifer above the land surface can produce groundwater flooding as a rapid response to extreme rainfall and a low water-absorbing capacity. This occurs, for example, in the chalk outcrops of Europe. Second, if groundwater levels rise to the surface in shallow unconsolidated sedimentary aquifers hydraulically connected to adjacent river networks, groundwater flooding can occur in response to heavy rainfall and high river water levels.

The aim of this study was to analyse recent trends in groundwater levels in the upper watershed of the Gangjeong-Koryeong barrage, a large new barrage in South Korea. The study was conducted using groundwater monitoring and field survey data. The applicability of a numerical model to forecast groundwater levels during heavy rain and during a dry season was also evaluated. The potential for inundation of the low-lying areas is also discussed.

STUDY AREA

The study area is located in Seongju-gun in Kyeongsang-bukdo Province in the south-eastern part of the Korean peninsula. The study area is a part of the Sincheon–Baekcheon basin which is drained by the Nakdong River. The study area is approximately 27 km2 in area (Figure 1). The hydrological outlet of the basin is in the southeast, where water flows into the Nakdong, the largest river in Korea. The northeastern and southern boundaries of the study area are marked by the Baekcheon and Sincheon streams, which are 28.0 km and 13.5 km in length, respectively. These streams flow into the Nakdong River at the southeastern end of the study area. The northern boundary is characterised by high-lying areas that act as a source of recharge.
Figure 1

Location of the study area and the Gangjeong–Koryeong Barrage.

Figure 1

Location of the study area and the Gangjeong–Koryeong Barrage.

The land surface elevation in the study area ranges from 20 to 25 m (above mean sea level, amsl) in the central flat area, and from 70 to 150 m (amsl) in the north-western high-lying area. Over 95 per cent of the central area is covered with greenhouses, which are used for melon cultivation from January to August. Water for irrigation is predominantly sourced from groundwater pumped from wells drilled close to each greenhouse. Melon cultivation increased strongly in the early 2000s to augment the income of growers and has now replaced rice-paddy cultivation.

The Gangjeong–Koryeong barrage was constructed 3 km downstream from the study area in the Nakdong River in 2012 (Figure 1). The height of the barrage is approximately 11.5 m, the length of the barrage crest is about 935.5 m, and the total capacity of the surface water in the barrage reservoir is about 92.3 million m3. The average river water level before barrage construction was approximately 15.0 m (amsl), but the water level is now controlled at 19.5 m (amsl) by the water gate of the barrage. The increase in the water level of 4.5 m is caused by the impoundment of the river.

The geological strata in the study area consist of Quaternary alluvium and bedrock (Figure 2). The bedrock is mostly comprised of Cretaceous–Jurassic sedimentary rocks, including shale, sandstone, conglomerate, and interbedded coal seams. The mean thickness of the Quaternary sediments is 7.48 ± 6.35 m. The sediments become deeper in the eastern part of the study area close to the Nakdong River. The bottom layer, 1–3 m thick, is composed of gravel and pebble, whereas the upper layer is mostly silty sand (Min & Chung 1985).
Figure 2

Geologic map of the study area.

Figure 2

Geologic map of the study area.

The alluvial deposits in the upstream valley of both streams are mainly gravel, pebble, and boulders, the latter having a maximum diameter of 40 cm. Smaller sediment particles are more frequently found closer to the eastern section of the main river flowing through the study area. Sand and gravel, including a small amount of pebbles, are the main components of the sediments in the alluvial fan region, which is in the western half of the central flat area (Figure 3). Silt and clay are mainly distributed in the lower sections of the two streams in the eastern part of the central flat area near the Nakdong River. This distribution of sediment produces large variations in the alluvial deposits in terms of their geohydrological properties, such as their hydraulic conductivity and layer distribution.
Figure 3

Sediments distribution along the two streams in the study area.

Figure 3

Sediments distribution along the two streams in the study area.

CHANGES IN WATER LEVELS

Manual measurements of groundwater levels and trend analysis

The study area is divided into two regions, a lower basin (Zone A) and an upper basin (Zone B), based on the topography, stratigraphic features, and groundwater levels (Figure 1). Zone A is a flat agricultural region where alluvial deposits are mainly fine sediments, such as clay, silt, and fine sand. Coarser sediments transported from upstream are deposited in Zone B, which is part of the alluvial fan. A total of 53 existing wells were selected for groundwater level analyses, with measurements conducted in March 2010, June 2011, October 2011, April 2012, and September 2012 (Figures 4 and 5). The reservoir behind the Gangjeong–Koryeong barrage was temporarily filled with water for several days in October 2011 for a test operation before a final completion of the barrage, and after March 2012, the barrage basin was permanently filled.
Figure 4

Locations of (a) groundwater wells used for manual measurements and (b) automatic monitoring wells.

Figure 4

Locations of (a) groundwater wells used for manual measurements and (b) automatic monitoring wells.

Figure 5

Distribution of groundwater levels for each measurement period.

Figure 5

Distribution of groundwater levels for each measurement period.

The average depth to groundwater was shallow in 2012 after the reservoir behind the barrage was filled, but some levels were still deep in April 2012 because water was pumped for irrigation during the growing season (Figures 5 and 6). Most greenhouses for melon cultivation need 5–15 m3/day water during the growing season. Therefore, if pumping wells are activated simultaneously at several greenhouses, a strong, continuous drawdown can be produced by the cumulative effects of pumping. Despite the effects of continuous groundwater pumping, the groundwater level is still high on average because of rising river water levels.
Figure 6

Box plots of groundwater levels measured in each period and zone (average values underlined). (a) Total; (b) March 2010; (c) June 2011; (d) October 2011; (e) April 2012; (f) September 2012.

Figure 6

Box plots of groundwater levels measured in each period and zone (average values underlined). (a) Total; (b) March 2010; (c) June 2011; (d) October 2011; (e) April 2012; (f) September 2012.

The trend in groundwater levels for Zones A and B were analysed in a series of box plots (Figure 6). The increasing trend in groundwater levels is more distinct in Zone A, which is close to the Nakdong River, than in Zone B. Over the period of measurement, the depth to groundwater in Zone A changed from 5.61 to 5.02 m, whereas in Zone B, the depth changed from 2.42 to 2.46 m. As Zone A is surrounded by two streams, which are directly connected to the Nakdong River, on two sides, the depth to groundwater can be more easily affected by the river fluctuation. In Zone A, the effect of groundwater pumping during the growing season from April to June was marked. The average depth to groundwater in this season (and its standard deviation) were larger than in Zone B because of the low hydraulic conductivity and yield capacity of the fine-material aquifer. The higher specific yield in Zone B explains why the groundwater levels were relatively stable, despite the large amount of pumping during the growing season. The high transmissivity in Zone B also allows rainwater to infiltrate easily into the soil, and therefore the depth to groundwater is relatively stable despite pumping.

Automatic measurements of groundwater levels and statistical analysis

Water levels in the Nakdong River are measured every 10 minutes at the Gangjeong gauging station located at the Gangjeong–Koryeong barrage, and groundwater levels are measured at six automatic monitoring wells (Figures 4 and 7). The river water levels before filling at the barrage fluctuated widely and were sensitive to rainfall, but after filling, they became more stable with the control of the barrage water gate.
Figure 7

Time series of river water levels and groundwater levels at six monitoring wells (bar: daily rainfall; red: groundwater levels; purple: river water levels). (a) W-1; (b) W-2; (c) W-3; (d) W-4; (e) W-5; (f) W-6. Please refer to the online version of this paper to see this figure in colour.

Figure 7

Time series of river water levels and groundwater levels at six monitoring wells (bar: daily rainfall; red: groundwater levels; purple: river water levels). (a) W-1; (b) W-2; (c) W-3; (d) W-4; (e) W-5; (f) W-6. Please refer to the online version of this paper to see this figure in colour.

Well W-1 is located far from the river and therefore shows no discernible correlation with the river water level. The groundwater levels at this well did not show any increasing response to the river level rise in March 2012. W-2 showed a rapid daily drawdown of groundwater during pumping from March to July and a moderate increase in groundwater levels during water-filling at the barrage. The correlation coefficient between the groundwater level at W-2 and the river water level was about 0.490, with a lag time of 4 days. This correlation suggests that a backwater effect from the barrage extends directly into the Baekcheon upstream region because there is no weir along this stream. There was no correlation between the groundwater levels and rainfall. W-3 is located in the lower basin with a stream on either side of it. The groundwater levels showed a weak response to fluctuations in the river water levels in March. Overall, the correlation between the groundwater levels and river levels was low and insignificant. This is also true of the correlation between the groundwater levels and rainfall. Well W-4 showed a rapid drawdown of groundwater levels in response to pumping activity from December to June, whereas a weak increase occurred in May when the reservoir behind the barrage was filled with water. W-5 is very close to the Nakdong River and therefore the correlation coefficient between the groundwater levels and river water levels was the highest in the study area, at 0.525. The response time for this well was only 2 days. Well W-6 is located in the centre of the basin, far from the Nakdong River. Therefore, the correlation between the groundwater levels and river water levels was low and the response time was relatively long. Three small weirs located on the Shincheon stream are used to store water for irrigation. These weirs control the water levels in the stream, and therefore the groundwater levels in the upper basin at W-1, W-4, and W-6 show little connectivity with the Nakdong River.

ESTIMATION OF GROUNDWATER LEVELS USING A NUMERICAL MODEL

Modelling configuration and approach

After construction of the barrage, some public complaints have been made regarding the inundation of low-lying land during very wet rainfall periods or during periods when there is no groundwater pumping. There are two possible causes for this inundation of the study area. First, the groundwater levels might rise abruptly with a rapid increase in the river water levels during heavy rainfall in a wet period, leading to the inundation of the low-lying areas. Second, the groundwater levels in the low-lying land could become high or start to overflow as a result of the high river water levels experienced during water filling at the barrage. This would typically occur during a dry season, to ensure an adequate supply of water, and would coincide with a period when pumping activity is low from October to April. These two scenarios were simulated using a numerical model, MODFLOW, to allow the prediction of the groundwater levels in the riversides. The hydrological conditions before the construction of the barrage were simulated as a steady-state, and the field survey data after water-filling were used to verify the accuracy of the numerical model by comparing the estimated and measured groundwater levels.

A digitised topographical map with a scale of 1:5000 was used to create a base map with the mapping software, Surfer (Figure 8). The model represents an area of 6.5 × 3.5 km using a grid size of 25 × 25 m, and two geological layers, an unconsolidated soil (Layer 1) and soft/hard rock (Layer 2). The depths of these layers were defined using existing drilling and geophysical survey data. The hydraulic conductivity of Layer 1 was determined using injection or pumping test data at 23 drilling sites, and ranged from 1.47 × 10–4 m/s to 2.33 × 10–6 m/s. Hydraulic conductivity for Layer 2 was also defined from the Lugeon test conducted for bedrock at 54 drilling sites and ranged from 2.31 × 10–6 to 8.44 × 10–8 m/s. A groundwater recharge rate of 128 mm was used, which is about 12.5% of the total rainfall (MLTM 2007).
Figure 8

(a) 3-D scheme of the modelled area and (b) the boundary of the area considered in modelling.

Figure 8

(a) 3-D scheme of the modelled area and (b) the boundary of the area considered in modelling.

Government regulations have set the representative monthly pumping rates for 17 categories of groundwater use based on monitoring data collected at 621 wells (MLTM 2007; Kim 2010, 2014, Kim et al. 2013). These representative pumping rates were assumed for the wells in the study area in the model configuration (Figure 9).
Figure 9

Monthly amounts of groundwater used for agricultural purposes in the study area.

Figure 9

Monthly amounts of groundwater used for agricultural purposes in the study area.

The numerical model was calibrated by trial and error to minimise selected measures of goodness of fit between the simulated and observed groundwater levels. The groundwater levels measured at 53 sites in October 2011 were used to calibrate the steady-state model. When the observed and simulated water levels were compared, the root mean squared error (RMSE) was approximately 1.6 m, and the normalised RMS was about 8.9%. The discrepancy between the inflow and outflow was only 0.02%.

Predicted groundwater levels during heavy rainfall

Short-term forecasts of groundwater levels during heavy rainfall events were made to predict the possibility of inundation in the low-elevation area. Numerical simulations were performed for heavy rainfall events occurring in June and July, 2012. Three main rainfall events occurred during this period: 30 mm on June 30; 43 mm from July 5 to July 7; and 141 mm from July 10 to July 19. The daily water levels of the Nakdong River, the Sincheon stream, and the Baekcheon stream were also measured and put into the model.

The results of the transient model for a short-term simulation of the groundwater levels during a heavy rainfall period from June 25 to July 24 are presented as a series of maps (Figure 10). The first rainfall of 30 mm on June 30 resulted in a slight increase in the groundwater levels in the eastern part of the study area (Figure 10(b)). The equipotential lines adjacent to the Baekcheon stream were parallel, indicating a linear distribution in this part of Zone A. The second rainfall of 43 mm also resulted in a slight increase in the groundwater levels in the vicinity of the Baekcheon stream, where the effects of the backwater from the Nakdong River are manifested (Figure 10(c)). The distance between the equipotential lines near the Baekcheon stream increased from July 10 to July 19 after the third rainfall, whereas the area surrounded by the 19.0 m (amsl) groundwater level decreased. The area surrounded by the 20.0 m (amsl) level increased after a heavy rainfall, when the water level in the Nakdong River rose to over 19.7 m (amsl) (Figure 10(d)). This short-term forecasting of groundwater levels indicates that the most rapid increase in groundwater levels during a rainy period occurs in the region closest to the Nakdong River. We found that the eastern area close to the two streams was inundated for three days because of a heavy rainfall of 40.5 mm, which occurred over two days in October 2013.
Figure 10

Map of the groundwater levels estimated with a numerical model for the period with heavy rainfall from June 25 to July 24, 2012. (a) After 1 day; (b) after 10 days; (c) after 20 days; (d) after 30 days.

Figure 10

Map of the groundwater levels estimated with a numerical model for the period with heavy rainfall from June 25 to July 24, 2012. (a) After 1 day; (b) after 10 days; (c) after 20 days; (d) after 30 days.

An increase in groundwater levels as the result of rainfall can reduce the infiltration because the soil moisture content is increased. When the groundwater levels reach the land surface, inundation of the low-lying land can occur. The short-term prediction of the groundwater levels before heavy rainfall offers an opportunity to lower river water levels and thus reduce flood damage to the land. Based on estimates of the groundwater levels and the reaction times to a reduction in the river water level, the water gate at the barrage could be opened to discharge the river water and lower the groundwater level at the riversides.

Groundwater levels and inundation potential during a dry season

The groundwater levels during a dry season on the riverside plains are mainly affected by surface water levels and groundwater pumping rates. The water level of the Nakdong River is kept constant during dry seasons by the management of the water gate at the barrage. Considering the interaction between the surface water and groundwater in the study area, the groundwater levels should also be kept constant, while avoiding heavy pumping, during this season. A rise in groundwater levels can result in the saturation of the soil and produce swampy conditions. A decreasing trend in groundwater levels during the irrigation season is followed by an increasing trend after August, when pumping activity is considerably reduced (Figure 7).

Table 1 shows the changes in the average daily water balance for the first dry season from Day 270 to Day 365 after the reservoir behind the barrage is filled with water. The total amount of water increased from 3262.3 to 4033.5 m3/d with leakage from the river, which resulted from an increase in the river water level. Groundwater use was considerably reduced because agricultural activity is reduced during this period and only a small amount of groundwater was pumped. The water inflow from the river through a leakage boundary increased from 431.3 m3/d before water filling to 3287.1 m3/d after water filling, and the water outflow into the river also changed from 2112.3 to 3010.4 m3/d. This indicates that the increased inflows and outflows are produced by an increase in the river water level. The change in groundwater quality after damming can also explain the inflow of surface water. Ion concentrations in groundwater samples from 11 wells were analysed for two cases, before and after damming, and the comparison explains some changes in groundwater quality. Before damming, the average contents of Ca2+, Mg2+, and Na+ were 69.46 mg/l, 17.08 mg/l, and 18.30 mg/l, respectively. However, after damming, they were 62.95 mg/l, 15.66 mg/l, and 15.48 mg/l, respectively. In contrast, the average contents of Cl and SO42– increased from 18.52 mg/l and 16.27 mg/l, respectively, to 20.08 mg/l and 17.83 mg/l, respectively. The reduction in cations and the increase in anions indicate that the groundwater inflow from the recharge area and the groundwater outflow into the river were blocked by the river when the river level was high after damming. This situation can result in a slow groundwater flow or alternatively a backward flow from the river, with a groundwater chemistry that is less characteristic of rock and more characteristic of river pollution.

Table 1

Changes in the average daily water balance before and after water-filling at the barrage (units: m3/day)

ConditionStorageConstant headWellsRiver leakageRechargeTotalIn minus out
Before water filling at the barrage In 0.0 0.0 0.0 431.3 2,831.0 3,262.3 −0.6 
Out 0.0 0.0 1,150.6 2,112.3 0.0 3,262.9  
After water filling at the barrage1 In 5.6 0.0 0.0 3,287.1 2,897.4 4,033.5 −0.6 
Out 342.7 0.0 681.1 3,010.4 0.0 4,034.1  
ConditionStorageConstant headWellsRiver leakageRechargeTotalIn minus out
Before water filling at the barrage In 0.0 0.0 0.0 431.3 2,831.0 3,262.3 −0.6 
Out 0.0 0.0 1,150.6 2,112.3 0.0 3,262.9  
After water filling at the barrage1 In 5.6 0.0 0.0 3,287.1 2,897.4 4,033.5 −0.6 
Out 342.7 0.0 681.1 3,010.4 0.0 4,034.1  

1Water balance for the first autumn season (September to November 2012).

A map of the distribution of groundwater levels estimated with the numerical model for the first autumn season after damming was drawn and compared with the actual groundwater levels (Figure 11, see also Figure 5). September 2012 marked the beginning of the first autumn season after the Nakdong River was dammed (Figure 5). The actual distribution of the depth to groundwater for this period was similar to that of the estimated groundwater levels. This similarity, reflected in the difference in groundwater levels between Zone A and Zone B and in the deepest and shallowest positions in each zone, indicates that forecasting with a numerical model effectively identifies areas with a high potential for groundwater flooding. A small difference existed between the estimated groundwater levels and the actual data near the boundary between the two zones. This difference probably originates from the uncertainty about the amount of pumping and the uneven hydraulic features of the aquifer. Given that swampy ground was observed near the Sincheon stream in the upper part of Zone B, this area shows a high potential for flooding during heavy rainfall or when the river water levels are high.
Figure 11

Map of the groundwater levels estimated with a numerical model for the first autumn season (September to November 2012).

Figure 11

Map of the groundwater levels estimated with a numerical model for the first autumn season (September to November 2012).

DISCUSSION AND CONCLUSION

The main aim of this study was to investigate how the increase in the river water levels after the construction of a barrage can affect groundwater levels in the riverside regions. Groundwater levels increased several days after water-filling at the Gangjeong–Koryeong barrage, with the average level rising by about 2–3 m. The greatest increase in water levels occurred in the eastern region of the study area, close to the Nakdong River, where the barrage is located. The stronger correlation between the groundwater levels and the river water levels in this region indicates that the river water levels contribute more to changes in groundwater levels than does rainfall.

The rate of increase in groundwater levels is relatively slow during a rainy period compared with that of river water levels. Similarly, high groundwater levels are also maintained for a longer period, and their rate of decline is slower. Therefore, a groundwater flooding event has a longer duration than a fluvial flooding event because of the interaction between the groundwater and surface water and the slow discharge capacity during heavy rainfall.

Groundwater levels can reflect the transmissivity of soil, land use type and extent, rainfall, nearby river water levels, groundwater pumping, and other natural or artificial factors. Multiple factors determine the groundwater dynamics during and after fluvial flood events, including the pre-event groundwater level, the pre-event soil moisture, groundwater recharge, the characteristics of the flood event itself, and the response of tributary streams.

In this study, two approaches to assessing groundwater flooding, during a heavy rainfall period and a dry season, were proposed and analysed. The analysis of groundwater flooding during heavy rain indicated that short-term inundation in a low-elevation area can occur through a combination of an increase in the groundwater level and a rapid increase in the river water level. However, the estimated response time to a change in the river water level and a significant correlation between the river water levels and groundwater levels suggest that the appropriate control of a water gate, to lower the river water levels at a barrage before heavy rainfall, can reduce the groundwater levels and minimise the risk of flooding.

Although high groundwater levels can be reduced by groundwater pumping at multiple wells during a dry season, they can still be high enough to produce swampy conditions in some low-lying areas. Therefore, we propose that a vulnerability map indicating the potential for groundwater flooding be developed, taking into account multiple site-specific factors. A discharge structure, such as an irrigation channel, should also be constructed to reduce the surface water level. Widening and deepening of the irrigation channels can increase the discharge rate of the surface water, and also reduce the groundwater levels and flood intensity. The introduction of a groundwater pumping system that operates in a systematic manner across multiple wells would be an effective approach to reducing groundwater levels in areas to which the drawdown effect of irrigation channels does not extend.

Measurements of groundwater levels, soil characteristics, land use, aquifer features, and topography could be used to develop a groundwater flood index and groundwater flood vulnerability map. Future research will be necessary to collect the relevant data and to develop a flood risk index and a groundwater flood hazard map in the study area. In practice, it will be difficult to take into consideration all the relevant factors in detail, because it would often be impossible to collect the necessary data sets required for complex flood risk assessment modelling tools. Statistical tools, such as a multiple linear regression model, could be useful in simplifying the factors involved. Flood risk assessment in South Korea must now consider groundwater levels after the construction of a large barrage. A full water balance, including the shallow and deep subsurface flow components and water flow exchanges between the groundwater and surface water, must be considered to achieve better results in flood risk assessment.

ACKNOWLEDGEMENTS

This work was supported by the research project ‘Advanced Technology for Groundwater Development and Application in Riversides (Geowater+)’ in the ‘Water Resources Management Program’ (code 11 Technology Innovation C05) of the Ministry of Land, Infrastructure and Transport, and the Korea Agency for Infrastructure Technology Advancement in Korea.

REFERENCES

REFERENCES
Anderson
M. P.
2005
Heat as a ground water tracer
.
Groundwater
43
(
6
),
951
968
.
Arumi
J. L.
Rivera
D.
Holzapfel
E.
Boochs
P.
Billib
M.
Fernald
A.
2009
Effect of the irrigation canal network on surface and groundwater interactions in the lower valley of the Cachapoal river, Chile
.
Chilean J. Agri. Res.
69
(
1
),
12
20
.
Constantz
J.
Stonestrom
D.
Stewart
A. E.
Niswonger
R.
Smith
T. R.
2001
Analysis of streambed temperatures in ephemeral channels to determine streamflow frequency and duration
.
Water Resour. Res.
37
,
317
328
.
Dreher
J.
1991
Groundwater management in the city of Vienna after construction of the new hydropower plant on the Danube River – a case study
. In:
Hydrological Basis of Ecologically Sound Management of Soil and Groundwater
(
Nachnebel
H. P.
Kovar
K.
, eds).
Proceedings of the Vienna Symp. IAHS Publ. No. 202
, pp.
229
238
.
Dreher
J.
Gunatilaka
A.
1996
Groundwater management in the city of Vienna. modelling, testing & monitoring for hydro power plants – II
. Int J Hydro Power Dams
2
,
545
554
.
Dreher
J.
Gunatilaka
A.
1998
Groundwater management system in Vienna – an evaluation after three years of operation
. In:
Artificial Recharge of Groundwater
(
Peters
J.
et al
, eds).
Balkema
,
Amsterdam
, pp.
167
172
.
Gunatilaka
A.
Dreher
J. E.
1996
Use of early warning systems as a tool for surface and groundwater quality monitoring
. In:
Proc. IAWQ and IWSA Symp. On Metropolitan Areas and Rivers 200–211
.
Rome
.
Harvey
F. E.
Lee
D. R.
Rudolph
D. L.
Frape
S. K.
1997
Locating groundwater discharge in large lakes using bottom sediment electrical conductivity mapping
.
Water Resour. Res.
33
,
2609
2616
.
Hassan
D. S.
2006
Review of Methods Used for Lowering Groundwater Levels at Archeological Sites, Egypt
.
Lund University
,
Sweden
.
Hughes
A. G.
Vounaki
T.
Peach
D. W.
Ireson
A. M.
Jackson
C. R.
Butler
A. P.
Bloomfield
J. P.
Finch
J.
Wheater
H. S.
2012
Flood risk from groundwater: examples from a chalk catchment in southern England
.
J. Flood Risk Manag.
4
(
3
),
143
155
.
Kim
G. B.
Cha
E. J.
Shin
K. H.
2014
Development and applicability of flood vulnerability index using groundwater levels
.
J. Geol. Soc. Korea
50
(
2
),
257
268
.
Kreibich
H.
Thieken
A. H.
Grunenberg
H.
Ullrich
K.
Sommer
T.
2009
Extent, perception and mitigation of damage due to high groundwater levels in the city of Dresden, Germany
.
Nat. Hazards Earth Syst. Sci.
9
,
1247
1258
.
LaBolle
E. M.
Ahmed
A. A.
Fogg
G. E.
2003
Review of the integrated groundwater and surface-water model (IGSM)
.
Groundwater
41
,
238
246
.
Macdonald
D. M. J.
Hall
R.
Carden
D.
Dixon
A.
Cheetham
M.
Cornick
S.
Clegg
M.
2007
Investigating the interdependencies between surface and groundwater in the Oxford area to help predict the timing and location of groundwater flooding and to optimize flood mitigation measures
. In:
Proceedings of 42nd DEFRA Flood and Coastal Management Conference
. pp.
04A.2.1
04A.2.12
.
MacDonald
A. M.
Hughes
A. G.
Vounaki
T.
Graham
M. T.
Lilly
A.
Mansour
M.
Stephens
C. A.
2008
Groundwater and the Forres (River Findhorn & Pilmuir) Flood Alleviation Schemes
.
British Geological Survey Commissioned Report CR/08/023
.
Nottingham
,
UK
.
Min
K. D.
Chung
C. D.
1985
Gravity survey of the subsurface structure between Waekwan-Pohang in Kyungsang basin
.
J. Korean Inst. Min. Geol.
18
(
4
),
321
329
.
MLTM (Ministry of Land, Transport, Maritime Affairs)
2007
Master Plan for Groundwater Management in Korea (2007–2011)
.
Seoul
,
Korea
.
MLTM (Ministry of Land, Transportation, Maritime Affairs)
2010
Plan for the Construction and Management of Groundwater Monitoring Wells at the Four Major Rivers Restoration Project Regions
.
Seoul
,
Korea
.
Paulsen
R. J.
Smith
C. F.
O'Rourke
D.
Wong
T.-F.
2001
Development and evaluation of an ultrasonic ground water seepage meter
.
Groundwater
39
,
904
911
.
Pennequin
D.
2010
Toward a water cycle approach for flood risk assessment
.
AQUAmundi
Am01002
,
7
12
.
Renken
R. A.
Dixon
J.
Koehmstedt
J.
Ishman
S.
Lietz
A. C.
Marella
R. L.
Telis
P.
Rogers
J.
Memberg
S.
2005
Impact of Anthropogenic Development on Coastal Ground-Water Hydrology in Southeastern Florida, 1900–2000, U.S. Geological Survey Greater Everglades Priority Ecosystems Science Program
.
US Geological Survey
,
Reston
.
Sommer
T.
Karpf
C.
Ettrich
N.
Haase
D.
Weichel
T.
Peetz
J. V.
Steckel
B.
Eulitz
K.
Ullrich
K.
2009
Coupled modeling of subsurface water flux for an integrated flood risk management
.
Nat. Hazards Earth Syst. Sci.
9
,
1277
1290
.
Sophocleous
M. A.
Townsend
M. A.
Bogler
L. D.
McClain
T. J.
Marks
E. T.
Coble
G. R.
1988
Experimental studies in stream-aquifer interaction along the Arkansas river in central Kansas: field testing and analysis
.
J. Hydrol.
98
,
249
273
.