Spatial data for landslide susceptibility, hazard, and vulnerability assessment: An overview

Landslide susceptibility mapping at Golestan Province, Iran: A comparison between frequency ratio, Dempster-Shafer, and weights-of-evidence models

The main objective of the present study was to compare the performance of a classifier that implements the Logistic Regression and a classifier that employs a Naive Bayes algorithm in landslide susceptibility assessments. The study provides an evaluation concerning the influence of model's complexity and the size of the training data, while it identifies the most accurate and reliable classifier. The comparison of the two classifiers was based on the assessment of a database containing 116 sites located at the mountains of Epirus, Greece, where serious landslides events have been encountered. The sites are classified into two categories, non-landslide and landslide areas. The identification of those areas was established by analysing airborne imagery, extensive field investigation and the examination of previous research studies. The geo-environmental conditions in those locations where analyzed in regard with their susceptibility to slide. In particular, seven variables where analyzed: engineering geological units, slope angle, slope aspect, mean annual rainfall, distance from river network, distance from tectonic features and distance from road network. Multicollinearity analysis and feature selection was implemented in order to estimate the conditional independence among the variables and to rank the variables according to their significance in estimating landslide susceptibility. By the above processes the construction of nine different datasets was accomplished. Further partition allowed creating subsets of training and validating data from the original 116 sites. Each dataset was characterized by the number of the variables used and the size of the training datasets. The comparison and validation of the outcomes of each model was achieved using statistical evaluation measures, the receiving operating characteristic and the area under the success and predictive rate curves. The results indicated that model's complexity and the size of the training dataset influence the accuracy and the predictive power of the models concerning landslide susceptibility. In particular, the most accurate model with high predictive power was the eighth model (five variables and 92 training data), with the Naive Bayes classifier having a slightly higher overall performance and accuracy than the Logistic Regression classifier, 87.50% and 82.61% on the validation datasets, respectively. The highest area under the curve was achieved by the Naive Bayes classifier for both the training and validating datasets (0.875 and 0.806 respectively) while the Logistic Regression classifier achieved a lower AUC values for the training and validating datasets (0.844 and 0.711, respectively). When limited data are available it seems that more accurate and reliable results could be obtained by generative classifiers, like Naive Bayes classifiers. Overall, landslide susceptibility assessments could serve as a useful tool for the local and national authorities, in order to evaluate strategies to prevent and mitigate the adverse impacts of landslide events.

Comparison of a logistic regression and Naïve Bayes classifier in landslide susceptibility assessments: The influence of models complexity and training dataset size

The main objective of this study is to evaluate and compare the results of evidential belief functions and fuzzy logic models for spatial prediction of landslide hazards in the Hoa Binh province of Vietnam, using geographic information systems. First, a landslide inventory map showing the locations of 118 landslides that have occurred during the last ten years was constructed using data from various sources. Then, the landslide inventory was randomly partitioned into training and validation datasets (70% of the known landslide locations were used for training and building the landslide models and the remaining 30% for the model validation). Secondly, nine landslide conditioning factors were selected (i.e., slope, aspect, relief amplitude, lithology, landuse, soil type, distance to roads, distance to rivers and distance to faults). Using these factors, landslide susceptibility index values were calculated using evidential belief functions and fuzzy logic models. Finally, landslide susceptibility maps were validated and compared using the validation dataset that was not used in the model building. The prediction-rate curves and area under the curves were calculated to assess prediction capability. The results show that all the models have good prediction capabilities. The model derived using evidential belief functions has the highest prediction capability. The model derived using fuzzy SUM has the lowest prediction capability. The fuzzy PRODUCT and fuzzy GAMMA models have almost the same prediction capabilities. In general, all the models yield reasonable results that may be used for preliminary landuse planning purposes.

Spatial prediction of landslide hazards in Hoa Binh province (Vietnam): A comparative assessment of the efficacy of evidential belief functions and fuzzy logic models

Landslide susceptibility and hazard assessments are the most important steps in landslide risk mapping. The main objective of this study was to investigate and compare the results of two artificial neural network (ANN) algorithms, i.e., multilayer perceptron (MLP) and radial basic function (RBF) for spatial prediction of landslide susceptibility in Vaz Watershed, Iran. At first, landslide locations were identified by aerial photographs and field surveys, and a total of 136 landside locations were constructed from various sources. Then the landslide inventory map was randomly split into a training dataset 70 % (95 landslide locations) for training the ANN model and the remaining 30 % (41 landslides locations) was used for validation purpose. Nine landslide conditioning factors such as slope, slope aspect, altitude, land use, lithology, distance from rivers, distance from roads, distance from faults, and rainfall were constructed in geographical information system. In this study, both MLP and RBF algorithms were used in artificial neural network model. The results showed that MLP with Broyden–Fletcher–Goldfarb–Shanno learning algorithm is more efficient than RBF in landslide susceptibility mapping for the study area. Finally the landslide susceptibility maps were validated using the validation data (i.e., 30 % landslide location data that was not used during the model construction) using area under the curve (AUC) method. The success rate curve showed that the area under the curve for RBF and MLP was 0.9085 (90.85 %) and 0.9193 (91.93 %) accuracy, respectively. Similarly, the validation result showed that the area under the curve for MLP and RBF models were 0.881 (88.1 %) and 0.8724 (87.24 %), respectively. The results of this study showed that landslide susceptibility mapping in the Vaz Watershed of Iran using the ANN approach is viable and can be used for land use planning.

/pdf/landslide-susceptibility-mapping-at-vaz-watershed-iran-using-sargdzvm0m.pdf

Landslide susceptibility mapping at Vaz Watershed (Iran) using an artificial neural network model: a comparison between multilayer perceptron (MLP) and radial basic function (RBF) algorithms

Landslides in Himalaya cause widespread damage in terms of property and human lives. It the present study, an attempt is made to derive information on causative parameters and preparation of landslide-susceptible map using fuzzy data integration in one of the seismically active region of Garhwal Himalaya that was recently devastated by a huge landslide. High-resolution remotely sensed data products acquired from Indian Remote Sensing Satellite before and after the landslide event were processed to improve interpretability and derivation of causative parameters. Spatial data sets such as lithology, rock weathering, geomorphology, lineaments, drainage, land use, anthropogenic factor, soil type and depth, slope gradient, and slope aspect were integrated using fuzzy gamma operator. The final map was reclassified in to five classes such as highly to lowly susceptible classes based on cumulative cutoff. The result shows around 72% of known landslide areas including the large Uttarkashi landslide in the high and very high susceptibility classes comprising of only 37% of the total area. The precipitation data from ground- and satellite-based observations were compared; the precipitation threshold and the role of seismic activity were analyzed for initiation of landslide.

Fuzzy-based method for landslide hazard assessment in active seismic zone of Himalaya

Debris flows, also referred to as mudflows are a common type of fast-moving landslide that generally occurs during intense rainfall on water saturated soil. Vegetation is an important factor influencing the occurrence of rainfall-triggered landslides. The study attempts to numerically simulate the transient hydrological conditions and resultant slope instability conditions occurring in Tikovil River basin of Kerala, India and thereby quantify the effect of vegetation on the initiation of debris flows in the region. The model used for the study was STARWARS+PROBSTAB (van Beek, 2002) and is realized in a dynamic GIS (PCRaster) environment.. The model was modified and additional parameters such as root induced cohesion and vegetation surcharge were incorporated. Root-induced cohesion was calculated using an empirical equation suggested by researchers from US Geological Survey. Hydrological Effects of vegetation such as interception and bulk throughfall were computed from MODIS-derived 16-day composite NDVI data. Interception was computed for every day using Aston’s equation. Reference Evapotranspiration (RET) was computed using Hargreaves method and later adjusted to Actual Evapotranspiration using Crop Factors derived from FAO Database. Calibration was carried out using observed ground water height data for the area. The model simulated transient hydrological and slope stability conditions on a daily time step, with and without considering vegetation effects. A detailed investigation revealed that the hydrological effects of vegetation are crucial for the long term stability of the study area. However, its effects on slope stability during high intensity rainfall are negligible. Mechanical effect of vegetation, especially root induced cohesion is the most significant effect of vegetation on slope stability. The model was also able to capture the relative difference of instability in both the years; year 2000 had less number of slides when compared to year 2001. The research leads to the general conclusion that physically based spatial models are ideal to quantitatively understand the contribution of a specific parameter towards debris flow initiation. * Corresponding Author ** Presenting Author 2 International Symposium on Geo-information for Disaster Management (Gi4DM) Remote Sensing and GIS Techniques for Monitoring and Prediction of Disasters; September 25-26, 2006, Goa India Introduction Debris Flows are destructive events caused when eroded and other loose geological materials are mixed with water to the point where they begin to move down a gradient as one semi-cohesive mass, usually defined where sediment concentrations are greater than 60% by volume or 80% by weight (Vallance and Scott, 1997). Often initiated by shallow landslides, they are the primary agent of landscape evolution and the dominant denudational process in humid forested areas (Iida, 1999). In Kerala they are known as ‘Urul Pottal’. The characteristic pattern of these phenomena is the swift and sudden down slope movement of highly water saturated overburden containing a varied assemblage of debris material ranging in size from soil particles to huge boulders destroying and carrying with it every thing that is lying in its path (Sankar, 2005). The occurrence of debris flows is an increasing concern in Kerala. Much of the increase in debris flow activity is observed in areas with frequent changes in landuse/landcover. Studies on a 40,000 km area in the state estimate that the annual rate of deforestation is 1.16%. (Jha et al., 2000).. The densely populated state is prone to debris flows due to its geomorphic setting as 40% of it lies in the most prominent orographic feature of peninsular India, The Western Ghats. Floods and landslides killed almost a 100 people in different parts of Kerala and caused damages (Figure 1) to the tune of Rs 50 crore in 2005 (Ajith, 2005). Figure 1: Debris Flow havoc in Calicut District, 2005 (Photo by: G. Sankar) The removal of natural vegetation from the slopes has exposed them to heavy rainfall. So also farmers practice monoculture, especially rubber, tea and cardamom plantations, which has aggravated slope instability. The situation is particularly evident in the upper reaches of Idukki and Kottayam districts of Kerala. A study by the Center for Earth Science Studies identified that ‘among landuse types the areas with degraded natural vegetation shows maximum slide intensity’ (Thampi et al., 1998). This necessitates research to quantify the influence of vegetation on slope stability in the region. Vegetation effect on slope stability may be broadly classified as either hydrological or mechanical in nature. The mechanical factors arise from the physical interactions of either the foliage or root system of the plant with the slope. The hydrological mechanisms are those intricacies of the hydrological cycle that exist when vegetation is present (Greenway, 1987). There have been a number of attempts in dynamic spatial modeling of the effects of landuse/landcover on debris flows (van Asch et al., 1999; van Beek, 2002). Most of the studies consider the effect of change in vegetation cover on debris flows and landslides in terms of the effective rainfall that reaches the earth’s surface. However, studies elsewhere have identified several other effects of vegetation on soil erosion in general and landsliding in particular (Lancaster and Grant, 1999). Landslide research works in India received deserving attention in the year 1994 through the report presented by the Ministry of Agriculture, Govt. of India, to the world conference on the IDNDR held in Japan. It is seen that considerable work has been done by various national agencies in the Himalayan region, while limited attention has been paid to the Western Ghats region (Thampi et al., 1998). Majority of the work in India was carried out using statistical models and no work could be identified which tried to utilize a physically based dynamic model for understanding and quantifying the causative parameters. 2 International Symposium on Geo-information for Disaster Management (Gi4DM) Remote Sensing and GIS Techniques for Monitoring and Prediction of Disasters; September 25-26, 2006, Goa India Aim The study aims at quantification of the hazard of debris flow initiation considering vegetation effects in the upper catchment of Tikovil River (Figure 2) that flows through Kottayam and Idukki districts of Kerala (India), using a physically based spatial modeling approach. Methodology The models used for the study were based on earlier work by Van Beek (2002). The two models (STARWARS and PROBSTAB) are implemented in a raster based GIS with an advanced Environmental Modeling Language, called PCRaster (www.pcraster.nl). Figure 2: Study Area The two models are coupled and consider both hydrological and mechanical effects of vegetation on slope instability. The detailed methodology with major input parameters are pictorially represented in Figure 3 and the indicated steps are detailed below. Due to data deficiency the original coupled model was uncoupled and two more models were integrated. Similarly modifications were made to incorporate new parameters. The procedure is centered on 4 models, namely 1) Aston’s model, 2) STARWARS, 3) Effective Degree of Saturation and 4) PROBSTAB. The detailed methodology with major input parameters are pictorially represented in Figure 3 and the indicated steps are detailed below. Step 1 – Aston’s Model: this model computes not only Interception from Aston’s equation (Aston, 1979) on a daily time step but also derives Bulk Throughfall, computes the remaining amount of moisture in the soil that is available for evapotranspiration and the effective rainfall that reaches the surface and is available for percolation. The base data for the computation is MODIS NDVI 16 day composite and Rainfall on a daily time step. MODIS NDVI was used to compute Fractional Vegetation Cover using the empirical equation suggested by Walthall et al., (2004). Leaf Area Index was derived using the relationship suggested by Campbell and Norman (1998) (Walthall et al., 2004). Canopy storage was computed from Leaf Area Index (LAI) using the equation proposed by Von Hoyningen-Huene (1981). The bulk throughfall coefficient was computed as suggested by van Beek (2002). The interception deducted rainfall, reaches the surface and is stored as surface detention. The computed outputs of the model are the remaining amount of moisture in the soil that is to be evapotranspired and the effective rainfall that reaches the surface and is available for percolation. These are inputs for the next model: STARWARS. Step 2 – STARWARS: this is an acronym for Storage and Redistribution of Water on Agricultural and Re-vegetated Slopes. The model simulates the spatial and temporal dynamics of moisture content and perched water levels in response to gross rainfall and evapotranspiration. Both interception and surface detention are subjected to evaporation. The evapotranspiration is accounted using Hargreaves equation, instead of the original Penman’s equation used by the model designer (van Beek, 2002). The model predicts percolation through the unsaturated zone and lateral saturated flow over a semipervious bedrock contact. The outputs of the model are Water Level from the bed rock on a daily time step and Volumetric Moisture Content. Calibration of the model was carried out based on available water level data for the study area. Step 3 – Effective Degree of Saturation: To facilitate the calculation of the unsaturated hydraulic conductivity and to check on the available water, all storages are expressed as 2 International Symposium on Geo-information for Disaster Management (Gi4DM) Remote Sensing and GIS Techniques for Monitoring and Prediction of Disasters; September 25-26, 2006, Goa India Figure 3: Methodology Flow Chart 2 International Symposium on Geo-information for Disaster Management (Gi

Effect of vegetation on debris flow initiation : conceptualization and parameterization of a dynamic model for debris flow initiation in Tikovil river basin, Kerala, India, using PCRaster

R.C. Lakhera

Papers

Fuzzy-based method for landslide hazard assessment in active seismic zone of Himalaya

Effect of vegetation on debris flow initiation : conceptualization and parameterization of a dynamic model for debris flow initiation in Tikovil river basin, Kerala, India, using PCRaster

A Quick appraisal of ground deformation in indian region due to the october 8, 2005 earthquake, Muzaffarabad, Pakistan

Effect of vegetation on debris flow initiation : poster

Natural disasters and their mitigation : a remote sensing perspective