KSCE Journal of Civil Engineering (2013) 17(1):233-243 DOI 10.1007/s12205-013-1609-x
Water Engineering
www.springer.com/12205
Spatially-Explicit Assessment of Flood Risk Caused by Climate Change in South Korea Il-Won Jung*, Heejun Chang**, and Deg-Hyo Bae*** Received June 20, 2011/Accepted May 17, 2012
···································································································································································································································
Abstract Identifying spatially-explicit risk is essential for efficient flood management with limited resources and budgets. We investigated national-scale relative flood risk for 139 sub-basins in South Korea under current and future climate conditions. A non-parametric index method was employed to calculate relative flood risk based on a sensitivity index, an exposure index, and an adaptive capacity index for each sub-basin. A dynamically downscaled climate simulation based on the A2 Greenhouse Gas (GHG) emission scenario, provided by the Korean Meteorological Research Institute (METRI), was used to examine possible change in flood risk. The estimated flood risk generally agreed with historical flood damage. The highly vulnerable sub-basins had high exposure indices, reflecting frequent heavy rainfall, as well as high sensitivity indices due to dense population at low elevation. Although rainfall intensity and frequency is likely to increase under the A2 GHG emission scenario, the spatial pattern of relative flood risk did not change remarkably. Our results indicate that reducing flood sensitivity levels and enhancing the adaptive capacity in vulnerable regions will be critical aspects of climate change preparation in South Korea. Keywords: flood, vulnerability, climate change, South Korea ···································································································································································································································
1. Introduction Changes in flood risk due to climate change will lead to a paradigm shift in flood management strategies in the future. Recent analyses reveal anthropogenic Greenhouse Gases (GHG) may increase the occurrence of extreme precipitation (Min et al., 2011) and thus flood risk (Pall et al., 2011). Additionally, several studies show that intensities of storms and tropical cyclones have increased since 1970 (e.g., Webster et al., 2005). This intensification of the water cycle caused by global warming could lead to an increasingly extreme climate, which would heighten the risk of flood occurrence (Huntington, 2006). Thus, numerous studies have examined possible changes in flood risk and associated uncertainties based on Global Climate Model (GCMs) simulations (e.g., Fowler and Hennessy, 1995; Tebaldi et al., 2006; Beniston et al., 2007; Cunderlik and Simonovic, 2005; Hamlet and Lettenmaier, 2007; Jung et al., 2011a). Spatially-explicit risk assessment is an effective method for identifying which regions are most vulnerable relative to others. In particular, national-scale risk analysis has the potential to provide powerful tools to decision-makers for developing targeted and synergistic mitigation strategies (Hall et al., 2005). Nationalscale risk analysis differs from flood mapping because it considers the impact of existing (or possible) infrastructure and measures
that reduce flood risk. Although a national-scale risk analysis requires a large amount of spatial data, Remote Sensing (RS) and Geographic Information System (GIS) technologies now facilitate the effective collection of such data sets. Based on such data, several investigations of national-scale flood risk have now been conducted. These studies help establish robust adaptation strategies in the face of global warming, land use change, urbanization, and population growth (e.g., England and Wales (Hall et al., 2005; Bradbrook et al., 2005), Czech Republic (Rodda, 2005), Australia (Merz et al., 2008), and the Netherlands (de Graaf et al., 2008)). These studies conclude that flood risk is likely to increase because of the combined forces of population growth and climate change. Coastal urban areas where the world’s major population is concentrated are particularly vulnerable due to sea level rise and increased storm runoff from inlands (Chang and Franczyk, 2008). Only a few studies have assessed and mapped flood risk in South Korea (e.g., Chang et al., 2009; Son et al., 2010; Lee et al., 2011). Chang et al. (2009) examined anthropogenic (land cover change) and natural causes of flood risks in Gangwon Province, located in Northeastern South Korea, using historical flood data from 1973-2005. This study emphasized that the flood vulnerability of the region will remain and is likely to increase under climate change unless adaptive flood management are imple-
*Post Doctoral Fellow, Institute for Sustainable Solutions, Dept. of Geography, Portland State University, OR 97201-0751, USA (Corresponding Author, E-mail:
[email protected]) **Professor, Institute for Sustainable Solutions, Dept. of Geography, Portland State University, OR 97201-0751, USA (E-mail:
[email protected]) ***Member, Professor, Dept. of Civil & Environmental Engineering, Sejong University, Seoul 143-747, Korea (E-mail:
[email protected]) − 233 −
Il-Won Jung, Heejun Chang, and Deg-Hyo Bae
mented. All previous Korean studies, however, focused on certain regions or watersheds. Therefore, identifying the relative vulnerability of various regions, which allows for prioritization of mitigation efforts, is still required to manage flood risk effectively. Additionally, a national-scale flood risk assessment under possible future climate conditions could improve our understanding of the linkage between human-induced global warming and regional flood risk. In South Korea, flooding causes more economic loss than any other natural hazard (NEMA, 2006). The National Emergency Management Agency (NEMA) (2006) reported that the average damage caused by natural hazards was approximately 2.4 billion USD a year for 1997-2006, and that flood damage related to typhoons and severe rainstorms was approximately 90% of total damage. To protect socio-economic property from floods, the Korean Ministry of Land, Transport and Maritime affairs (MLTM) has extensively employed structural approaches such as constructions of dams and embankments, river channel improvement, reinforcement of the flood control capacities of dams, and the use of flood forecasting and alert system (MLTM, 2008). Consequently, the annual loss of human life and inundation area has decreased since the 1980s. However, the cost of property damage is increasing by about 3.2 times per ten-year period (MLTM, 2008). Recent increases in flood damage in South Korea can be attributed to changes in both natural and anthropogenic systems. Increases in the frequency and intensity of heavy rainfall (e.g. Choi, 2002; Chang and Kwon, 2007; Jung et al., 2011b) are primary natural stressors. Significant increasing trends in heavy rainfall since the 1970s (over 80 mm/day) have been observed. Anthropogenic factors such as urbanization and industrialization near flood plains, which increase danger to people, property, and infrastructure, are also responsible for increasing flood damage (Chang et al., 2009). Thus, the national-scale flood risk assessment should include both natural factors and socio-economic factors to obtain reliable results (Chang and Franczyk, 2008). We investigate spatially-explicit flood risk in South Korea using a non-parametric index method that calculates relative flood risk based on three indices: sensitivity, exposure, and adaptive capacity. The estimated flood risk is compared with historical flood damage to assess the applicability of the methodology. Changes in flood risk for two future time-slices, 2011-2040 and 2051-2080, are projected based on a dynamically downscaled climate change simulation. This study does not consider future changes in socio-economic factors. We discuss the caveats of the study in Section 5.2.
2. South Korea and Floods Located in Northeastern Asia, South Korea is composed of five large river basins (see Fig. 1). The topography of the western and southern regions is relatively flat and low in elevation, and the eastern region is steeper, with higher elevations. Major cities are located at the outlets of the five large river basins (Fig. 1). In
Fig. 1. Main Rivers, Major Cities, and Five Large Basins in South Korea
2000, approximately 65% of the land cover in Korea was forest and 20% was agricultural land (WAMIS, 2011). More than 70% of the people live in urban areas, but urban land use makes up only 4% of South Korea’s total area. The population of Korea was 47.28 million in 2005 (KOSIS, 2011). More than 52% of the population lives in the Han River Basin, which includes the Seoul metropolitan area; 27% live in the Nakdong River Basin; 12% in the Gum River Basin; and less than 10% reside in the Youngsan River Basin and the Sumjin River Basin, respectively. South Korea is highly vulnerable to flood damages. The climate of South Korea has strong seasonality; over 70% of annual precipitation occurs during the summer monsoon season (June to September) (e.g., Jung et al., 2011b). Therefore, South Korea experiences flooding in summer and drought in spring. Most rivers in South Korea have short channel length and steep channel slope due to mountainous terrain, and low soil moisture holding capacity due to shallow soil (Bae et al., 2008a, 2008b). These natural conditions result in the occurrence of high peak flows, because rainfall often flows over the surface, thus rapidly entering rivers, rather than being infiltrated into the soil zone (MLTM, 2008). Recent intense and frequent rainfall events (e.g. Chang and Kwon, 2007; Jung et al., 2011b) contributed to increased flood damage. In the 2000s, South Korea suffered massive flood damage from typhoon Rusa (2002, 6,930 million USD damage), typhoon Maemi (2003, 4,880 million USD), and typhoon Ewiniar (2006, 1,942 million USD). These are the most damaging floods between 1916 and 2006 (NEMA, 2006).
− 234 −
KSCE Journal of Civil Engineering
Spatially-Explicit Assessment of Flood Risk Caused by Climate Change in South Korea
3. Methodology This study analyzes flood risk for the 139 sub-basins delineated in the Korean Water resources Unit map (KWU) (WAMIS, 2011); 47 are located in the Han River basin, 33 in the Nakdong River basin, 27 in the Gum River basin, 14 in the Sumjin River basin, and 14 in the Youngsan River basin. We first calculate three indices using their associated proxy variables (see Fig. 2). The proxy variables are collected and interpolated into the subbasin mapping units according to the area ratio when the extent (or resolution) of proxy variables mismatches sub-basin boundaries (Jung, 2008). We estimate flood risk and compare it with historical flood damage. To examine climate change impacts, we then calculate the relative change in proxy variables (frequency and intensity of heavy rainfall) using future climate simulation in the near future (2011-2040) and distant future (2050-2080) relative to the reference (1971-2000) period. Finally, we analyze the possible changes in flood risk according to the relative change in proxy variables. 3.1 Proxy Variables for Sensitivity, Exposure, and Adaptive Capacity The flood risk of a certain region can be determined by the magnitude and frequency of its hydrologic extremes and by its vulnerability level. Thus, assessing vulnerability is a key aspect of risk analysis. Vulnerability assessments using proxy variables have been widely applied in previous studies on risk and natural hazards (e.g., Hurd et al., 1999; Jung, 2008; Lee et al., 2011). Vulnerability can be defined as the combination of exposure, sensitivity, and adaptive capacity (or resilience) in diverse fields (e.g., McCarthy et al., 2001; Turner et al., 2003; Gallopin, 2006; Young et al., 2006; Chang et al., in press). Exposure represents the degree, duration, and extent of a system’s contact with external perturbations (Gallopin, 2006). The sensitivity is the degree to which a system is affected by external forcing. The adaptive capacity is the coping or response capacity of a system to the external perturbation (McCarthy et al., 2001). Based on these definitions, this study assumes that exposure to flood risk is influenced by external factors such as frequency and intensity of rainfall extremes (Jung, 2008). Two rainfall indices,
the number of days of daily rainfall above 80 mm (PN80) and a maximum daily rainfall (PX1D), are commonly employed as the criteria of frequency and intensity of extremes, respectively (e.g. Choi, 2002; Chang and Kwon, 2007; Jung et al., 2011b). If a certain region has higher PN80 and PX1D values than others, this region has higher exposure to flood risk than others. For the sensitivity proxy variables, we selected geographical and socio-economic factors, including mean elevation of each sub-basin (ELEV), population density (POP), and declared land value (DLV) in 2000. ELEV accounts for the fact that lowelevation regions are generally more vulnerable to flooding than upland regions. Regions near the coast will also be affected by sea level rise in a changing climate. Due to the limited scope of this research, sea level rise effects are not considered here. POP and the DLV in flood zone strongly relate to flooding damage to property, and POP could represent possible loss of human life. DLV captures the different property values of metropolitan, agricultural, and forested areas, even though these are located in similar elevations. Adaptive capacity is closely related to technological resources, infrastructures, and institutions, which mitigate flood damage and recover flood loss. We used three proxy variables for representing adaptive capacity; the PCI (% channel improvement), the FWPC (Flood Water Pump Capacity), and the DFCC (dam flood control capacity). The PCI includes channel alignments, which allow water to flow away quickly, constructed river banks, and riverside parks and picnic areas, which are safely inundated during flooding, thus decreasing flooding downstream. FWPC indicates comprehensive flood control systems to protect communities, businesses and cities from major flood disasters. DFCC measures the flood control capacity of sub-basins located in both downstream and upstream regions, including multi-purpose dams. We consider that flood control capacity decreases as the distance from a dam to a downstream sub-basin increases. While a flood forecasting and a flood alarm system are operating in South Korea, they are not considered in this study because they are difficult to quantify. Table 1 presents the index and proxy variables used in this study. Table 1. Description of Flood Risk Index and Associated Proxy Variables Proxy variables PN80 Exposure PX1D ELEV Sensitivity POP DLV Index
PCI Adaptive capacity Fig. 2. Overview of the National Flood Risk Assessment Methodology used in This Study Vol. 17, No. 1 / January 2013
− 235 −
FWPC DFCC
Description Number of days of precipitation above 80 mm/day Maximum one-day precipitation (mm) Mean elevation of the watershed (m) Population density (persons/km2) Declared land value per unit area (USD/m2) Completed channel improvements divided by total channel improvement required for the flood protection within sub-basins (%) Flood water pumping capacity within sub-basins (m3/minute) Flood control capacity of multi-purpose dams (106 m3)
Il-Won Jung, Heejun Chang, and Deg-Hyo Bae
To estimate the eight proxy variables, we collected climatic, geographic, and socio-economic data from the Water Management Information System (WAMIS) website (http://www.wamis. go.kr/) maintained by the MLTM. WAMIS provides extensive data (hydro-climatic data, Digital Elevation Model (DEM), PCI, and dam information, etc.) relevant to Korean water resources. Since the early 2000s, the MLTM has conducted a comprehensive basin investigation project for the five large river basins of South Korea. They have collected scattered water-related data and interpolated unevenly-spaced data using the KWU (WAMIS, 2011). The PN80 and PX1D are estimated using daily precipitation data for each sub-basin and averaged for 1971-2000. ELEV is calculated by averaging elevation within each sub-basin using a 30m DEM. POP and DLV data were acquired from the Korea National Statistical Office (KNSO, 2011) and the annual report on declared land value was obtained from the MLTM (http://www.mltm.go.kr). For POP and DLV, we used data from 2000. 3.2 Non-parametric Flood Risk Assessment To reliably analyze flood risk, it is essential to conceptualize the dynamic interaction between endogenous (e.g., geophysical regimes) and exogenous processes (e.g., global warming, institutional and technological factors) (Young et al., 2006). A conceptual approach is easier than an empirical or a physical approach to address these complex interactions between socio-economic conditions and geophysical conditions (e.g., Turner et al., 2003; Adger, 2006). To quantify relative flood risk (Rf), we employed a nonparametric, conceptual approach (Jung, 2008). Eq. (1) expresses that when the sensitivity index (Is) and exposure index (Ie) are higher and when the adaptation index (Ia) is lower, the flood risk index has a larger value. The index combines proxy variables that represent quantifiable factors such as population, elevation, infrastructure, assets, and so on. These factors must be normalized because they have different units and dimensions. We normalize the factors (P) of each region to a range of 0 to 1 using Eq. (2). The normalized factors (NP) are merged to calculate the index of each region using Eq. (3). A weighting coefficient (α) for proxy variables was assumed as 1.0 in this study. Before calculating the relative flood risk, the indexes of exposure, sensitivity, and adaptive capacity are normalized again. Finally, the relative flood risk is determined using Eq. (1). Is × Ie Rf = ----------Ia Pi, j, k – Pmin, j, k NPi, j, k = ---------------------------------Pmax, j, k – Pmin, j, k
(1) (2)
n
I i, k =
∑ αj, k × NPi, j, k
(3)
j=1
where P is a proxy variable, i is each sub-basin, j is a proxy variable, k is an index, n is the number of proxy variables, Pmin is the minimum value of proxy variables, and Pmax is the maximum value of proxy variables.
3.3 Climate Change Simulation High resolution data is essential for spatially-explicit analysis because the complex mountainous terrain of South Korea induces spatial and temporal variation in precipitation (Im et al., 2010, 2011). High resolution climate simulations may capture possible spatial variation of PN80 and PX1D. Dynamical downscaling with a Regional Climate Models (RCM) is a useful way to generate high-resolution climate simulations from coarse GCM simulations (Bae et al., 2008b). The Korean Meteorological Research Institute (METRI) disaggregated the simulation of the ECHO-G with the SRES A2 GHG emission scenario (CO2 concentrations about 890ppm by 2100) using the Fifth-Generation NCAR/Penn State Mesoscale Model (MM5) at 27 km nested grid spacing (Boo et al., 2006). We determine the relative change in PN80 and PX1D using downscaled daily precipitation for 2011-2040 and 2051-2080 relative to 19712000. The relative changes are multiplied by the estimated PN80 and PX1D to project the future exposure index. Climate impact studies always involve uncertainties arising from the projection of future GHG emissions, different GCM and RCM structures, and different downscaling methods. This study does not investigate these uncertainties because of the small number of RCM simulations. Thus, our results would likely be different with other climate simulations. We discuss the limitations of the current approach in detail later.
4. Results 4.1 Patterns in the Proxy Variables The estimated proxy variables vary spatially. PN80 shows the largest value (above 2 days per year) in downstream regions of the Han River and southern coastal regions, and exhibits the lowest value (below 1.0 days per year) in the upper and middle sections of the Nakdong River. PX1D shows a similar pattern to PN80 (see Fig. 3). POP and DLV exhibit larger values in major cities and urbanizing regions than in rural and mountainous regions. Since channel improvements have been concentrated in urbanized regions, PCI displays a spatial pattern similar to those of POP and DLV. FWPC is remarkably centralized in highly urbanized regions because flood water pumping systems are a critical component of highly developed sewer systems (see Fig. 3). This study investigates relations between historical flood damage and proxy variables (see Fig. 4). FWFC, flood water pumping capacity, has high positive correlation with historical flood damage. This may be because the Korean MLTM has preferentially constructed flood water pumping systems in areas that were frequently damaged. Most proxy variables show positive relations to flood damage except for ELEV and DFCC. DFCC has no significant relation to flood damage (99% confidence level) because only a small proportion of sub-basins is influenced by multi-purpose dams (see Fig. 3). Although most proxy variables show significant correlations with historical flood damage, each proxy variable exhibited large scatter in
− 236 −
KSCE Journal of Civil Engineering
Spatially-Explicit Assessment of Flood Risk Caused by Climate Change in South Korea
Fig. 3. Spatial Distribution of Proxy Variables for Each Sub-basin (Gray lines indicate sub-basin boundaries and the bold black lines indicate large river basins.)
Fig. 4. The Relationship of Flood Damage Costs for 1971-2000 and Proxy Variables (r is the correlation coefficient. The flood damage and proxy variables were normalized to a range of 0 to 1 using Eq. (2).)
relation to the flood damage. This may mean that a combination of multiple sources of risk drives regional flood damage, Vol. 17, No. 1 / January 2013
indicating that these sources could be included in a flood risk analysis of South Korea. A multiple regression model using all
− 237 −
Il-Won Jung, Heejun Chang, and Deg-Hyo Bae
Table 2.Multiple Regression Analysis between Normalized Historical Flood Damage and Proxy Variables (An asterisk (*) indicates significance at 99% confidence level.) Model Sensitivity only Exposure only Adaptive capacity only Sensitivity + Exposure Sensitivity + Adaptive capacity Exposure + Adaptive capacity Sensitivity + Exposure + Adaptive capacity
Equation Damage = -0.31ELEV+0.46POP-0.04DLV Damage = 0.26PN80+0.20PX1D Damage = 0.20PCI+0.60FWPC-0.12DFCC Damage = 0.18PN80+0.14PX1D-0.20ELEV+0.41POP+0.08DLV Damage = -0.27ELEV-0.18POP+0.03DLV+0.13PCI+0.71FWPC-0.13DFCC Damage = 0.35PN80+0.05PX1D+0.17PCI+ 0.57FWPC-0.17DFCC Damage = 0.23PN80+0.14PX1D-0.14ELEV-0.29POP-0.08DLV+0.13PCI +0.76FWPC-0.17DFCC
proxy variables can explain historical flood damage more accurately than ones using fewer proxy variables (see Table 2). An interesting finding is that sensitivity alone and adaptive capacity alone can each explain the variation of flood damage much better than exposure alone. This can be interpreted as an indication that the management of existing sensitivity and adaptive capacity is important in mitigating flood risk. 4.2 The Relative Flood Risk Assessment for the Current Situation This study calculates sensitivity, exposure, and adaptive capacity indices using Eqs. (2) and (3) for 139 sub-basins (see Fig. 5). The exposure index has the highest values in the western Han River basin and the southern coastal regions because of intense and frequent heavy rainfall events. Heavy rainfall (above 80 mm/day) in the western Han River Basin has increased significantly since the 1970s (e.g., Jung et al., 2011b). The exposure index in the Northern and middle Nakdong River Basin shows the lowest values. These regions have the lowest annual precipitation in South Korea. In general, the degree of exposure is in accordance with the distribution of annual precipitation. The sensitivity index shows relatively low values in mountainous regions and higher values in urbanized regions because the urbanized regions have dense population, expensive declared
R2 0.40** 0.21** 0.45** 0.48** 0.51* 0.59** 0.63**
land value, and are generally low in elevation. The highest adaptive capacity indices are detected near major cities, where intensive flood management infrastructure has been installed because these regions also have high sensitivity to flood risk. This is especially true in the sub-basins located in the metropolitan city, Seoul, which show the largest index values for both adaptive capacity and sensitivity. The estimated regional flood risk derived from Eq. (1) shows that the highly vulnerable regions (above 0.6) were located in the downstream regions of the large river basins (see Fig. 6), because most of them have lower adaptive capacity and higher sensitivity than others. The proportion of relatively vulnerable sub-basins that have a flood risk value above 0.4 (moderate) and 0.6 (high) are 37% and 15%, respectively (see Table 3). The ratio of very low vulnerability sub-basins (below 0.2) is approximately 25%. The high vulnerability sub-basins are characterized by both high exposure indices (PN80, PX1D) and high sensitivity indices (POP, ELEV). By contrast, the very low vulnerability sub-basins have low exposure indices and low sensitivity indices, but also have low adaptive capacity indices. The very high vulnerability sub-basins (above 0.8) had the highest exposure indices and the lowest adaptive capacity indices. The spatial pattern of the calculated relative flood risk shows significant correlation to historical flood damage at the 99%
Fig. 5. Flood Exposure, Sensitivity, and Adaptive Capacity for Each Sub-basin − 238 −
KSCE Journal of Civil Engineering
Spatially-Explicit Assessment of Flood Risk Caused by Climate Change in South Korea
Fig. 7. Change in Annual Precipitation and Annual Mean Temperature Relative to 1971-2000 under the A2 GHG Emission Scenario over South Korea
Fig. 6. Relative Flood Risk (Left) and Historical Flood Damage (Right) for the Reference Period, 1971-2000 Table 3. Mean Values for Proxy Variables According to Relative Flood Risk Relative flood risk Very low Low Moderate High Very high < 0.2 0.2~0.4 0.4~0.6 0.6~0.8 0.8 < Number of 35 52 31 12 9 sub-basins (25%) (37%) (22%) (9%) (6%) PN80 0.97 1.57 1.91 2.26 2.71 PX1D (mm/day) 95.3 109.7 124.1 135.5 141.7 ELEV (m) 427 181 155 93 162 POP (persons/km2) 129.8 304.6 1002.6 1525.5 216.7 DLV (USD/m2) 1.127 1.450 2.049 1.500 1.418 PCI (%) 66.8 74.0 74.9 71.6 44.0 FWPC (m3/minutes) 1.55 3.70 11.18 9.41 0.38 DFCC (106 m3) 0.03 0.29 0.36 0.20 0.03 Proxy variables
confidence interval (Pearson’s correlation coefficient, r = 0.53). However, some regions with historically enormous flood damage, especially Daejeon and Gwanju, were estimated as less vulnerable regions. One reason is that these historically vulnerable regions now have high adaptive capacities, even though they have high sensitivity index. Since the 1980s, the MLTM has concentrated on installing countermeasures (e.g., channel improvement, construction of pumping system) in these regions. 4.3 The Relative Flood Risk Assessment for Future Scenario The dynamically downscaled climate simulation projects that both annual precipitation and temperature will increase by up to +20% and +6oC in the 2090s in Korea under the A2 GHG emission scenario (see Fig. 7). Annual mean temperature increases gradually, but change in annual precipitation shows a different direction before the 2040s. The mean change in temperature and precipitation shows +2°C (+5°C), +0.3% (+13%) for 2011-2040 (2051-2080). PN80 and PX1D are projected to increase under climate change in most sub-basins in South Korea (see Fig. 8). Mean Vol. 17, No. 1 / January 2013
Fig. 8. Relative Change of Flood Sensitivity Index for 2011-2040 (Upper Panel) and 2051-2080 (Lower Panel) Relative to the Reference Period
PN80 increases by about +8.6% for 2011-2040, and +19.0% for 2051-2080. Also, the PX1D increases by about +6.6% for 20112040, and +12.2% for 2051-2080. These results indicate that the rainfall intensity and frequency are likely to increase more dramatically in the distant future than in the near future. In the flood risk analysis under the climate simulation, the number of high vulnerability sub-basins (above 0.6) increases slightly (see Table 4). In addition, due to spatially varying changes in PN80 and PX1D, some sub-basins of the low group move to
− 239 −
Il-Won Jung, Heejun Chang, and Deg-Hyo Bae
Table 4. Changes of Mean PN80 and PX1D According to Relative Flood Risk for the 2040s and the 2080s under the A2 Scenario Period
2040s
2080s
Proxy variables Number of sub-basins PN80 PX1D (mm/day) ELEV (m) POP (persons/km2) DLV (USD/m2) PCI (%) FWPC (m3/minutes) DFCC (106 m3) Number of sub-basins PN80 PX1D (mm/day) ELEV (m) POP (persons/km2) DLV (USD/m2) PCI (%) FWPC (m3/minutes) DFCC (106 m3)
Very low < 0.2 41 (29%) 1.09 99.9 369 173.2 1.188 68.2 2.09 0.05 39 (28%) 1.19 107.6 397 119.4 1.091 68.5 2.02 0.08
Low 0.2~0.4 39 (28%) 1.75 118.3 203 202.3 1.336 74.3 3.93 0.22 43 (30%) 1.79 117.9 178 315.1 1.396 74.3 3.81 0.14
Relative flood risk Moderate 0.4~0.6 38 (27%) 2.10 131.2 153 531.5 1.639 74.0 3.34 0.31 33 (24%) 2.36 141.5 166 712.4 1.775 75.4 5.88 0.33
High 0.6~0.8 16 (12%) 2.63 154.8 130 2919.3 2.341 80.6 27.70 0.59 16 (12%) 2.73 160.5 122 1832.4 2.229 76.8 17.4 0.65
Very high 0.8 < 5 (4%) 2.27 130.2 149 179.5 1.361 23.4 0.00 0.01 8 (6%) 3.00 150.4 145 216.5 1.318 41.3 0.38 0.01
the moderate group at the middle portion of the Han River basin and to the very low group in the lower portion of the Youngsan River Basin. However, as shown in Fig. 9, the spatial distribution of future flood risk for 2011-2040 (r = 0.87), and 2051-2080 (r = 0.91) does not change in comparison to reference flood risk (see Fig. 9 upper panel), though flood risk changes for both periods (see Fig. 9 lower panel).
5. Discussion
Fig. 9. Relative Flood Risk (Upper Panel) and Change in Flood Risk (Lower Panel) for 2011-2040 and for 2051-2080 from the Reference Period
5.1 Flood Management Strategies in Changing Climate Our results show that the flood risk is not evenly distributed over South Korea, suggesting that a spatially-explicit approach can be useful in developing flood risk mitigation strategies. The highly vulnerable sub-basins were characterized by high exposure indices due to frequent heavy rainfall (high PX1D and PN80), and high sensitivity indices, indicating dense population (POP) and low elevation (ELEV) (see Tables 3 and 4). The future relative regional flood risk did not change very much, although the PX1D and PN80 are projected to increase under the A2 GHG scenario. This indicates that the strategies of reducing the sensitivity level and enhancing the adaptive capacity in more vulnerable regions are important for flood risk management in a changing climate. To reduce sensitivity to flood risk, people and property (e.g., houses, schools, hospitals, etc.) within or near floodplains should be relocated to safer places. Locating property outside the floodplain is a prime way to reduce flood risk (see Table 2). However, it is difficult to persuade people and industry to move properties located in high-risk areas to safer places, as people and industry are not typically willing to give up their current benefits in major
− 240 −
KSCE Journal of Civil Engineering
Spatially-Explicit Assessment of Flood Risk Caused by Climate Change in South Korea
urban areas. Hence, to reduce flood risk in those areas, adaptive capacity should be enhanced. There are two ways to increase adaptive capacity; structural measures and non-structural flood management measures. Traditionally, Korean central and provincial governments have emphasized structural measures such as flood pumps, dike reinforcement, river channel regulation and improvement, multi-purpose dam construction, and construction of facilities for flood detention and storage in upstream areas (Chang et al., 2009). However, recent flood damages in South Korea caused by typhoons and severe rainstorms showed that structural flood risk management was not sufficient to prevent flood damage in South Korea (MLTM, 2008). Non-structural measures, such as flood forecasting and alert systems, efficient flood control operation and management techniques, and the formulation and modification of laws and regulations for allocating flood volume, are currently under way in Korea (Chang et al., 2009). This national-scale flood risk assessment also contributes to flood risk management policy and investment priorities. This assessment can be routinely updated when new evidence becomes available or when climatic simulations improve (Fig. 2). 5.2 Uncertainty in Flood Risk Assessment This study has two major sources of uncertainty. One is the uncertainty of future changes in extreme rainfall as projected by GCMs and GHG emission scenarios. In general, the impacts of future changes in extreme rainfall may be underestimated because GCMs often underestimate the observed increase in heavy rainfall with warming (e.g., Min et al., 2011; Lee et al., 2012; Jung et al., 2012). Therefore, our results may also underestimate the changes in PN80 and PX1D. Additionally, an analysis using different GCMs and GHG emission scenarios could be different from our result. However, some previous studies (e.g. Boo et al., 2006; Im et al., 2010, 2011) consistently demonstrated increases in the number of days of heavy rainfall as well as mean precipitation amount over South Korea, although they used different RCMs and GHG emission scenarios (e.g., A2 and B2 scenarios). Jung et al. (2012) also reported a possible increase in wet season flow based on seasonal runoff projections using 39 climate simulations. These results indicate that flood risk is likely to increase in South Korea under climate change. To clarify this problem, further research will need to be conducted based on multiple or ensemble climate scenarios (e.g., Wilby and Harris, 2006; Jung et al., 2011a; Bae et al., 2011). The second major source of uncertainty arises from the assumption that socio-economic conditions will not change in future. Because flood damage is closely related to human life loss and property, damage to social infrastructure, and hydro-ecological resources, credible future socio-economic conditions are important for projecting future risk. However, projecting accurate future socio-economic conditions relevant to flood risk is very difficult. One alternative option is to use possible socio-economic scenarios such as GHG emission scenarios. These scenarios include future Korean land use and land cover change, populaVol. 17, No. 1 / January 2013
tion growth, urban planning, and long-term flood-related water management plans. Consequently, developing national socioeconomic scenarios will be an essential part of improving our forecasting capacity and our understanding of the flood risk change caused by human-induced climate change (e.g., Hall et al., 2005; Chang and Franczyk, 2008; Jung et al., 2011a).
6. Conclusions We examine national-scale flood risk in current and possible future periods using a dynamically downscaled climate simulation. A conceptual, non-parametric index method is employed to identify the relative degree of exposure, sensitivity, and adaptive capacity to flood risk for 139 sub-basins in South Korea. The estimated flood risk agrees well with historical flood damage for 1970-2000, suggesting that our method is useful for spatially-explicit flood risk analysis in South Korea. The highly vulnerable regions are characterized by high exposure indices and high sensitivity indices. The major metropolitan areas had both higher sensitivity and adaptive capacity. Future rainfall is projected to increase both in frequency (PN80) and in intensity (PX1D). Although PN80 and PX1D show increases over South Korea, spatial patterns of relative flood risk do not change much. This study shows that reducing the sensitivity level and enhancing the adaptive capacity of vulnerable regions will be a critical aspect of adaptive preparation for climate change and climate variability.
Acknowledgements This work was funded by the Korea Meteorological Administration Research and Development Program under Grant CATER 2012-3100. Additional support was provided by a grant from the Institute for Sustainable Solutions (ISS) at Portland State University. We appreciate Dr. Won-Tae Kwon and Dr. Kyung-On Boo of the Korean National Institute of Meteorological Research for providing the statistically downscaled climate simulation. We thank Madeline Steele of Portland State University for proofreading the manuscript and for providing helpful comments.
References Adger, W. N. (2006). “Vulnerability.” Global Environ. Change, Vol. 16, No. 3, pp. 268-281. Bae, D. H., Jung, I. W., and Chang, H. (2008a). “Long-term trend of precipitation and runoff in Korean river basins.” Hydrol. Process., Vol. 22, pp. 2644-2656. Bae, D. H., Jung, I. W., and Chang, H. (2008b). “Potential changes in Korean water resources estimated by a high resolution climate simulation.” Clim. Res., Vol. 35, No. 3, pp. 213-226. Bae, D. H., Jung, I. W., and Lettenmaier, D. P. (2011). “Hydrologic uncertainties in climate change from IPCC AR4 GCM simulations of the Chungju basin, Korea.” J. Hydrol., Vol. 401, Issues 1-2, pp. 90-105. Beniston, M., Stephenson, D. B., Christensen, O. B., Ferro, C. A. T.,
− 241 −
Il-Won Jung, Heejun Chang, and Deg-Hyo Bae
Frei, C., Goyette, S., Halsnaes, K., Holt, T., Jylhä, K., Koffi, B., Palutikof, J., Schöll, R., Semmler, T., and Woth, K. (2007). “Future extreme events in european climate: And exploration of regional climate model projections.” Climatic Change, Vol. 81, No. 1, pp. 7195. Boo, K. O., Kwon, W. T., and Baek, H. J. (2006). “Change of extreme events of temperature and precipitation over Korea using regional projection of future climate change.” Geophys. Res. Lett., Vol. 33, No. L01701, DOI: 10.1029/2005GL023378. Bradbrook, K., Waller, S., and Morris, D. (2005). “National floodplain mapping: Datasets and methods - 160,000 km in 12 months.” Nat. Hazards, Vol. 36, Nos. 1-2, pp. 103-123. Chang, H. and Franczyk, J. (2008). “Climate change, land use change, and floods: Toward an integrated assessment.” Geogr. Compass, Vol. 5, No. 2, pp. 1549-1579. Chang, H., Franczyk, J., and Kim, C. (2009). “What is responsible for increasing flood risks? The case of Gangwon Province, Korea.” Nat. Hazards, Vol. 48, No. 3, pp. 339-354. Chang, H., Jung, I. W., Strecker, A., Wise, D., Lafrenz, M., Shandas, V., Moradkhani, H., Yeakley A., Pan, Y., Bean, R., Johnson, G., and Psaris, M. “Water supply, demand, and quality indicators of the spatial distribution of water resources vulnerability in the Columbia River basin, USA.” Atmosphere-Ocean (In press). Chang, H. and Kwon, W. T. (2007). “Spatial variations of summer precipitation trends in South Korea, 1973-2005.” Environ. Res. Lett., Vol. 2, Nos. 1-9, DOI: 10.1088/1748-9326/2/4/045012. Choi, Y. (2002). “Trends in daily precipitation events and their extremes in the southern region of Korea.” Korean Soc. Environ. Impact Asses., Vo. 11, pp. 189-203. Chung, Y. S., Yoon, M. B., and Kim, H. S. (2004). “On climate variations and changes observed in South Korea.” Climatic Change, Vol. 66, Nos. 1-2, pp. 151-161. Cunderlik, J. M. and Simonovic, S. P. (2005). “Hydrological extremes in a southwestern Ontario river basin under future climate conditions.” Hydrolog. Sci. J., Vol. 50, No. 4, pp. 631-654. de Graaf, R. E., van de Giesen, N., and van de Ven, F. (2008). “Alternative water management options to reduce vulnerability for climate change in the Netherlands.” Nat. Hazards, Vol. 51, No. 3, pp. 407-422. Fowler, A. M. and Hennessy, K. J. (1995). “Potential impacts of Global Warming on the frequency and magnitude of heavy precipitation.” Nat. Hazards, Vol. 11, No. 3, pp. 282-303. Gallopin, G. C. (2006). “Linkages between vulnerability, resilience, and adaptive capacity.” Global Environ. Change, Vol. 16, No. 3, pp. 293303. Hall, J. W., Sayers, P. B., and Dawson, R. J. (2005). “National-scale assessment of current and future flood risk in England and Wales.” Nat. Hazards, Vol. 36, Nos. 1-2, pp. 147-164. Hamlet, A. F. and Lettenmaier, D. P. (2007). “Effects of 20th Century warming and climate variability on flood risk in the Western U.S.” Water Resour. Res., Vol. 43, No. W06427, doi:10.1029/2006WR00 5099. Huntington, T. G. (2006). “Evidence for intensification of the global water cycle: review and synthesis.” J. Hydrol., Vol. 319, pp. 83-95. Hurd, B. H., Leary, N., Jones, R., and Smith, J. B. (1999). “Relative regional vulnerability of water resources to climate change.” J. Am. Water Resour. As., AWRA, Vol. 35, No. 6, pp. 1399-1410. Im, E. S., Jung, I. W., and Bae, D. H. (2011). “The temporal and spatial structure of trends in extreme indices over Korea simulated by regional climate model: Validation (1971-2000) and projection
(2000-2100).” Int. J. Climatol., Vol. 31, No. 1, pp. 72-86. Im, E. S., Jung, I. W., Chang, H., Bae, D. H., and Kwon, W. T. (2010). “Hydroclimatological response to dynamically downscaled climate change simulations for Korean basins.” Climatic Change, Vol. 100, No. 3, pp. 485-508. Intergovernmental Panel on Climate Change (IPCC) (2007). Climate change 2007: The physical science basis, Contribution of WG I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, Available at http://ipcc-wg1.ucar.edu/wg1/wg1-report.html (Assessed June 06, 2011) Jung, I. W. (2008). Climate change impact assessments on Korean water resources using hybrid downscaling technique, Doctoral Dissertation, Sejong University, Seoul, Korea. Jung, I. W., Bae, D. H., and Kim, G. (2011b). “Recent trends of mean and extreme precipitation over South Korea.” Int. J. Climatol., Vol. 31, No. 3, pp. 359-370. Jung, I. W., Bae, D. H., and Lee, B. J. (2012). “Possible change in Korean streamflow seasonality based on multi-model climate projections.” Hydrol. Process., DOI: 10.1002/hyp.9215. Jung, I. W., Chang, H., and Moradkhani, H. (2011a). “Quantifying uncertainty in urban flooding analysis considering hydro-climatic projection and urban development effects.” Hydrol. Earth Syst. Sc., Vol. 15, pp. 617-633. Kay, A. L., Davies, H. N., Bell, V. A., and Jones, R. G. (2009). “Comparison of uncertainty sources for climate change impact: Flood frequency in England.” Climatic Change, Vol. 92, Nos. 1-2, pp. 4163. Korea National Statistical Office (KNSO) (2011). Available at http:// www.nso.go.kr. (Assessed June 6, 2011) Korean Statistical Information Service (KOSIS) (2011). Available at http://www.kosis.kr/. (Assessed June 6, 2011) Lee, J. J., Kwon, H. H., Kim, T. W. (2012). “Spatio-temporal analysis of extreme precipitation regimes across South Korea and its application to regionalization.” J. Hydro-environment Res., KWRA, Vol. 6, pp. 101-110. Lee, M. H., Jung, I. W., and Bae, D. H. (2011). “Korean flood vulnerability assessment on climate change.” J. Korea Water Resour. As., KWRA, Vol. 44, No. 8, pp. 653-666. McCarthy, J. J., Canziani, O. F., Leary, N. A., Dokken, D. J., and White, K. S. (eds.) (2001). Climate change 2001: Impacts, adaptation, and vulnerability, United Nations Intergovernmental Panel on Climate Change. Available at: www.grida.no/climate/ipcc_tar/wg2/001.htm. (Assessed June 6, 2011). Merz, R., Blöschl, G., and Humer, G. (2008). “National flood discharge mapping in Austria.” Nat. Hazards, Vol. 46, No. 1, pp. 53-72. Min, S. K., Zhang, X., Zwiers, F. W., and Hegerl, G. (2011). “Human contribution to more-intense precipitation extremes.” Nature, Vol. 470, No. 7334, pp. 378-381. Ministry of Land, Transport and Maritime Affairs (MLTM) (2008). Water resources in Korea 2007, Water Resources Bureau of MLTM, Edited by Korea Institute of Construction Technology, Geonet, Seoul, Available at http://www.adb.org/Documents/Books/AWDO/ 2007/br04.pdf (Assessed June 6, 2011). National Emergency Management Agency (NEMA) (2006). Annual disaster reports 2006. Pall, P., Aina, T., Stone, D. A., Stott, P. A., Nozawa, T., Hilberts, A. G. J., Lohmann, D., and Allen, M. R. (2011). “Anthropogenic greenhouse gas contribution to flood risk in England and Wales in autumn 2000.” Nature, Vol. 470, No. 7334, pp. 382-385.
− 242 −
KSCE Journal of Civil Engineering
Spatially-Explicit Assessment of Flood Risk Caused by Climate Change in South Korea
Rodda, H. J. E. (2005). “The development and application of a flood risk model for the Czech Republic.” Nat. Hazards, Vol. 36, Nos. 1-2, pp. 207-220. Son, K. H., Lee, B. J., and Bae, D. H. (2010). “Assessment on flood characteristics changes using multi-GCMs climate scenario.” J. Korea Water Resour. As., KWRA, Vol. 43, No. 9, pp. 789-799. Tebaldi, C., Hayhoe, K., Arblaster, J. M., and Meehl, G. A. (2006) “Going to extremes, an intercomparison of model simulated historical and future changes in extreme events.” Climatic Change, Vol. 79, Nos. 3-4, pp.185-211. Turner, B. L., Kasperson, R. E., Matson, P. A., McCarthy, J., Corell, R. W., Christensen, L., Ecklely, N., Kasperson, J. X., Luers, A., Martello, M. L., Polsky, C., Pulsipher, A., and Schiller, A. (2003). “A framework for vulnerability analysis in sustainability science.” P. Natl. Acad. Sci. USA., Vol. 100, No. 14, pp. 8074-8079. Usul, N. and Turan, B. (2006). “Flood forecasting and analysis within
Vol. 17, No. 1 / January 2013
the Ulus Basin, Turkey, using geographic information systems.” Nat. Hazards, Vol. 39, No. 2, pp. 213-229. WAter resources Management Information System (WAMIS) (2011). Available at http://www.wamis.go.kr/. (Assessed June 6, 2011). Webster, P. J., Holland, G. J., Curry, J. A., and Chang, H. R. (2005). “Changes in tropical cyclone number, duration, and intensity in a warming environment.” Science, Vol. 309, No. 5742, pp. 18441846. Wilby, R. L. and Harris, I. (2006). “A framework for assessing uncertainties in climate change impacts: Low-flow scenarios for the River Thames, UK.” Water Resour. Res., Vol. 42, No. W02419, doi:10. 1029/2005WR004065. Young, O. R., Berkhout, F., Gallopin, G. C., Janssen, M. A., Ostrom, E., and van der Leeuw, S. (2006). “The globalization of socio-ecological systems: An agenda for scientific research.” Global Environ. Change, Vol. 16, No. 3, pp. 304-316.
− 243 −