Original Article
Shallow Landslide Susceptibility Modeling Incorporating Rainfall Statistics: A Case Study from the Deokjeok-ri Watershed, South Korea 1
2
1
1
Ananta Man Singh PRADHAN , Paolo TAROLLI , Hyo-Sub KANG , Ji-Sung LEE 1 and Yun-Tae KIM* 1 Dept. of Ocean Engineering, Pukyong National University (45 Yongso-ro, Nam-gu, Busan, 608-737, Korea) 2 Dept. of Land and Agroforest Environments, University of Padova, Pgripoli, Italy *E-mail:
[email protected]
A physically based slope stability model was applied to predict topographic and climatic control on shallow landslide initiation processes in mountainous terrain. We applied two simple hydrological models, coupled with the infinite slope stability analysis, to the July 2006 landslide event in Deokjeok-ri, South Korea. The rainfall predicted to cause instability in each topographic element is characterized by duration and frequency of occurrence. The incorporation of a rainfall frequency–duration relationship into assessment of landslide susceptibility provides a practical way to include climate information into estimation of the relative potential for shallow landsliding. A GIS-based landslide inventory map of 748 landslide locations was prepared using data from previous reports, aerial photographic interpretation, and extensive field work. This landslide inventory was used to document sites of instability and to provide a test of model performance by comparing observed landslide locations with model predictions. The area under curve of QD-SLaM was 0.79, which means that the overall accuracy of the landslide susceptibility is 79% and the prediction result is good. Key words: Deokjeok-ri, GIS, Shallow landslide, QD-SLaM, Susceptibility modeling
including statistical classification methods and process-based models [Carrara et al., 1991; Claessens et al., 2007]. Montgomery and Dietrich [1994] developed a simple physically based model, based on digital terrain data, which couples a steady-state shallow saturated subsurface model with an infinite slope stability model. The steady-state model incorporates a scheme based on the formulation proposed by O'Loughlin [1986] to simulate the topographic dependence of runoff generation during transient storm events in humid environments. Landslides constitute a major natural hazard in the Republic of Korea due to high rates of weathering, abundant rainfall, steep topography, and infrastructure development. Korea is experiencing changes in climate parameters, including annual temperature and precipitation [Chung et al. 2004]. During the past 30 years in Korea, average rainfall in summer season increased from 661 mm for the period 1976~1985, to 710 mm for the period 1996
1. INTRODUCTION Rainfall is an important triggering mechanism for slope failure, which accompanies severe damage and losses of life and property. The occurrence of precipitation-triggered shallow landslides is of concern in hydro-geomorphic and natural hazards science due to the high ranking of such events among natural disasters in terms of both the number of people affected globally and the proportion of fatalities on the affected population. Shallow landsliding can evolve into debris flows, resulting in high risk where vulnerable targets are involved [Petley, 2012]. Landslide susceptibility maps represent one of the key elements for landslide risk management. Landslide susceptibility is the probability of the spatial occurrence of a slope failure, given a set of geo-environmental conditions [Guzzetti et al., 2005]. Many methods have been proposed to evaluate landslide susceptibility at the basin scale, 18
common slope instability in weathered soil deposits is an initial shallow translational slide followed by flow of the disturbed mass [Turner 1996], which occurs following periods of heavy rainfall. The infiltration of rain [i.e., saturation from above] and the formation and rise of a temporary perched water table, which is in contact with the less permeable bedrock, increase the pore water pressure and cause a reduction in the shear strength of the soil.
2. STUDY AREA Fig. 1 Location and landslide inventory of Deokjeok-ri Watershed
Deokjeok-ri Creek is located in the northeastern sector of Korea in Inje County. The site lies at 38°04′07″–38°05′42″N latitude and 28°11′11″– 128°18′E longitude; it occupies approximately 33.4 km2 and is surrounded by steep mountains, shown in Fig. 1. The study area mainly consists of massive older granite of the Mesozoic era, and the Creek in the lower reaches is partly composed of banded gneiss of the Precambrian era. The valleys are shaped by faults, causing the ground to be very weak. The study area contains a range of similar soils reflecting the same type of parent material composition and texture, age of soil development, climate, relief, and landscape position and drainage. In the study area, most of the landslides have a failure surface within an intermediate saturated layer of saprolithic material. In the steep area, the failure surface forms at the point of contact between the residual soils and the bedrock. Many shallow landslides and debris flows occurred in the Deokjeok-ri Creek area following Typhoons Ewiniar and Bilis, which left 17 people dead, 12 missing, and 29 injured. The combined effects of extreme precipitation and weathered granite soil covering steep slopes led to large risks of slope instability, slope movement, and slope failure [Pradhan and Kim, 2014]. Such type disastrous event had not experienced before in Korea. Landslide inventory maps was prepared by consulting satellite imagery and aerial photographs coupled with field surveys using GPS. Remote sensing technique for landslide inventory is important to reduce cost and time for work. To produce a detailed and reliable landslide inventory map, landslide locations obtained from remote sensing technique was verified by extensive field surveys of Deokjeok-ri Watershed. At Deokjeok-ri, a total of 748 landslides were mapped and subsequently digitized in ArcGIS 10.2 environment for further analysis; the dominant failures were shallow translation landslides that sometimes
~2005, and to 814 mm for the period 1996~2005 [Seo and Lee 2011]. Due to these concentrated higher rainfalls, rainfall-induced slope failure became more serious in Korea. Usually a concentrated short duration and high intensity rainfall from July to August has caused slope instability in Korea. Landslides in Korea usually are induced by intensive rainfall in summer season [Kim et al., 2000]. Typhoons Yanni (in 1998), Rusa (2002), Maemi (2003), and Ewinia (2006) brought particularly heavy rains that caused hundreds of landslides in Korea [Park et al. 2013] Statistics from the National Disaster Management Institute (NDMI) shows that landslides are responsible for at least 34% of all fatalities from natural disaster in the period of 2002 to 2011, in Korea. Weathered granite soil is a generic term for soils generated via weathering of granite and granitic gneiss in such a way that the residual soils remain in situ. The mechanical properties of decomposed granite soil following weathering are affected by the content of the primary minerals, which may include quartz, mica, and feldspar, by bedrock structures, and by the natural conditions, such as the climate and drainage conditions [Kwon and Oh, 2011]. Degradation of the mechanical properties of the bedrock may be affected by changes in the water content, e.g., cycles of wetting and drying, softening [Picarelli, 2000], or infiltration of different aqueous solutions [Calvello et al., 2005]. Gravitational forces acting on these disaggregated materials may cause them to move down the slope and accumulate in topographic depressions. The aim of this research is to investigate and evaluate the predictive power of a quasi-dynamic process-based landslide model [Borga et al., 2002] to predict the spatial distribution of shallow landslide susceptibility. QD-SLaM, a previously published quasi-dynamic physically based slope stability model for shallow landsliding, was applied to a granitic watershed in South Korea. The most
19
a
c
flow and preferential flow in a large number of landscapes susceptible to shallow landsliding, including the study area. These processes may have a considerable impact on the response timescale of vertical infiltration and lateral groundwater. Decades of work have shown the significance of vertical and lateral flow through soils containing macro pores under unsaturated soil conditions. The influence of these processes on the triggering of shallow landslides has been analysed in detail by Krzeminska [2013]. Coupling the groundwater model and the infinite slope stability model under the assumption that the failure plane is on the impeding layer [Taylor, 1948; Haefeli, 1948] provides a relationship for the critical steady state rainfall rate rc(d), in other words the rainfall rate required to trigger slope failure for the specific topographic element. The relationship reads as follows:
b
d
Fig. 2 Slope failure observed in the study area: a, b, c translation shallow landslide and d oblique aerial view of landslides.
resulted in debris flows during periods of intense rain as depicted in Fig. 2. A DEM provide a topographic basis for the model [Pradhan and Kim, 2015]. The DEM was constructed using 1:5, 000 scale LIDAR data, and a 10 × 10 m grid was used, giving a spatial resolution of the topographic data of 10 m.
rc (d ) =
T sin θ a(d )
ρ Cr + Cs W tan θ 1 − + s + ρ gz cos θ tan φ ρ ρ tan φ w gz w w
(1) where C combines soil and root cohesion, θ slope, W is the vegetation surcharge, ϕ is the internal friction angle of the soil, ρs is the wet soil density, ρw is the density of water, g is the gravitational acceleration, and T is the soil transmissivity, defined as the product of the saturated lateral hydraulic conductivity Ks and soil thickness z. Since root strength produces significant reinforcement in vegetation-covered slopes, the formulation of QD-SLaM includes the effective soil cohesion due to vegetation. For a predefined storm duration d, Eq. (1) allows for the determination of the minimum uniform rainfall rc needed to cause instability, which is the meaning of the critical rainfall. By introducing a dynamic drainage area, the QD-SLaM may offer an efficient way to model the subsurface flow response at short temporal scales [Barling et al., 1994]. Moreover, the model provides a framework to relate the characteristics of the critical rainfall (rate and duration) to their probability of exceedance. Application of Eq. (1) allows the definition of three slope stability conditions: unconditionally stable, unconditionally unstable and conditionally unstable. A slope is defined as (a) unconditionally stable if it is stable even when it is saturated, (b) unconditionally unstable if it is unstable even when the soil is dry, and (c) conditionally unstable if the slope instability depends on rainfall conditions. One should note that the model, as applied here, does not include the description of the effects of road drainage on surface/subsurface flow dynamics.
3. THE SHALLOW LANDSLIDE MODEL In this study, the QD-SLaM model based on the coupling of a topography-driven model of subsurface flow with an infinite slope, Mohr–Coulomb failure model to describe the shallow landsliding process was adopted. The QD-SLaM model [Borga et al., 2002; Tarolli et al., 2008; 2011] is used in this work to predict the spatial distribution of shallow landslide susceptibility. The model predicts the duration and intensity of rainfall required for landslide initiation using a quasi-dynamic wetness index (QDWI) to predict the spatial distribution of soil saturation in response. The QDWI is the ratio between the effective contributing area and the local slope [Borga et al., 2002]. The effective contributing area a (d) is the fraction of the total specific contributing area which contributes subsurface flow to the contour segment within a specified drainage period d corresponding to a rainfall duration. This is based on the hypothesis that all precipitation infiltrates and that vertically infiltrating volumes quickly redistribute and produce lateral subsurface flow. The assumption of instantaneous infiltration, which leads to disregarding the effects of infiltration on near-surface pore pressure distributions and consequent slope stability, is motivated by the observation of the importance of macro pores, pipe
20
parameters ζ1 and m can be estimated by linear regression of mean values of annual maxima of precipitation depth against their durations, after log transformation. Combining Eqs. (1) and (3) yields the following equation for the value of yTr (Tarolli et al., 2011) : y Tr =
1− m ρ Cr + Cs W tan θ d cr 1 T sin θ π 1 − ⋅ ⋅ 1 − + s + −ξ a (d ) ρ w gz cos θ tan φ ρ w ρ w gz tan φ ς 1 6 CV
(5) Based on Equation (5), the value of exp( yTr k ) can be determined for each topographic element. Although our study area is mainly covered by pine trees, the vegetation surcharge and root cohesion were not considered in the following analysis.
Fig. 3 Regional depth-duration-frequency relationship for the study area for return periods ranging from 10 to 300 yr.
5. RESULTS AND DISCUSSION
4. COMBINING THE SLOPE STABILITY MODEL WITH THE SIMPLE PRECIPITATION MODEL
The largest average monthly precipitation in the Inje County typically occurs during July. Typhoon Ewiniar in July 2006 brought record-breaking rainfall in the study area. The monthly precipitation in July 2006 was 2.4 times greater than the annually averaged monthly rainfall. This rainfall was caused by two factors: a rain front from North Korea and vapor developing in China due to Typhoon Bilis. The collected rainfall information from the gage (Fig. 1) shows the total rainfall from July 14 to July 16, 2006, was 402 mm. The maximum daily rainfall of 227 mm occurred on July 15, 2006. The greatest rainfall intensity was from 8 am to 11 am on July 15; 168 mm of rainfall was measured during these 3 hours. The maximum hourly rainfall on July 15 was 62 mm/hour, and the maximum rainfall over a 24-hour period was 192 mm. Rainfall amounts and corresponding durations and return periods for the rainfall data collected at the rain gauge are reported in Fig. 3, showing that the return period for durations of 1 to 6 hours ranges from 10 to 300 yr. Six disturbed and three undisturbed samples corresponding to shallow scars were collected and analyzed in laboratory. According to the unified soil classification system (USCS), the soil samples were characterized by non-plastic well- to poorly graded sand. 100 mm diameter undisturbed samples were used for triaxial test. The specimens have an approximate 2:1 height-to-diameter ratio. The average cohesion was 2.14 kPa, and the average internal frictional angle was 35.61°, that were determined in triaxial shear test. The in-situ permeability tests were performed in various soil depths. And shear strength parameters and physical properties of soil were determined in the laboratory
Equation (2) provides the critical rainfall rate for a precipitation of a given duration, thus offering a way to quantify the return time of the critical rainfall. The variability of storm intensity with duration for a specified frequency level is often represented by the intensity-duration-frequency (IDF) relation. A power function is often used to model the IDF relation [Koutsoyiannis et al., 1998]:
rF (d ) = ζ F d mF −1
(2)
where rF is the rainfall rate which can be exceeded with a probability of (1-F), and ςF and mF are model parameters. The scaling properties of the statistical moments of rainfall depths of different durations are used in this work to derive the IDF relationship [Ceresetti et al., 2010]. Aronica et al. [2012] showed that a generalized extreme value (GEV) simple scaling model described well the distribution of annual maximum series of rainfall for the study region. The GEV simple scaling distribution of the rainfall rate rF (d) can be determined as 6 rF (d ) = ς1 ⋅ 1 − CV ⋅ ⋅ (ξ + yTr ) ⋅ d m −1 π
(3)
where ξ is the Euler’s constant (approximately 0.5772) and yTr can be derived by the following relation: T y Tr = ln ln r Tr − 1
(4)
where Tr, recurrence interval, corresponds to the exceedance probability (1-F). The values of the
21
Table 1 Hydraulic and geotechnical characteristics
Properties In-situ permeability Cohesion Internal frictional angle Density Assumed soil depth
Average 1.87×10-5(m/s) 2.14 (kPa) 35.61 1.68 (kg/m3) 1 (m)
Table 2 Parameters estimated for Gumbel equation (3)
Parameter
ς1 CV m
Value 58 mm 0.46 0.512
shown in Table 1. Six permeability measurements were carried out at depths of 20 cm, 40 cm, and 60 cm using a 2800K1 Guelph Permeameter and the constant head method. Soil depth surveys are time consuming, and the soil depth is often difficult to measure even for a small basin like Deokjeok-ri, the soil depth was estimated through field observations of the 50 landslide initiation scarps. The value adopted for the entire basin was considered to be equal to the soil depth at which the slope failure occurred, and the average soil depth was estimated as 1m. For rainfall analysis, a record of 20 years of annual maximum rainfall data of 1, 3, 6, 12, and 24hr data were collected at the meteorological stations located nearby study area. Table 2 represents the analyzed data, shows the heavy rainfall regime is much stronger. The QD-SlaM model shows seven classes that vary from stable to unconditionally unstable as shown in Fig. 4. At each terrain element, the output from the QD-SlaM is a return period, whereas that from the steady-state model is a critical rainfall. Comparison of observed landslide locations with model predictions provides an assessment of model reliability. The comparison is obtained by mapping observed landslide on a map of predicted critical return period. Better model performance would be reflected in a larger difference between fractions of catchment and observed landslide area corresponding to low values of critical rainfall. Two types of error can be identified in this way: (1) a site is characterized by the model as unconditionally unstable or with a low value of critical rainfall, but no scars were observed on it; (2) a site is predicted as unconditionally stable or characterized by a high value of critical rainfall, but slope failures were mapped on it.
Fig. 4 a Return period of critical rainfall based susceptibility map, b enlarged portion of inset. Table 3 Percentage of catchment area and of observed landslide
Freq of critical rainfall (return period, Year) Unconditionally Unstable 0-2 2-10 10-50 50-100 >100 Unconditionally Stable
Catchment area (%) 4.74 3.40 6.83 11.76 15.59 15.0 42.67
Landslide area (%) 10.10 8.65 24.52 24.04 24.04 7.69 0.96
The comparison of observed landslide locations with model predictions is reported in Table 3. Results reported in Table 3 indicate that a total of 4.74% of the area was categorized as being unconditionally unstable, 3.40 % of the area had return period 0–2 yr, 6.83 % had critical rainfall of 2–10 yr, 11.76 % of the area was categorized as unstable with 10–50 yr of critical rainfall, and 15.59 % was delineated as unstable with 50–100 yr. Similarly, 15 % of the area was identified as unstable with 100 yr return period. Also, 42.67 % of the area of the watershed was found to be unconditionally stable. The QD-SLaM approach to instability modeling was quite successful in describing slope failure in the study area, identifying 10.10 % of observed landslides in unconditionally unstable areas, 8.65 %
22
1.0
settlements.
0.8
6. CONCLUSION QD-SLaM has been applied to the 33.4 km2 Deokjeok-ri Watershed. The model uses a ‘quasi-dynamic’ wetness index to predict the spatial distribution of soil saturation in response to a rainfall of specified duration. The approach provides a way to capture both topographic and climatic forcing on shallow landsliding. At each terrain element, the output from the QD-SLaM is a return period. The return period shows seven classes that vary from unconditionally unstable to unconditionally stable. A total of 748 landslides were mapped and the dominant failures were shallow translation slide in the study area. Comparison with mapped landslides shows that a total of 10.10% of mapped landslide flagged in unconditionally unstable terrain. Similarly, 8.65 % in regions with return period of 0–2 yr, 24.52 % in areas with return period of 2–10 yr, 24.04 % landslides occurred in return period of 10–50 yr, 24.04 % in return period of 10–50 yr, and 7.69 % of landslides in return period of >100 yr. less than 1% landslide were identified in unconditionally stable may be considered as a measure of the error in the model. The area under curve of QD-SLaM was 0.79, which means that the overall accuracy of the landslide susceptibility is 79% and the prediction result is good. This result demonstrate the QD-SLaM was successful in identifying the unstable areas under return critical rainfall. A coupled model of infinite model and rainfall is important for hazard mitigation in a mountainous area.
FL(q)
0.6
0.4
QD-SLaM AUC 0.79
0.2
0.0 0.0
0.2
0.4
0.6
0.8
1.0
FB(q)
Fig. 5 Relationship between cumulative frequencies FL(q) and FB(q)
in regions with return period of 0–2 yr, 24.52 % in areas with return period of 2–10 yr, 24.04 % in areas with return period of 10–50 yr, 24.04 % in areas with return period of 10–50 yr, and 7.69 % of landslides in areas with return period of >100 yr. The fact that 0.96 % of landslides occurred in areas that were identified as unconditionally stable may be considered as a measure of the error in the model. The western portion of the watershed has more unconditionally unstable land lower return interval areas of instability because those areas are simply steeper in topography. For the model performance assessment, the approach developed by Borga et al. [2002] was adopted in this study. Let q be an index that is predictive of slope instability, which decreases with increasing hazard. Two empirical distribution functions FB(q) and FL(q) are defined for the terrain elements and for the observed landslide cells, respectively. FB(q) represents the percentage of terrain elements with q less than the given value, and FL(q) represents the percentage of landslide cells with q less than the given value. An empirical function is defined to express the relationship between FL(q) and FB(q) for a given q-index model, by reporting fraction of basin area FB(x) on the X-axis and fraction of landslide area FL(x) on the Y-axis. Fig. 5 represents the functions for the QD-SLaM. The area under curve of QD-SLaM was 0.79, which means that the overall accuracy of the landslide susceptibility is 79% and the prediction result is good. This result demonstrate the QD-SLaM was successful in identifying the unstable areas under return critical rainfall. The shallow landslide susceptibility incorporating rainfall statistics revealed several potentially affected areas by rainfall-triggered shallow landslides and debris flows, which corresponds to
ACKNOWLEDGMENT: The authors are thankful to the reviewers for their valuable comments that were very useful in bringing the manuscript into its present form. This research was supported by the Public Welfare and Safety Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Science, ICT, and Future Planning (grant No. 2012M3A2A1050977), a grant (13SCIPS04) from Smart Civil Infrastructure Research Program funded by Ministry of Land, Infrastructure and Transport (MOLIT) of Korea government and Korea Agency for Infrastructure Technology Advancement (KAIA) and the Brain Korea 21 Plus (BK 21 Plus). REFERENCES Aronica, G. T., Brigandí, G., and Morey, N. (2012): Flash floods and debris flow in the city area of Messina, north-east part of Sicily, Italy in October 2009: the case of the
23
Kwon Y. and Oh S. (2011): Physical and mechanical properties of decomposed granite soils sampled in Cheongju, Korea. Int J Physical Sciences. Vol. 6(24), pp: 5777–5794. Montgomery D.R. and Dietrich W.E. (1994): A physically based model for the topographic control on shallow landsliding. Water Resources Research vol. 30, pp: 1153–1171. O’Loughlin E.M. (1986) Prediction of surface saturation zones in natural catchments by topographic analysis, Water Resources Research, vol. 22(5), pp: 794–804. Park, S., Choi, C., Kim, B., Kim, J. (2013): Landslide susceptibility mapping using frequency ratio, analytic hierarchy process, logistic regression, and artificial neural network methods at the Inje area, Korea. Environmental earth sciences, vol: 68(5), pp: 1443-1464. Petley, D. N. (2012): Global patterns of loss of life from landslides, Geology, Vol. 40, pp. 927–930. Picarelli L. (2000): Mechanisms and rates of slope movements in fine grained soils. Engineering Geology. Vol., pp: 1618–1670. Pradhan A.M.S. and Kim Y.T. (2014): Relative effect method of landslide susceptibility zonation in weathered granite soil: a case study in Deokjeok-ri Cree, South Korea., Natural Hazards, Vol. 72(2), pp: 1189–1217. Seo K.H. and Lee L.J. (2011): A White Book of Changma. Korea Meteorological Administration, 268 pp. Tarolli, P., Borga, M., and Dalla Fontana G. (2008): Analyzing the influence of upslope bedrock outcrops on shallow landsliding, Geomorphology, vol.93, pp: 186–200. Tarolli, P., Borga, M., Chang, K. T., Chiang, S. H. (2011): Modeling shallow landsliding susceptibility by incorporating heavy rainfall statistical properties. Geomorphology, Vol: 133(3), pp: 199-211. Taylor, D. W. (1948): Fundamentals of Soil Mechanics, Wiley, New York. Turner A.K. (1996): Colluvium and talus. In: AK Turner and RL Schuster, (Eds.), Landslides Investigation and Mitigation. Special Report, vol. 247, pp: 525–554.
Giampilieri catchment, Nat.Hazards Earth Syst. Sci., 12, 1295–1309, doi: 10.5194/nhess-12-1295-2012. Barling, D. B., Moore, I. D., and Grayson, R. B. (1994): A quasi-dynamic wetness index for characterizing the spatial distribution of zones of surface saturation and soil water content, Water Resour. Res., vol.30, pp: 1029–1044. Borga, M., Dalla Fontana, G., and Cazorzi, F. (2002): Analysis of topographic and climatic control on rainfall-riggered shallow landsliding using a quasi-dynamic wetness index, J. Hydrol., vol.268, pp: 56–71. Calvello M., Lasco M., Vassallo R., and Di Maio C. (2005): Compressibility and residual shear strength of smectitic clays: influence of pore aqueous solutions and organic solvents. Italian Geotechnical Journal. vol.1, pp: 34–46. Carrara A., Cardinali M., Detti R., Guzzetti F., Pasqui V., and Reichenback P. (1991): GIS techniques and statistical models in evaluating landslide hazard. Earth Surf. Processes Landforms, vol.16. pp: 427–445. Claessens, L., Schoorl, J. M., Veldkamp, A. (2007): Modelling the location of shallow landslides and their effects on landscape dynamics in large watersheds: an application for Northern New Zealand. Geomorphology, 87(1), 16-27. Ceresetti, D., Molinié, G., and Creutin, J.-D. (2010): Scaling properties of heavy rainfall at short duration: A regional analysis, Water Resour. Res., vol.46, W09531, doi: 10.1029/2009WR008603. Chung Y.S., Yoon M.B., and Kim H.S. (2004): On climate variations and changes observed in South Korea. Climatic Change. Vol. 66, pp: 151–161. Guzzetti F., Reichenbach P., Cardinali M., Galli M., and Ardizzone F. (2005): Probabilistic landslide hazard assessment at the basin scale. Geomorphology vol.72, pp: 272–299. Haefeli, R. (1948): The stability of slopes acted upon by parallel seepage, paper presented at International Conference on Soil Mechanics and Foundation Engineering. Kim, W., Kim, K., Chae, B., Cho, Y. (2000): Case study of landslide types in Korea. J Eng Geol, vol.10, pp: 18–35. Koutsoyiannis, D., Kozonis, D., and Manetas, A. (1998): A mathematical framework for studying rainfall intensity-duration-frequency relationships, J. Hydrol., vol.206, pp: 118–135. Krzeminska, D. (2013): The influence of fissures on landslide hydrology, PhD Thesis, University of Delft.
Received: 31 December, 2014 Accepted: 26 January, 2016
1
24
