Journal of Hydrology 399 (2011) 263–273

Contents lists available at ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier .com/ locate / jhydrol

Soil erosion risk in Korean watersheds, assessed using the revised universalsoil loss equation

Soyoung Park a, Cheyoung Oh a, Seongwoo Jeon b, Huicheul Jung b, Chuluong Choi a,⇑a Dept. of Geoinformatic Engineering, Pukyung National University, 599-1 Daeyeon 3-Dong, Nam-Gu, Busan 608-737, South Koreab Korea Adaptation Center for Climate Change, Korea Environment Institute, 290 Jinheung-Ro, Eunpyong-Gu, Seoul 122-706, South Korea

a r t i c l e i n f o s u m m a r y

Article history:Received 20 August 2010Received in revised form 11 December 2010Accepted 9 January 2011Available online 22 January 2011

This manuscript was handled by G. Syme,Editor-in-Chief

Keywords:Analytic hierarchy processEnvironmental Conservation ValueAssessment MapFrequency ratioLogistic regressionSoil erosion

0022-1694/$ - see front matter Crown Copyright � 2doi:10.1016/j.jhydrol.2011.01.004

⇑ Corresponding author. Tel.: +82 10 7747 6272; faE-mail addresses: 100yac@hanmail.net (S. Park)

swjeon@kei.re.kr (S. Jeon), hchjung@kei.re.kr (H. Jung)

Soil erosion reduces crop productivity and water storage capacity, and, both directly and indirectly,causes water pollution. Loss of soil has become a problem worldwide, and as concerns about the environ-ment grow, active research has begun regarding soil erosion and soil-preservation policies. This studyanalyzed the amount of soil loss in South Korea over a recent 20-year period and estimated future soilloss in 2020 using the revised universal soil loss equation (RUSLE). Digital elevation (DEM) data, detailedsoil maps, and land cover maps were used as primary data, and geographic information system (GIS) andremote sensing (RS) techniques were applied to produce thematic maps, based on RUSLE factors. Usingthe frequency ratio (FR), analytic hierarchy process (AHP), and logistic regression (LR) approaches, landsuitability index (LSI) maps were developed for 2020, considering the already established EnvironmentalConservation Value Assessment Map (ECVAM) for Korea. Assuming a similar urban growth trend and 10-,50-, and 100-year rainfall frequencies, soil loss in 2020 was predicted by analyzing changes in the cover-management factor and rainfall–runoff erosivity factor. In the period 1985–2005, soil loss showed anincreasing trend, from 17.1 Mg/ha in 1985 to 17.4 Mg/ha in 1995, and to 20.0 Mg/ha in 2005; the 2005value represents a 2.8 Mg/ha (16.6%) increase, compared with 1985 and is attributable to the increasedarea of grassland and bare land. In 2020, the estimated soil loss, considering the ECVAM, was 19.2–19.3 Mg/ha for the 10-year rainfall frequency, 36.4–36.6 Mg/ha for the 50-year rainfall frequency, and45.7–46.0 Mg/ha for the 100-year rainfall frequency. Without considering the ECVAM, the amount of soilloss was about 0.4–1.6 Mg/ha larger than estimates that did consider the ECVAM; specifically, the valueswere 19.6–19.9 Mg/ha for the 10-year rainfall frequency, 37.1–37.8 Mg/ha for the 50-year frequency, and46.7–47.5 Mg/ha for the 100-year frequency. In 2010, without considering the ECVAM, the soil loss was0.3–1.8 Mg/ha more than that estimated when considering the ECVAM. These results indicate that ifurban areas are developed such that they damage areas of high value, as defined environmentally andlegislatively, the amount of soil loss will increase, whereas if such areas are preserved, erosion willdecrease slightly. Thus, when planning urban development, the environmental and legislative value ofpreservation should be considered to minimize erosion and allow for more sustainable development.

Crown Copyright � 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction

In recent decades, soil erosion by water has become a world-wide issue, with climate change and progressive declines in theratio of natural resources to human populations. Moreover, variouspractices expose soils to greater risks of erosion, including inappro-priate agricultural practices, deforestation, overgrazing, forest fires,and construction activities (Terranova et al., 2009). Soil erosion hasnegative impacts on ecology and can lead to reduced crop produc-tivity, worsened water quality, lower effective reservoir water

011 Published by Elsevier B.V. All

x: +82 51 629 6653., leeieel@nate.com (C. Oh),, cuchoi@pknu.ac.kr (C. Choi).

levels, flooding, and habitat destruction (Oh and Jung, 2005).Concern for the environment has also increased worldwide and,thus, various studies have examined soil conservation. In particu-lar, the need for environmentally sensitive development alterna-tives in watersheds with multiple usage pressures and the needto forecast erosion and minimize the environmental impacts ofdevelopment have been noted (KMOE, 2001).

To calculate the risk of soil erosion and identify ways to controlerosion, quantitative analysis is required to calculate how quicklysoil erodes in its natural state. However, precise predictions are dif-ficult because soil loss is influenced by complex factors, such as soilconditions, surface cover, and environmental factors. Thus, variousforecasting formulae based on statistical and basic approacheshave been developed (Oh and Jung, 2005).

rights reserved.

264 S. Park et al. / Journal of Hydrology 399 (2011) 263–273

Techniques for predicting soil erosion are classified as physical,analog, and digital types, with the digital type further subclassifiedinto physically based, stochastic, and empirical types. Most modelsfor calculating the amount of soil loss belong to the ‘‘gray-box’’type of the empirical approach. These models select only the mostimportant factors related to soil erosion and predict the amount ofsoil erosion using statistical techniques with materials observedand calculated in the field and laboratory. Recently, there has beentremendous effort to develop a ‘‘white-box’’ or physically-basedmodel after issues regarding a mechanical understanding of theerosion process were raised (National Institute for Disaster Preven-tion, 1998).

Major empirical models include the Pacific Southwest Inter-agency Committee (PSIAC, 1968), universal soil loss equation(USLE) (Wischmeier and Smith, 1978), soil loss estimation forSouthern Africa (SLEMA) (Elwell and Stocking, 1982), and Morganand Finney methods (Morgan et al., 1984). Among these models,the USLE was the first and most important empirical model; itwas developed based on thousands of experimental data pointscollected by the Soil Conservation Service and the Agricultural Re-search Service in 37 US states. A revision of the USLE model, calledRUSLE (Renard et al., 1997), has been applied to erosion over ex-tended areas and in different contexts, including forests, rangeland,and disturbed areas (Terranova et al., 2009).

Physically-based models, based on physics, include the arealnon-point source watershed environment response simulation(ANSWERS) (Beasley et al., 1980), Water Erosion Prediction Project(WEPP) model (Nearing et al., 1989), chemical, runoff, and erosionfor agricultural management system (CREAMS) (Knisel, 1980), andthe European soil erosion model (EuroSEM) (Morgan et al., 1990).

Because most models deal with many variables displaying greatspatial and temporal variability, the use of remote sensing and geo-graphical information system (GIS) techniques makes soil erosionestimation and its spatial distribution feasible, with reasonablecosts and accuracy over large areas (Millward and Mersey, 1999;Wang et al., 2003).

The target area of this study was the Korean peninsula, locatedat the far eastern part of the Asian continent. As South Korea hasundergone rapid industrialization and urbanization since the1960s, the population has become intensively concentrated in ur-ban areas, as shown by the increase in the urbanization ratio from39.1% in the 1960s to 90.1% in 2005. Urbanization has been accom-panied by reckless large-scale residential land development, withcontinuous destruction of agricultural and forest lands. Lookingat trends in land use for individual categories surveyed by StatisticsKorea, the area of rice paddy decreased from 7596 km2 in 1980 to1421 km2 in 2005. In the same period, the area of farmland was re-duced by 570 km2 from 12,152 km2, and the forest area was re-duced by 1324–64,805 km2 (Statistics Korea Homepage, http://www.index.go.kr). In addition, extraordinary weather conditionshave brought unusually heavy local rain. For example, Inje-gunin Gangwon Province recorded a maximum daily rainfall of350 mm during Typhoon Ewiniar in 2006; this amount corre-sponds to a rainfall event of 80–500 year frequency and was thelargest daily rainfall recorded in Korea since meteorological obser-vation began (Lee et al., 2009). Thus, both urbanization andextraordinary weather conditions in South Korea are acceleratingthe risk of soil erosion.

A few foreign researchers have estimated soil loss in Korea.Walling (1983), for example, estimated soil loss of 500–750 ton/km2/year, while Lvoivich et al. (1991) proposed a range of 1000–5000 ton/km2/year. However, observational data were lacking.Additionally, domestic researchers, such as Park (2003), Kimet al. (2007), Oh and Jung (2005), Lee et al. (2006), and Lee andHwang (2006) calculated the amount of soil loss using GIS andRUSLE and analyzed the distribution characteristics by rating each

category according to the risk of soil erosion. Most studies havetargeted a specific watershed because of spatial and temporal lim-itations of data collection and analysis. Recently, thanks to Kimet al.’s (2009) research, maps showing the distribution and riskclass of soil loss have been completed. However, time-series anal-yses of soil loss and future forecasting have not yet beencompleted.

Thus, this study quantitatively analyzed time-series changes insoil loss over the last 20 years in Korea. Compared with natural soilerosion, soil loss due to human activities, such as agricultural, ur-ban, and road development, may be increased by a few to a fewmillion times, even under the same rainfall conditions (Goldmanet al., 1986). This study analyzed the amount of urban growthand forecasted the amount of 2020 soil loss, due to changes in rain-fall. As primary data, this study used digital elevation model (30-mDEM) data from the Korea National Geographic InformationInstitute, 1:25,000 detailed soil maps from the Korea NationalInstitute of Agricultural Science and Technology, and 1:25,000 landcover maps from the Korea Ministry of Environment. These de-tailed data, combined with the use of GIS and remote sensing(RS) analysis techniques, allowed for an accurate analysis.

For the analysis of urban growth, a 1:25,000 land suitability in-dex (LSI) map was completed, considering the Environmental Con-servation Value Assessment Map (EVCAM) for Korea by modelingthe relationship between drivers and urban growth, using the fre-quency ratio (FR), logistic regression (LR), and analytic hierarchyprocess (AHP) methods. The ECVAM shows the environmentaland legislative regions of the country, ranked by grades of 1–5.An LSI map is a probability map that analyzes urban growth thatpreserves the environment as well as urban growth that focuseson development, rather than the environment. An LSI map is a cov-er-management factor in RUSLE. Rainfall frequencies for 10, 50,and 100 years were applied for the rainfall–runoff erosivity factor.On the basis of these factors, the amount of soil loss was predictedand the differences with and without considering the ECVAM wereanalyzed.

This quantitative analysis and forecast of soil loss will contrib-ute to efforts to minimize the environmental impacts of develop-ment and prevent natural disaster due to soil loss. Moreover, theresults can serve as primary data for planning short- and long-termpolicies and in developing methods to preserve and manage soilresources.

2. Methodology

2.1. Study area

The study area was all land areas of South Korea (hereafter,Korea), except some smaller islands, such as Jeju Island, UlleungIsland, and Dokdo. The target area lies between 34�1804200N and37�2204300N and 124�1903000E and 130�5203100E, covering100560.87 km2 (Fig. 1). The research period focused on 1985 and2005.

Korea has a continental monsoon climate. Within Korea’s rela-tively small land area, climate conditions vary widely from northto south and from east to west. Seasonal variations are also dis-tinct. Summer is hot and humid with severe rain storms, and win-ter is cold and very dry. The annual average temperature (basedon the average temperature between 1971 and 2000) ranges fromapproximately 10 to 16 �C except in the central mountainousareas. The hottest month, August, has an average temperature of27 �C, while in the coldest month, January, temperatures can fallto �7 �C. In addition, the annual average precipitation (based onthe average precipitation between 1971 and 2000) is 1000–1400 mm in the central region, 1000–1800 mm in the southern

Fig. 1. Study area.

S. Park et al. / Journal of Hydrology 399 (2011) 263–273 265

region. Of total annual precipitation, 50–60% falls in the summerseason (Korea Meteorological Administration Homepage, http://www.kma.go.kr). The National Institute of Meteorological Re-search (2009) analyzed potential weather changes in the Koreanpeninsula for the A1B greenhouse gas concentration scenario;the analysis examined 27-km-resolution data obtained for theglobe by ECHO-G with 20C3M (historical 20th century climate)specifications and integration data obtained for the A1B climatechange scenario using the weather model MM5 for the East Asianregion. The report, comparing weather at the end of the 21st cen-tury (2071–2100) to weather at the end of the 20th century(1971–2000), suggests that temperature will increase by 4 �C(3.8 �C in the continental region of South Korea) and precipitationwill increase by 17% (13% in the continental region of SouthKorea).

With respect to geologic and soil characteristics, the Koreanpeninsula contains various rocks from the Archaeozoic era to theCenozoic era. These include metamorphic rocks of the Archaeozoicand Proterozoic eras in the Gyeonggi and Yeongnam massifs,metasedimentary rocks and lime rocks of the Paleozoic era in theOgcheon belt, and sedimentary rocks of the Mesozoic era in theGyeongsang basin. The Korean soil has high silicic acid and lowcation content, creating barren conditions, and receives more pre-cipitation than evaporation loss. Thus, acid conditions exist inKorean soil due to the loss of soil components. In addition, the hightemperature and humidity in summer result in fast decompositionby microorganisms so that the soil has a low level of organicmatter.

With regard to geographic features, over 65% of the land ismountainous. In east-tilting mountains (e.g., the great Baekdumountain range), the eastern slopes are steep, whereas westernslopes are gentle. The southern region is composed of gentle topog-raphy. Within mountainous areas, only 2.4% of the land has less

Table 1River basins of South Korea.

Basin Major river Length (k

Han River Bukhan, Namhan, Anseong River, Eastern coast 481.7Nakdong River Nakdong, Taehwa, Hyungsan 521.5Keum River Keum, Mankyung, Dongjin, Sapgyo River 395.9Yeongsan River Yeongsan, Seomjin, Tamjin 138

a Major water system refers to the major rivers in a particular region, such as the Ha

than 15% gradient. These areas have a low fertility level and arethus limited in terms of lumber production and water conversa-tion. Most farming in mountainous areas takes place on erodedplains (KMOE, 2009).

The watersheds of Korea are divided into four major basins,named after the four main rivers: the Han River, Nakdong River,Keum River, and Yeongsan River. The total basin area is99,133 km2, and 65 different national rivers flow in these riverbasins, with 55 local rivers at level 1 and 3773 rivers at level 2.In terms of river length, the Nakdong River is the longest at521.5 km. By basin area, the Han River Basin covers 32,300 km2,followed by the Nakdong River Basin at 23,800 km2, the KeumRiver Basin at 17,767 km2, and the Yeongsan River Basin at16,886 km2 (Table 1).

2.2. Revised universal soil loss equation (RUSLE)

The universal soil loss equation (USLE) model was proposed byWischmeier and Smith (1965), based on the breakaway and trans-port of soil particles due to rainfall, to calculate the amount of soilloss in agricultural regions. However, while the departure processof particles may be the main erosion mechanism in areas of short,gentle slopes, the model may be less useful in actual basins wherediverse, complex environments exist. To address this problem, Re-nard et al. (1997) suggested a new revised universal soil loss equa-tion (RUSLE) with modified climate factors, based on thedevelopment of erosion factors, due to seasonal changes, modifica-tion of slope-length and slope factors, and creation of a new pro-cess to calculate vegetation-cover factors.

The RUSLE model can be applied to a wide range of areas usingsimulated and observed rainfall and runoff data and erosion the-ory. The RUSLE equation model, structured from the USLE model,is as follows:

m) Area (major water systema) (km2) Local level-II rivers (number)

32,200 (26,018) 77632,280 (23,817) 85317,767 (9810) 71716,886 (3371) 479

n River, Nakdong River, Deum River, and Yeongsan River.

266 S. Park et al. / Journal of Hydrology 399 (2011) 263–273

A ¼ R� K � LS� C � P ð1Þ

where A is the mean soil loss per year (Mg/ha/year), R is the rain-fall–runoff erosivity factor (MJ mm ha�1 h�1 year�1), K is the soilerodibility factor (Mg h MJ�1 mm�1), LS is the slope-length andsteepness factor, C is the cover-management factor, and P is thesupport-practice factor.

2.2.1. Rainfall–runoff erosivity factorThe rainfall–runoff erosivity factor R is determined by the strike

energy of rain, kinetic energy of rainfall, and maximum rainfallintensity. Rainfall is measured at numerous rain gauge stationsthroughout Korea, including 77 run by the Korea MeteorologicalAdministration, 429 by the Ministry of Land, Transportation andMaritime Affairs, nine by the Rural Development Administration,and 163 by the Korea Water Resource Corporation. Among these,rain gauge data from the Korea Meteorological Administration sta-tions undergo a verification process and are used to predict climatechange in Korea. Therefore, to calculate the R factor for each fre-quency, precipitation data over the last 30 years (1977–2006) werecollected from each of the 58 gauging stations of the Korea Mete-orological Administration, from which the maximum precipitationdata from 1 to 24 h were derived. For rainfall intensity over 30 min,frequency analysis of rainfall data (FARD), developed by NationalInstitute for Disaster Prevention using the hourly distribution coef-ficient of local storm designs (Ministry of Land, Transportation andMaritime Affairs, 2000), was used.

For the distribution type used in the frequency analysis, 12methods including Normal, Gamma2, and Log-Normal were ap-plied. Through Kolmogorov–Smirnov (K–S), x2, and Cramer-vonMises (CVM) tests, goodness of fit was determined for the fre-quency analysis of all data from the meteorological stations from

Fig. 2. Distribution of the RUSLE factors for 1985, 1995, and 2005. (A) Rainfall erosivity fcover management and support-practice factor, ‘CP’, for 1985, (E) ‘CP’ for 1995, and (F)

2-year to 200-year frequencies through the bivariate logarithmicnormal distribution. On the basis of these data, this study calcu-lated the R factors for each frequency (Kim et al., 2009; Korea Re-mote Sensing Center Homepage, http://krsc.kict.re.kr).

2.2.2. Soil erodibility factorThe soil erosion factor K indicates the resistance of soil to ero-

sion by rainfall and is related to the soil particles, the distribution,structure, and organic matter content, and the size of voids. Typi-cally, the K value ranges from 0.13 to 0.91 Mg/ha/R; the K factoris smaller in soils with large amounts of very fine sand and silt,high permeability, and high organic matter content (Brady andWeil, 1996). In this study, K values calculated by the Ministry ofConstruction, Korea Institute of Construction Technology (1992),were input as values for a detailed numerical soil map (1:25,000)and thematic maps of soil factors, as in Fig. 2.

2.2.3. Slope-length and steepness factorThe geographical factor LS indicates the impact of length and

slope of terrain on soil erosion and has a large value if the lengthand slope of terrain are large. The slope-length is measured fromwhere water starts to flow to where sedimentation begins. The gra-dient is the average slope gradient of the surface and is expressedas a percentage of the height difference over horizontal distance.To calculate the LS factor, this study used a DEM with a 30-m gridand Remortel et al.’s (2001) LS factor calculation program, whichwas developed in ARC Macro language (AML) by ArcInfo.

2.2.4. Cover-management factorThe vegetation-cover factor C is a quantitative indicator of

the extent to which vegetation prevents erosion. The value of C

actor ‘R’, (B) soil erodibility factor ‘K’, (C) slope-length and steepness factor, ‘LS’, (D)‘CP’ for 2005.

Table 2The cover-management factor and support-practice factor for land cover items.

C factor (Shin, 1999) P factor

Urban and built-up 0.1 1.0Farmland 0.322 0.5Cropland 0.1 (0.2–0.3) 0.5Paddy land 0.5 (0.3–0.4) 0.5Forest 0.001 (0–0.1) 1.0Grassland 0.2 (0.1–0.2) 1.0Wetland 0.05 1.0Bare land 0.35 1.0Water 0.01 1.0

Table 3The change of area and annual growth rate in urban.

Urban area(km2)

Annual growthrate (%)

Annual growthrate (%)

1985 1850 85–90 6.8 85–05 6.01990 2571 90–95 4.6 85–00 5.21995 3227 95–00 4.1 90–00 4.42000 3941 00–05 8.6 90–05 5.72005 5941 85–95 5.7 95–05 6.3

S. Park et al. / Journal of Hydrology 399 (2011) 263–273 267

depends on the size of cover plants controlling the surface erosion,the state of the surface area, the plant roots, the surface roughness,and the amount of contained water. High C values, nearing 1, occuron bare land with little vegetation, while low values, of less than0.1, are found in areas of dense forest or grain cover (Lee et al.,2008; Kim et al., 2007). The C value was calculated based on valuessuggested by Shin (1999) and Dawen et al. (2006), with farmlanddivided into cropland and paddy fields. However, this study alsoused land cover maps from 1985 to 1995, which did not distin-guish cropland and paddy fields. Based on the estimated percent-ages of cropland and paddy fields, the C value was calculated as0.322 (Table 2).

2.2.5. Soil preservation factorThe soil preservation factor P is defined as the ratio of soil loss

from an upward and downward slope of an inclined plane where asoil preservation policy has been put in place (Kim et al., 2009).This factor was suggested by Wischmeier and Smith (1978) andused to evaluate effect of contours, target planting, bench work,sub-surface drainage, and lighting in dry farmland. The P factor isinfluenced by various types of farmland and slope, but this studyonly considered P values due to types of land cover, using the val-ues suggested by Dawen et al. (2006).

2.3. Producing the land suitability map

This study used the FR, LR, and AHP approaches to model therelationships of each driver for generating LSI maps. Here, the driv-ers are topographic (elevation, slope, aspect), geographic (land use,distance from road, road ratio, distance from urban areas), and so-cial (grade of environmental value, grade of legal restriction zones)factors.

The FR is based on the observed spatial relationship betweenthe location of urban growth and each urban growth-related factor.This relationship was used to determine the rating of each factor inthe overlay analysis (Lee and Talib, 2005). The correlation analysisinvolved dividing the scope of individual factors or the dimensionratio of urban occurrence between types by the occupancy ratio ofeach factor grade for the whole dimension (Lee et al., 2000).

The AHP is a kind of multi-criteria decision-making (MCDM) ap-proach that involves pair-wise comparisons of decision variables(e.g., objectives, alternatives). In constructing the pair-wise com-parison matrix, this study used ‘‘B,’’ a factor-wise weight, calcu-lated through LR in SPSS 13 after standardization of factormeasurements (Wu, 2002). Next, the matrices were normalizedby adding the column elements and dividing each element bythe respective weights. The rows of the normalized matrix ofobjectives were averaged to yield the relative weights (Bantayanand Bishop, 1998; Dai et al., 2001).

McFadden (1973) developed LR, a type of multivariate analysismodel. The key to LR is that the dependent variable is dichoto-mous. The independent variables are predictors of the dependent

variable and can be measured on a nominal, ordinal, interval, or ra-tio scale. The relationship between the dependent variable andindependent variables is nonlinear (Yesilnacar and Topal, 2005).Quantitatively, the relationship between an occurrence and itsdependency on several variables can be expressed as

p ¼ 11þ e�z

ð2Þ

where p is the suitability of urban growth occurring and z repre-sents the linear combination:

z ¼ b0 þ b1x1 þ b2x2 þ � � � þ bnxn ð3Þ

in which b0 is the intercept of the model, bi(i = 0, 1, 2, . . . , n) repre-sents the coefficients of the logistic regression model, and xi(i = 0, 1,2, . . . , n) represents the independent variables (Lee and Sambath,2006).

2.4. Setting scenarios for soil erosion assessment in 2020

The amount of soil loss increased over the 20 years from 1985to 2005 (Table 5); based on this information, the followingassumptions were made for forecasting the trend up to 2020:

(1) The urban trend is similar to that for 1985–2005 and rainfallhas 10-year frequency.

(2) The urban trend is similar to that for 1985–2005 and rainfallhas 50-year frequency.

(3) The urban trend is similar to that for 1985–2005 and rainfallhas 100-year frequency.

The trend for 1985–2005 refers to the following (Table 3). Overthe 20 years, the area of urban land use tripled, from 1850 km2 in1985 to 5941 km2 in 2005. With regard to the annual averagegrowth rate in urbanized areas, the average growth rates from2000 to 2005, 1985 to 1990, and 1995 to 2005 were relatively high(over 6.0%), while rates in other periods were 6.0% or less. To avoidover- or under-growth of the urbanized areas in the future, themaximal average growth rate from 2000 to 2005 (8.6%) and theminimal average growth rate from 1995 to 2000 were excludedin calculating the average. This adjusted average growth rate wasapplied. As a consequence, this study assumed the average growthrate in the urban areas to be 5.6% in 2020.

Thus, assuming for the C factor that urban areas grow at 5.6% ormore annually on average, the LSI map was drawn, depending onconsideration of the ECVAM, using FR, AHP, and LR to calculatethe areas for each land cover item. In addition, the R factor used10-year, 50-year, and 100-year rainfall frequencies.

3. Results

3.1. Soil erosion change detection over 1985–2005

The distribution of soil loss per unit over the 20 years from 1985to 2005 is shown in Table 4 and Fig. 3. Soil loss amounted to17.1 Mg/ha in 1985, 17.4 Mg/ha in 1995, and 20.0 Mg/ha in 2005,

Table 4Estimated changes in soil erosion over 1985, 1995, and 2005. (unit: Mg/ha).

1985 1995 2005 Mg/ha % %/year

Keum River BasinKeum River 16.9 16.2 19.6 2.7 15.8 0.7Keum River (West Sea) 24.3 25.0 23.1 �1.2 �4.8 �0.2Mankyung, Dongjin River 12.0 11.6 12.1 0.1 1.1 0.1Sapgyo River 22.0 23.1 21.3 �0.7 �3.1 �0.2

Sub-total 18.7 18.4 20.0 1.3 7.2 0.3

Nakdong River BasinNakdong River 13.0 12.8 15.7 2.7 20.5 0.9Nakdong River (South Sea) 18.1 21.0 19.2 1.1 5.9 0.3Nakdong River (East Sea) 10.1 12.0 11.0 0.9 8.7 0.4Taehwa River 15.4 23.6 21.0 5.5 35.9 1.5Hyungsan River 12.3 16.9 15.1 2.8 22.5 1.0Heoya, Suyoung River 25.3 32.3 29.9 4.7 18.6 0.9

Sub-total 14.2 15.0 16.9 2.6 18.5 0.9

Yeongsan River BasinSeomjin River 13.6 13.5 19.9 6.3 46.3 1.9Seomjin River (South Sea) 24.3 23.8 25.4 1.1 4.4 0.2Yeongsan River 22.1 19.3 24.4 2.3 10.6 0.5Yeongsan River (South Sea) 18.2 15.3 17.7 �0.5 �2.7 �0.1Yeongsan River (West Sea) 16.1 15.5 14.5 �1.6 �9.9 �0.5Tamjin River 31.9 25.7 38.3 6.4 20.2 0.9

Sub-total 21.5 20.1 24.5 2.9 13.7 0.6

Han River BasinAnseong River 19.2 22.5 22.1 2.8 14.8 0.7Han River 13.3 13.9 16.5 3.2 23.8 1.1Han River (East Sea) 10.3 9.7 12.7 2.4 23.5 1.1Han River (West Sea) 14.8 17.1 16.1 1.3 8.7 0.4

Sub-total 17.1 18.0 20.9 3.8 22.3 1.0

Total 17.1 17.4 20.0 2.8 16.6 0.8

268 S. Park et al. / Journal of Hydrology 399 (2011) 263–273

showing an increase of 2.9 Mg/ha, at an average annual growthrate of 0.8%.

Among the four river basins, the largest increase was in the HanRiver Basin at 3.8 Mg/ha, followed by the Yeongsan River Basin(2.9 Mg/ha), the Nakdong River Basin (2.6 Mg/ha), and the KeumRiver Basin (1.3 Mg/ha).

Soil loss generally increased in each river basin, with someexceptions, over the last 20 years. The Seomjin River showed thelargest increase in soil loss from 13.6 Mg/ha in 1985 to 19.9 Mg/ha in 2005 (46.3% [6.3 Mg/ha] increase). The lowest increase wasfound in the Mankyung-Dongjin River in which the soil loss in-creased from 12.0 Mg/ha in 1985 to 12.1 Mg/ha in 2005 (0.1 Mg/ha increase).

On the other hand, soil loss for the Yeongsan River (South Sea),Sapgyo River, Keum River (West Sea), and Yeongsan River (WestSea) decreased by 2.7–9.9% in 2005, compared with 1985. Amongthese rivers, the largest decline was in the Yeongsan River (West

Fig. 3. RUSLE maps of soil erosion

Sea), which had a value of 16.1 Mg/ha in 1985; by 2005, this valuehad decreased to 14.5 Mg/ha, with an annual average decrease rateof 0.5%.

3.2. Assessment of factors related to the scenarios

This study predicted the amount of soil loss in 2020 by estab-lishing scenarios involving changes in the rainfall–runoff erosivityfactor and cover-management factor. These two factors were as-sessed as follows. For the rainfall–runoff erosivity factor, valuesfrom the Korea Institute of Construction Technology, as well asthe 10-year frequency, were used. Fig. 4 shows the resulting rain-fall–runoff erosivity factors for the 50- and 100-year frequencies.

The cover-management factor was evaluated using the valuesin Table 2, obtained from the completed 2020 LSI map, constructedconsidering the ECVAM and using FR, AHP, and LR (Table 5). As awhole, the FR, AHP, and LR LSI maps showed similar distributionsdue to each land use-land cover (LULC) item. All categories, exceptfor water bodies, decreased, compared with 2005. Farmland andbarren land showed higher reduction rates (20% or above) thanother land types, followed by grassland with an approximately15% reduction rate. In contrast, forest and wetland showed reduc-tion rates of 5% or less.

When considering the ECVAM, the FR, LR, and AHP resultsshowed grassland decreases of 15.2%, 12.7%, and 12.7%, respec-tively. For farmland, a decrease of 22.1% was shown by FR, withsimilar values of 24.7% by the AHP and 24.0% by LR. By the FR,LR, and AHP approaches, bare land decreased by 29.3%, 21.1%,and 23.4%, respectively. The FR and AHP results showed wetlanddecreases of 5.4% and 3.0%, respectively, while the LR showed anincrease of 1.0%. The FR, AHP, and LR results also showed slight de-creases in forest cover, of 1.6%, 0.8%, and 1.1%, respectively.

The areas of farmland and bare land on LSI maps made withoutconsidering the ECVAM increased more than those areas whenconsidering the ECVAM. Among them, the largest increase wasshown for farmland, which increased by 4.7%, 2.4%, and 1.4% inthe LR, FR, and AHP results, respectively. On the other hand, theareas of forest, grassland, and wetland decreased on the LSI mapsthat considered the ECVAM. The LR, AHP, and FR LSI maps indi-cated that wetlands decreased the most, by 3.2%, 2.6%, and 1.9%,respectively (see Fig. 5).

3.3. Scenario-based analysis of 2020 soil loss

3.3.1. Scenario 1: the urban area shows a similar trend to 1985–2005and rainfall has a 10-year frequency

In scenario 1, the amount of 2020 soil loss considering the EC-VAM was 19.2 Mg/ha by the LR and 19.3 Mg/ha by both the AHPand FR. Values without considering the ECVAM were 19.9 Mg/ha

risk for 1985, 1995, and 2005.

Fig. 4. Distribution of the R factors used for the respective scenarios.

Fig. 5. Percentage contribution of each LULC category in each 2020 LSI map.

S. Park et al. / Journal of Hydrology 399 (2011) 263–273 269

by the LR approach and 19.6 Mg/ha by both the FR and AHP ap-proaches. Compared with 2005, the amount of soil loss in 2020when considering the ECVAM decreased by 0.7–0.8 Mg/ha, whilethat without considering the ECVAM decreased by 0.3-0.4 Mg/ha(Table 6).

The largest amount of soil erosion was found in the TamjinRiver Basin and the lowest was predicted for the Nakdong River(East Sea). When considering the ECVAM, the FR results showeda decrease of 2.1 Mg/ha in the Tamjin River Basin, from 38.8 Mg/ha in 2005 to 36.7 Mg/ha in 2020. For the same period and basin,the AHP showed a 1.3 Mg/ha decrease, to 37.5 Mg/ha, and the LRshowed a 1.7 Mg/ha decrease, to 37.1 Mg/ha. Without consideringthe ECVAM, sediment loss decreased by 1.2, 1.4, and 1.1 Mg/ha bythe FR, AHP, and LR approaches, to 37.6, 37.5, and 37.8 Mg/ha,respectively, in 2020.

When considering the ECVAM, for the Nakdong River (East Sea),the FR results showed soil loss of 11.0 Mg/ha and the AHP and LRboth estimated a loss of 11.1 Mg/ha; these values are 0.4–0.5 Mg/ha lower than the 11.6 Mg/ha estimate for 2005. Without consider-ing the ECVAM, the FR and AHP produced estimates of 11.2 and11.4 Mg/ha, respectively, representing decreases of 0.2–0.3 Mg/ha; the LR result showed soil loss of 11.6 Mg/ha.

3.3.2. Scenario 2: the urban area shows a similar trend to 1985–2005and rainfall has a 50-year frequency

In scenario 2, when considering the ECVAM, the amount of 2020soil loss was 36.4 Mg/ha using LR and 36.6 Mg/ha using FR andAHP. Without considering the ECVAM, soil loss was estimated tobe 37.8, 37.3, and 37.1 Mg/ha by the LR, FR, and AHP methods,

Table 5Change of area in 2020 land cover using LSI maps (unit: km2, %).

Using ECVAMa Not using ECVAMb 2005

FR Diff.c AHP Diff. LR Diff. FR Diff. AHP Diff. LR Diff.

Urban 13,106 120.6 13,106 120.6 13,106 120.6 13,052 119.7 13,092 120.4 13,106 120.6 5941Farmland 18,959 �22.1 18,322 �24.7 18,496 �24.0 19,540 �19.7 18,666 �23.3 19,639 �19.3 24,332Forest 58,318 �1.6 58,761 �0.8 58,615 �1.1 57,895 �2.3 58,459 �1.3 57,551 �2.9 59,247Grassland 2373 �15.2 2443 �12.7 2442 �12.7 2300 �17.8 2385 �14.8 2358 �15.8 2799Wetland 204 �5.4 209 �3.0 213 �1.0 200 �7.4 203 �5.6 206 �4.3 215Bare land 1026 �29.3 1145 �21.1 1113 �23.4 999 �31.2 1180 18.8 1127 �22.4 1453Waters 2229 0.0 2229 0.0 2229 0.0 2229 0.0 2229 0.0 2229 0.0 2229

Total 96,216 96,216 96,216 96,216 96,216 96,216 96,216

a In case ECVAM is included in the factors related to urbanization in drawing LSI maps.b In case ECVAM is not included in the factors related to urbanization in drawing LSI maps.c Amount of differences between LSI maps and area values in 2005.

Table 6The amount of 2020 soil loss for each river basin in scenario 1 (unit: Mg/ha).

FR LSI map AHP LSI map LR LSI map 2005 LULC map

With Without With Without With Without

Keum River BasinKeum River 19.3 19.6 19.3 19.6 19.2 19.9 20.0Keum River (West Sea) 25.7 26.0 25.7 26.7 25.5 27.1 27.1Mankyung, Dongjin River 12.1 12.2 12.0 12.3 12.0 12.5 12.8Sapgyo River 20.9 21.1 20.9 21.4 20.7 21.9 22.0

Sub-total 19.2 19.4 19.2 19.6 19.0 19.9 20.0

Nakdong River BasinNakdong River 15.5 15.7 15.5 15.6 15.4 15.9 15.9Nakdong River (South Sea) 32.0 34.4 32.1 32.8 31.7 34.2 32.9Nakdong River (East Sea) 11.0 11.2 11.1 11.4 11.1 11.6 11.6Taehwa River 20.2 21.2 20.4 21.0 20.1 21.9 21.1Hyungsan River 14.4 15.1 14.7 15.2 14.7 15.6 15.4Heoya, Suyoung River 32.7 35.3 30.4 31.6 30.1 34.5 31.2

Sub-total 16.4 16.8 16.4 16.6 16.3 17.0 16.9

Yeongsan River BasinSeomjin River 19.6 19.8 19.7 19.8 19.7 19.9 20.3Seomjin River (South Sea) 25.5 26.0 25.7 25.8 25.6 26.3 26.2Yeongsan River 23.8 24.3 24.0 24.2 23.9 24.9 25.1Yeongsan River (South Sea) 32.0 32.5 32.3 32.5 32.2 33.1 33.3Yeongsan River (West Sea 21.5 22.1 21.9 22.2 21.5 22.5 22.9Tamjin River 36.7 37.6 37.5 37.5 37.1 37.8 38.8

Sub-total 23.5 23.9 23.7 23.8 23.6 24.2 24.5

Han River BasinAnseong River 21.9 22.1 21.4 22.2 21.2 23.0 22.9Han River 20.5 20.8 20.5 20.7 20.4 21.0 21.1Han River (East Sea) 14.6 14.9 14.9 15.1 14.8 15.4 15.3Han River (West Sea) 29.0 30.5 28.6 29.8 28.3 31.3 30.4

Sub-total 20.2 20.6 20.2 20.5 20.1 20.9 20.9

Total 19.3 19.6 19.3 19.6 19.2 19.9 20.0

270 S. Park et al. / Journal of Hydrology 399 (2011) 263–273

respectively, representing increases of 1.4, 0.7, and 0.5 Mg/ha,respectively (Table 7).

Among the four main river basins, the Yeongsan River Basin wasestimated to have the most soil loss in 2020, followed by the HanRiver Basin, Keum River Basin, and Nakdong River Basin. Whenconsidering the ECVAM, the FR, AHP, and LR estimated soil lossesin the Yeongsan River Basin were 44.4, 44.8, and 44.6 Mg/ha,respectively. Without considering the ECVAM, the FR, AHP, andLR estimates were 45.2, 45.1, and 45.7 Mg/ha, respectively; thesevalues differed from the ECVAM estimate by 0.3–1.2 Mg/ha.

Within the main river basins, the FR, AHP and LR estimated thelargest soil losses for the Tamjin River when considering theECVAM: 73.1, 74.6, and 73.8 Mg/ha, respectively. Without consid-ering the ECVAM, estimates for the Tamjin were 74.8 Mg/ha(1.7 Mg/ha increase compared to the ECVAM estimate) by FR,74.6 Mg/ha (the same) by AHP, and 75.2 Mg/ha (1.3 Mg/ha

increase) by LR. For the Mankyung-Dongjin River, FR, AHP, andLR estimates were 21.8, 21.6, and 21.5 Mg/ha when consideringthe ECVAM and 21.8, 22.1, and 22.4 Mg/ha without consideringthe ECVAM, respectively.

3.3.3. Scenario 3: the urban area trend shows a similar trend to 1985–2005 and rainfall has a 100-year frequency

In scenario 3, when considering the ECVAM, soil loss was esti-mated to be 45.9, 46.0, and 45.7 Mg/ha by the FR, AHP, and LR ap-proaches, respectively. Without considering the ECVAM, theamounts were 46.8, 46.7, and 47.5 Mg/ha by FR, AHP, and LR,respectively, and thus 0.9, 0.7, and 1.8 Mg/ha larger than the for-mer estimates (Table 8).

Like in scenario 2, the amount of 2020 soil loss was the highestin the Yeongsan River Basin, followed by the Han River Basin,Keum River Basin, and Nakdong River Basin. For the Yeongsan Riv-

Table 7The amount of 2020 soil loss for each river basin in scenario 2 (unit: Mg/ha).

FR LSI map AHP LSI map LR LSI map

With Without With Without With Without

Keum River BasinKeum River 35.9 36.4 35.8 36.4 35.6 37.0Keum River (West Sea) 48.8 49.3 48.8 50.6 48.4 51.5Mankyung, Dongjin River 21.8 21.8 21.6 22.1 21.5 22.4Sapgyo River 38.7 39.1 38.6 39.6 38.4 40.6

Sub-total 35.7 36.1 35.6 36.4 35.4 37.0Nakdong River BasinNakdong River 28.4 28.8 28.4 28.7 28.3 29.1Nakdong River (South Sea) 60.7 65.3 60.8 62.3 60.2 64.9Nakdong River (East Sea) 23.6 24.0 23.8 24.4 23.8 24.9Taehwa River 40.7 42.9 41.1 42.3 40.6 44.1Hyungsan River 31.6 33.0 32.2 33.3 32.2 34.2Heoya, Suyoung River 64.4 69.5 59.9 62.3 59.2 68.1

Sub-total 30.9 31.7 30.8 31.3 30.6 32.0Yeongsan River BasinSeomjin River 36.0 36.3 36.2 36.3 36.1 36.4Seomjin River (South Sea) 49.7 50.6 50.3 50.5 50.0 51.3Yeongsan River 44.5 45.5 44.9 45.4 44.7 46.5Yeongsan River (South Sea) 62.5 63.6 63.2 63.5 63.0 64.7Yeongsan River (West Sea) 39.2 40.3 40.0 40.5 39.2 41.1Tamjin River 73.1 74.8 74.6 74.6 73.8 75.2

Sub-total 44.4 45.2 44.8 45.1 44.6 45.7Han River BasinAnseong River 41.2 41.5 40.2 41.7 39.7 43.1Han River 39.1 39.7 39.1 39.6 39.0 40.1Han River (East Sea) 32.1 32.8 32.8 33.3 32.5 33.9Han River (West Sea) 56.2 59.1 55.4 57.8 55.0 60.8

Sub-total 39.1 39.8 39.1 39.6 38.9 40.3

Total 36.6 37.3 36.6 37.1 36.4 37.8

Table 8The amount of 2020 soil loss for each river basin in Scenario 3 (unit: Mg/ha).

FR LSI map AHP LSI map LR LSI map

With Without With Without With Without

Keum River BasinKeum River 44.8 45.4 44.7 45.3 44.3 46.1Keum River (West Sea) 61.3 61.9 61.3 63.6 60.8 64.7Mankyung, Dongjin River 26.8 26.9 26.6 27.2 26.4 27.6Sapgyo River 48.1 48.6 48.1 49.3 47.8 50.5

Sub-total 44.5 45.0 44.4 45.4 44.1 46.2Nakdong River BasinNakdong River 35.2 35.8 35.3 35.6 35.1 36.1Nakdong River (South Sea) 76.1 81.9 76.3 78.2 75.5 81.4Nakdong River (East Sea) 31.1 31.6 31.3 32.1 31.3 32.7Taehwa River 52.2 54.9 52.7 54.3 52.1 56.6Hyungsan River 42.0 43.9 42.7 44.3 42.8 45.4Heoya, Suyoung River 81.9 88.4 76.2 79.3 75.3 86.5

Sub-total 38.7 39.8 38.6 39.3 38.4 40.1Yeongsan River BasinSeomjin River 44.6 45.0 44.8 45.0 44.7 45.2Seomjin River (South Sea) 63.0 64.2 63.8 64.0 63.4 65.0Yeongsan River 55.6 56.8 56.0 56.7 55.8 58.1Yeongsan River (South Sea) 79.3 80.6 80.2 80.6 79.9 82.0Yeongsan River (West Sea) 48.5 49.9 49.5 50.1 48.5 50.8Tamjin River 93.2 95.4 95.1 95.1 94.1 95.8

Sub-total 55.6 56.6 56.2 56.5 55.9 57.3Han River BasinAnseong River 51.4 51.9 50.2 52.1 49.6 53.9Han River 49.3 50.0 49.3 49.8 49.1 50.5Han River (East Sea) 42.6 43.5 43.5 44.1 43.1 45.0Han River (West Sea) 71.0 74.7 70.1 73.0 69.5 76.8

Sub-total 49.4 50.3 49.4 50.1 49.2 51.0

Total 45.9 46.8 46.0 46.7 45.7 47.5

S. Park et al. / Journal of Hydrology 399 (2011) 263–273 271

272 S. Park et al. / Journal of Hydrology 399 (2011) 263–273

er Basin, the FR, AHP, and LR estimates were 55.6, 56.2, and55.9 Mg/ha when considering the ECVAM and 56.6, 56.5, and57.3 Mg/ha without considering the ECVAM, respectively. The dif-ference between the estimates with and without consideration ofthe ECVAM ranged from 0.3 to 1.5 Mg/ha.

The largest soil loss was found for the Tamjin River, where FR,AHP, and LR results considering the ECVAM were 93.2, 95.1, and94.1 Mg/ha, respectively, while those without considering the EC-VAM were 95.4 (2.2 Mg/ha larger than the ECVAM estimate),95.1 (the same), and 95.8 Mg/ha (1.7 Mg/ha larger), respectively.The lowest soil loss was estimated for the Mankyung-Dongjin Riv-er. The FR, AHP, and LR estimates considering the ECVAM were26.8, 26.6, and 26.4 Mg/ha, respectively, while those without con-sidering the ECVAM were 26.9, 27.2, and 27.6 Mg/ha, respectively.

4. Discussion and conclusion

Soil is the foundation for vegetation growth and various ecosys-tems, as well as a key factor in maintaining healthy forests andwater circulation. In particular, surface soil is an important re-source, and regions with a high potential for soil erosion face prob-lems of decreased crop productivity and water storage capacity,which, directly and indirectly, cause water pollution. As concernsabout the environment grow, the importance of soil loss is beingincreasingly recognized and research on methods to preserve soilis actively underway. Information on soil loss and erosion preven-tion can serve as primary data for minimizing environmental im-pacts and developing policies and plans based on predicted soilloss.

For this purpose, this study quantitatively analyzed soil loss inmajor rivers of Korea using RUSLE. DEM data, detailed soil maps,and soil cover maps were used as data and analyzed by GIS andRS techniques. To predict the amount of 2020 soil loss, LSI mapswere created using FR, AHP, and LR approaches, and the amountof soil loss of each river basin was calculated under three scenarios,each assuming a similar urban area trend to that in 1985–2005, butwith 10-, 50-, and 100-year rainfall frequencies.

The OECD (2001) evaluated the amount of soil erosion accord-ing to five rating categories for the risk of soil erosion. By these rat-ings, soil erosion of 0–5 Mg/ha/year is tolerable, 6–10 Mg/ha/yearis low, 11–21 Mg/ha/year is moderate, 22–33 Mg/ha/year is high,and 33 Mg/ha/year or more is severe. Annual soil loss in Koreawas 20 Mg/ha or less, making it moderate by this rating scheme.However, the amount of soil loss from the Keum River (WestSea), Sapgyo River, Heoya Suyoung River, Seomjin River (SouthSea), Yeongsan River, and Anseong River ranged from 22 to33 Mg/ha/year, indicating a high degree of soil loss.

The Tamjin River Basin had the largest amount of soil loss, withvalues of 31.9 Mg/ha in 1985 and 25.7 Mg/ha in 1995, a decrease of6.2 Mg/ha. However, in 2005, soil loss had increased by 12.7–38.3 Mg/ha and the degree of soil loss changed from high to severe.The Tamjin River is the only region with a severe risk of soil lossamong the 20 rivers basins.

The amount of 2020 soil loss in each scenario was as follows.Scenario 1 analyzed the pattern of 2020 soil loss under the sameconditions as from 1985 to 2005. When considering the ECVAM,soil loss was predicted at to be 19.2–19.3 Mg/ha, a decrease of0.7–0.8 Mg/ha, compared with 2005. Without considering the EC-VAM, soil loss was 19.6–19.9 Mg/ha, a decrease of 0.3–0.4 Mg/ha,compared with 2005.

Scenarios 2 and 3 reflected factors that are expected to change,for example climate and changes in urban areas. Additionally, dif-ferences in rainfall frequencies were examined. In scenario 2,assuming a 50-year frequency, soil loss was estimated at 36.4–36.6 Mg/ha when considering the ECVAM and 37.1–37.8 Mg/ha

without considering the ECVAM. In scenario 3, assuming a 100-year rainfall frequency, soil loss was predicted to be 45.7–46.0 Mg/ha considering the ECVAM and 46.7–47.5 Mg/ha withoutconsidering it.

Thus, in scenario 1, the amount of soil loss was lower in 2020than in 2005, due to decreases in the areas of farmland, grassland,and bare land among the LULC items. Farmland decreased by5373–6010 km2, from 24,332 km2 in 2005; grassland decreasedby 356–426 km2, from 2799 in 2005; and bare land decreased by307–426 km2 from 1453 km2.

As a whole, without considering the ECVAM, the amount of soilloss was higher than the amount when considering the ECVAM.This is because, without considering the ECVAM, the areas of forest,wetland, and grassland decreased; these categories have highersoil preservation value and legislative limitations. At the sametime, without considering the ECVAM, the areas of farmland andbare land increased. These results indicate that the amount of soilloss will increase if urban areas are developed at the cost of dam-age to areas of high environmental preservation and legislative va-lue. In contrast, the amount of soil loss will decrease if urban areasare developed while preserving these valuable regions.

In each scenario, the risk of soil loss of the Heoya, Suyoung, andTamjin river basins was severe. The degree of soil loss in the Tam-jin River was especially severe currently and was also predicted tobe severe in 2020. For regions with a high risk of soil loss, includingthe Tamjin River, periodic monitoring must be carried out andmethods must be found to minimize soil loss, such as environmen-tally friendly agriculture, improved cultivation methods, and theplanting of buffer-zone forests.

Furthermore, in areas of high environmental and legislative va-lue, urban development should be carried out with maximumpreservation of the environment to reduce the amount of soil loss.Moreover, further research should be conducted to improve theaccuracy of erosion estimates using rainfall frequencies shorterthan 10 years and using actual rainfall data to calculate the rain-fall–runoff erosivity factor.

This study calculated the amount of soil loss over a 20-year per-iod in Korea using RUSLE and predicted the amount of soil losswith and without consideration of the ECVAM, assuming a similarurban growth trend to that of 1985–2005 and 10-, 50-, and 100-year rain frequencies. The amount of soil loss increased over theperiod from 1985 to 2005, because of increases in the areas ofgrassland and bare land. When considering the ECVAM, soil lossin 2020 was predicted to 19.2–19.3 Mg/ha with the 10-year rainfallfrequency, 36.4–36.6 Mg/ha with the 50-year rainfall frequency,and 45.7–46.0 Mg/ha with the 100-year rainfall frequency. With-out considering the ECVAM, soil loss is expected to increase, by0.3–0.7 Mg/ha at the 10-year rainfall frequency, by 0.5–1.5 Mg/haat the 50-year frequency, and by 0.7–1.8 Mg/ha at the 100-year fre-quency. Thus, to reduce soil loss, urban development must takeenvironmental preservation into account.

Acknowledgement

This work was researched by the supporting project to Yeong-nam Sea Grant.

References

Bantayan, N.C., Bishop, I.D., 1998. Linking objective and subjective modeling forlanduse decision-making. Landscape Urban Plan. 43, 35–48.

Beasley, D.B., Huggins, L.F., Monke, E.J., 1980. ANSWERS: a model for watershedplanning. Trans. ASAE 23 (4), 938–944.

Brady, N.C., Weil, R.R., 1996. The Nature and Properties of Soils, 11th ed. PrenticeHall, New Jersey.

Dai, F.C., Lee, C.F., Zhang, X.H., 2001. GIS-based geo-environmental evaluation forurban land-use planning: a case study. Eng. Geol. 61, 257–271.

S. Park et al. / Journal of Hydrology 399 (2011) 263–273 273

Dawen, Y., Kanae, S., Oki, T., Koike, T., Musiake, K., 2006. Global potential soil erosionwith reference to land use and climate changes. Hydrol. Process. 17, 2913–2928.

Elwell, H.A., Stocking, M.A., 1982. Developing a simple yet practical method of soilloss estimation. Trop. Agric. 59 (1), 43–48.

Goldman, S.J., Jackson, K., Bursztynsky, T.A., 1986. Erosion and Sediment ControlHandbook. Mc-Graw Hill. pp. 5.1–5.32.

Kim, J.H., Lee, C.D., Kim, K.T., Choi, Y.S., 2007. The soil loss analysis using landcoverof WAMIS. J. Korean Assoc. Geogr. Inform. Stud. 10 (4), 122–131 (in Korean).

Kim, J.H., Kim, K.T., Lee, H.J., 2009. Analysis of Korea soil loss and hazard zone. J. GISAssoc. Korea 17 (3), 261–268 (in Korean).

KMOE (Korea Ministry of Environment), 2001. A Study on the Conservation ofSurface Soil and Erosion Control, pp. 30–57 (in Korean).

KMOE (Korea Ministry of Environment), 2009. The Soil Conservation Master Plan,2010–2019, pp. 11–12 (in Korean).

Knisel, W.G., 1980. CREAMS, A Field Scale Model for Chemicals, Runoff, and Erosionfrom Agricultural Management Systems, US Department of AgricultureResearch Service Report No. 26.

Korea Institute of Construction Technology, 1992. Sediment Yield Rate Research forthe Design of the Dam in the Watershed, p. 287 (in Korean).

Korea Meteorological Administration Homepage. <http://www.kma.go.kr.>Korea Remote Sensing Center Homepage. <http://krsc.kict.re.kr>.Lee, G.S., Hwang, E.H., 2006. The influence analysis of GIS-based soil erosion in

water-pollutant buffering zone. J. Korean Soc. Civil Eng. 26 (2D), 335–340 (inKorean).

Lee, S., Sambath, T., 2006. Landslide susceptibility mapping in the Damrei Romelarea, Cambodia using frequency ratio and logistic regression models. Environ.Geol. 50, 847–855.

Lee, S., Talib, J.A., 2005. Probabilistic landslide susceptibility and factor effectanalysis. Environ. Geol. 47 (7), 982–990.

Lee, S., Kim, Y.J., Min, J.D., 2000. Development of spatial landslide informationsystem and application of spatial landslide information. J. GIS Assoc. Korea 8,141–153 (in Korean).

Lee, J.Y., Yang, D.Y., Kim, J.Y., Hong, S.S., Kim, J.K., 2006. The relationship betweensoil erosion and cultivation in the Bockha stream watershed area. J. KoreanAssoc. Geogr. Inform. Stud. 9 (1), 56–65 (in Korean).

Lee, M.B., Kim, N.S., Jin, S., Kim, H.D., 2008. A study on the soil erosion by landuse inthe Imjin River basin, DMZ of central Korea. J. Korean Geogr. Soc. 43 (3), 263–275 (in Korean).

Lee, C.J., Park, E.S., Yoo, N.J., 2009. A study on disaster and recovery of landslides atInje province on Korea. J. Ind. Technol. 29, 9–17 (in Korean).

Lvoivich, M.I., Karasik, G.Y., Bratseva, N.L., Medvedeva, G.P., Meleschko, A.V., 1991.Contemporary Intensity of the World and Intracontinental Erosion,Meshduvedomstv. Geophisich, Comotet pri Prisendeume Akad, Nauk SSSR,Moskva.

McFadden, D., 1973. Conditional logit analysis of quantitative choice behavior. In:Zarembka, P. (Ed.), Frontiers in Econometrics. Academic Press, New York, pp.105–142.

Millward, A.A., Mersey, J.E., 1999. Adapting the RUSLE to model soil erosionpotential in a mountainous tropical watershed. CATENA 38 (2), 109–129.

Ministry of Construction and Transportation, 2000. Production of Probable IsohyetalChart in Korea (in Korean).

Morgan, R.P., Morgan, D.D., Finney, H.J., 1984. A predictive model for the assessmentof soil erosion risk. J. Agric. Eng. Res. 30, 245–253.

Morgan, R.P.C., Quinton, J.N., Rickson, R.J., 1990. Structure of the soil erosionprediction model for the European community. In: Proceedings of InternationalSymposium on Water Erosion, Sedimentation and Resource Conservation, 9–13October, 1990, Dehradun, India. Central Soil and Water Conservation Researchand Training Institute, CSWCRTI, Dehradun, India, pp. 49–59.

National Institute for Disaster Prevention, 1998. Study on the Sediment YieldEstimation Due to Land Development (I), pp. 117–154 (in Korean).

National Institute of Meteorological Research, 2009. Understanding of ClimateChange, pp. 76–80 (in Korean).

Nearing, M.A., Foster, G.R., Lane, L.J., 1989. A process-based soil erosion model forUSDA water erosion prediction project. Trans. ASAE 32 (5), 1587–1593.

OECD, 2001. Environmental Indicators for Agriculture, vol.3. Part 4. Soil QualityOEDC, Paris, France, pp. 197–226.

Oh, J.H., Jung, S.G., 2005. Potential soil prediction for land resource management inthe Nakdong River basin. J. Korea Soc. Rural Plan. 11 (2), 9–19 (in Korean).

Park, K.H., 2003. Soil erosion risk assessment of the Geumho River watershed usingGIS and RUSLE methods. J. Kor. Assoc. Geogr. Inform. Stud. 6 (4), 24–36.

PSIAC, 1968. Pacific Southwest Inter-Agency Committee. Report of the WaterManagement Sub-Committee. Sedimentation Task Force. 10. ASCE. 98. ReportHY12.

Remortel, R.V., Hamilton, M., Hickey, R., 2001. Estimating the LS factor for RUSLEthrough iterative slope length processing of digital elevation data withinArcInfo Grid. Cartography 30 (1), 27–35.

Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D.K., Yoder, D.C., 1997. PredictingSoil Erosion by Water: A Guide to Conservation Planning with the RevisedUniversal Soil Loss Equation (RUSLE), Agriculture Handbook, No.703, USDepartment of Agriculture Research Service, Washington, DC, USA, p. 348.

Shin, K.J., 1999. The Soil Loss Analysis Using GSIS in Watershed. A Master Thesis,Kangwon National University, Gangwon-do, Korea, p. 116 (in Korean).

Statistics Korea Homepage. <http://www.index.go.kr>.Terranova, O., Antronico, R., Coscarelli, R., Iaquinta, P., 2009. Soil erosion risk

scenarios in the Mediterranean environment using RUSLE and GIS: anapplication model for Calabria (southern Italy). Geomorphology 112, 228–245.

Walling, D.E., 1983. The sediment delivery problem. J. Hydrol. 65, 209–237.Wang, G., Gertner, G., Fang, S., Anderson, A.B., 2003. Mapping multiple variables for

predicting soil loss by geostatistical methods with TM images and a slope map.Photogramm. Eng. Rem. Sens. 69 (8), 889–898.

Wischmeier, W.W., Smith, D.D., 1965. Predicting Rainfall Erosion Losses fromCropland East of the Rocky Mountains, US, Agriculture Handbook, No. 282, USDepartment of Agriculture Research Service, Washington, DC, USA, 45 pp.

Wischmeier, W.H., Smith, D.D., 1978. Predicting Rainfall Erosion Losses, AgricultureHandbook, No. 537, US Department of Agriculture Research Service,Washington, DC, USA, 58 pp.

Wu, F., 2002. Calibration of stochastic cellular automata: the application to rural–urban land conversions. Int. J. Geogr. Inform. Sci. 16 (8), 795–818.

Yesilnacar, E., Topal, T., 2005. Landslide susceptibility mapping: a comparison oflogistic regression and neural networks methods in a medium scale study,Hendek region (Turkey). Eng. Geol. 79, 251–266.