 |
 |
||||
|
|||||
|
|
|||||||
LANDSLIDE HAZARD ZONATION USING REMOTE SENSING AND GIS BASED WEIGHT OF EVIDENCE MODEL Navigation: Main page » Engineering Articles Author: GOUTHAM PRIYA CHAPTER 1 INTRODUCTION
1.1. GENERAL
1.1.1 LANDSLIDES
Landslides, one of the major catastrophes, always cause a major problem by killing hundreds of people every year besides damaging the properties and blocking the communication links. Most of the terrains in mountainous areas have been subjected to slope failure under the influence of a variety of terrain factors, and triggered by events such as extreme rainfall or earthquake. The frequency and the magnitude of slope failures can increase due to the human activities, such as deforestation or urban expansion. The problem of landslides becomes more aggravated, especially during the rainy season; through the main causative factors for the instability are often geological and geomorphological in nature. The problem of the stability of slopes, both natural and excavated, has to be faced in many fields of human activity, particularly in civil engineering. When slope stability is disturbed, a great variety of sliding movements takes place. Rapid movements of sliding rocks, separated from the underlying stationary part of the slope by a definite plane of separation, are designated as landslides in the strict sense. The sliding of slopes are not uncommonly caused by human activity, such as bad construction methods, etc. the great diversity of forms and the complexity of interrelationships, as well as practical relevance of landslides, can be recognized only by systematic and thorough study. Therefore landslide is a downward movement of a part of a slope, detritus or soil, along a sliding surface where shear failure occurs. They are a type of denudation caused by an interaction between external and internal earth forces and are closely related to climate.
1.1.2 CAUSES OF LANDSLIDES
In general, the factors which influence whether a landslide will occur typically include slope angle, climate, weathering, water content, vegetation, overloading, and geology and slope stability. How these factors are interrelated is important in understanding what causes landslides along with an understanding of the impact humans have on these factors by altering natural processes. Typically, a number of elements will contribute to a landslide, but often there is only one which triggers the movement of material. Natural causes include:
1.1.3 LANDSLIDE HAZARD ZONATION
Landslide hazard zonation (LHZ) refers to "the division of the land surface into homogenous areas or domains and their ranking according to degrees of actual/potential hazard caused by mass movement".
1.2. NEED FOR THE STUDY
The aim of such study is to use a "working methodology" in which geoenvironmental parameters are statistically analyzed to develop a model to predict landslide prone areas using remotely sensed data products and geographic information system (GIS). Population pressures & increasing urbanization along with the local zoning laws have been forcing man from flat lying areas & flood plains onto the adjacent hill-slopes. The hill-slope as an environmental setting can help us to avoid the severe flooding but can be equally hazardous because of potentially unstable earth materials. Landslides are common phenomena in a tectonically fragmented and sensitive mountainous terrain like the Nilgiris. The high susceptibility to landslides of the study area is mainly due to complex geological setting with contemporary adjustments in slope and relief, rainfall and anthropogenic effects. Therefore recognition of landslide prone topography is becoming increasingly important in landuse decisions.
CHAPTER 2 REVIEW OF LITERATURE
2.1. GENERAL
Literatures or research articles related to landslide hazard mapping is presented in this chapter. 2.2 RESEARCH ARTICLES John Mathew Jha et al. (2007) accessed the utility of weight of evidence model for landslide susceptibility mapping by using IRS-1D (LISS III & PAN), Topographical maps (1:50000). In this study, this model has been used to find out probability of occurrence of landslides for unique combinations of evidential themes. Lithology, structure, slope, slope aspect, landuse/landcover, drainage, distance to road are the evidential themes were considered. ROC analyses have shown 84.6% accuracy. Using land slide location and a spatial database containing information such as Topography, soil, forest, geology, land cover, lineament the weight of evidence model was applied to calculate each relevant factor's rating. Lee S et al. (2004) conducted Tests of conditional independence for the selection allowing 43 combinations of factors. He used Aerial photographs, Landsat TM imagery, IRS imagery, topographic maps, geological maps, soil maps and forest maps. The combination of slope, curvature, topography, timber diameter, geology & lineament showed the best results. The results were validated by calculating the correlation between the observed landslide location and the results. To examine the influence of various parameters like landuse/landcover, geology, slope, tectonics on landslides, PVSP Prasada Raju et al. (1999) used an in-house built software 'DECISION SPACE' to find the impact of each parameter on landslides using SPOT PLA190, IRS-1C PAN 1996 & SOI Toposheet. Decision rules such as Analytical hierarchy process (AHP) and factor of importance (FOI) were applied to assign weightages. Detection of translational landslides Barlow J. et al. (2003) used segmentation of Landsat ETM+ & DEM data in the northern Cascade Mountains, British Columbia. Aerial photography, Landsat ETM+, TRIM (1:20000), Field data were used. This study made attempts to overcome the problem of expensive and time-consuming aerial photographic interpretation. A hierarchical classification system was developed such that the normalized difference vegetation index (NDVI) and slope data eliminated all areas in the image that were both vegetated and on a gradient of less than 15. The remaining areas were classified using a supervised classification based on a spectral, geometric, and shape as criteria. The combination of data and techniques used in the detection of landslides for this study yielded an accuracy of 75% for errors of commission. A real-time prediction system for a global view of rainfall-triggered landslides using satellite remote sensing and geospatial datasets was developed by Yang Hong et al. (2007), by combining two necessary components: surface landslide susceptibility (LS) and a real-time space-based rainfall analysis system. First, a global LS map is derived. Second, an adjusted empirical relationship between rainfall-intensity and landslide occurrence is used to assess landslide hazards at areas with high susceptibility. A major outcome of this paper is the availability for the first time of a global assessment of landslide hazards. The effectiveness of this system is compared to several recent landslide events that occurred during the TRMM operational period. A GIS slope-stability model was applied to site-specific landslide prevention in Honduras. Zaitchik B.F et al. (2003), evaluated a simplified slope-stability model for developing landslide-susceptibility maps categorized according to local slope gradient & relative wetness the variables that most influence slope stability. Stability Index Mapping (SINMAP) is a physical slope-stability model. Stability predictions were empirically evaluated and zones of predicted instability were subsequently categorized according to the slope gradient derived from DEM and W based on steady-state hydrology for hurricane conditions. It indicates that site-specific management strategies required for slope stabilization in the study area. Spatially distributed model have provided a tool for rapid characterization of landslide susceptibility in large land areas. A further advantage is that slopes can be classified according to gradient & wetness through field survey when model is not available A new technique for quickly accessing extensive areas of large scale landslides that uses DEM extracted from high-resolution satellite images has been developed by Ken Tsutsui et al. (2007) using SPOT-5 images. The proposed technique observes the elevation changes by using multitemporal DEMs. Five-meter-resolution DEMs from SPOT-5 images are applied to two large-scale landslide disasters. Both events yielded elevation changes in excess of 10 m. we assess the DEM s produced by the proposed method and their landslide application. The developed technique well supports damage assessments of large-scale landslides because the location, depth, and volume can be quantitatively determined by remote sensing. The elevation diff accuracy depends on the slope angle. The proposed technique well estimated the average depth & volume of large-scale landslides. Landslide-hazard mapping using an expert system and a GIS was done by Kavitha Muthu et al. (2007). To create landslide warning & alert maps a rule-based expert system was used. GIS maps were used in vector format-geology, rainfall, and earthquake. The difference with other systems is in the use of change input data. The system is tested with 16 documented landslides. It can predict 11 of the 16 landslides to be among the 10% most at risk pixels in the study, which covers 100 sq km. A map containing the severity of the event class and a warning map is created by implementing a point system based on the accumulation of changes. Finally an alert map based on the combination of the above two maps is prepared. To analyze the hazard of landslides based on the data fusion to remotely sensed images of the same scene collected from multiple sources & further to classify & estimate the potential risks Yang-Lang chang et al. (2007). Multisensor, multitemporal, multiresolution, multifrequency image data were used. A framework for data fusion of multisource remotely sensed images, which consists of two approaches, referred to as the band generation process (BGP) and the positive Boolean function (PBF) classifier. PBF-hyper spectral image classification. A multiple adaptation BGP is introduced to create a new set of additional bands especially accommodated to landslide classes. The performance of the proposed method is evaluated by fusing Syteme Pour l' Observation de la Terre images and DEM information for land cover classification during the post 921 Earthquake period in Taiwan. The advantages are its discrete & nonlinear binary properties.
CHAPTER 3
OBJECTIVES OF THE STUDY
CHAPTER 4 STUDY AREA
4.1 GENERAL
The Nilgiris (meaning the Blue Mountains) is an ancient land mass thrust upwards at the junction of the two major mountain ranges near the southern end of India some 70 million years ago. 57% of the surface of the Nilgiri hills rises over 1000 m above the mean see level and 47% of that towers over 1800 m with the pinnacle formed by the Big Mountain at 2670 m. The Nilgiris, which is an administrative district of the state of Tamil Nadu, covers an area of 2478.63 km2. Described as "a cold tropical island rising above the warm tropical sea of southern India", the Nilgiris is an elevated physiographic zone - a massif. The Nilgiri massif is cloaked on its sides by dense growing on steep slopes, surmounted by the hilly plateau of grassland, woodland and savanna. The climatological and geo-ecological differentiations that abound within the ecosystem of the Nilgiris are considered remarkable by the scientific community. Although it is a relatively small region, the Nilgiri Massif shows a remarkable variability in landforms, soils, flora, fauna, microclimates, primates and human settlement patterns.
4.2 DESCRIPTION
Severe landslides happen every year in this area, especially during rainy season, causing a lot of damage to property and infrastructure, and inconvenience to the tourist. The total area accounts to 33 sq km. It lies between 11˚ 19' to 11˚ 21' N latitudes and 76˚ 48' to 76˚ 53' E longitudes. Criteria for route selection-tourist route, vulnerable area, geologically weak zones.
4.3 DATA USED
4.3.1 REMOTE SENSING DATA
4.4 SOFTWARE USED
CHAPTER 5 METHODOLOGY
5.1. GENERAL LANDSLIDES are one of the most common natural hazards in the Nilgiris terrain, causing widespread damage to property and infrastructure, besides loss of human lives almost every year. Appropriate management measures taken at the right time will reduce the risk from potential landslides. In order to prioritize the area for hazard-mitigation efforts, it is beneficial to have a Landslide Hazard Zonation (LHZ) map prepared depicting the ranking of the area based on actual and/or potential threat of slides in future. LHZ mapping is being carried out using qualitative or quantitative approaches. Statistical methods are based on the mathematical relationship between the observed landslides and their controlling factors. Such methods reduce the subjectivity and ensure better reproducibility of the hazard zonation processes and their outputs. The model developed for one area may not exactly give the same type of result in a different area. The statistical methods involve both bivariate as well as multivariate techniques for LHZ mapping. The commonly used multivariate statistical methods in landslide hazard assessment are linear regression, discriminant analysis and logistic regression. The bivariate statistical methods utilize the normalized landslide densities derived using the landslide occurrence in each parameter class to arrive at the hazard map. Information value method and weights of evidence modeling are two common bivariate methods applied in LHZ mapping process. The flow chart of the methodology of the present study is shown in Fig 5.1.
Figure 5.1 THE FLOW OF WORK
5.2. DESCRIPTION
For the land slide hazard analysis, the main steps were data collection and construction of a spatial database from which the relevant factors were extracted, followed by assessment of the landslide hazard using the relationship between landslide and landslide-related factors with validation of the results. A key assumption of this approach is that the potential (occurrence) possibility of landslides will be comparable with the actual frequency of landslides. Landslide occurrence areas were detected in the study area by interpretation of aerial photographs. Maps of recent landslides were developed from 1:20000 scale aerial photographs, and this was used to evaluate the frequency and distribution of landslides in the area. Topography, soil, forest, geology, land cover and lineament databases were constructed for the analysis. Maps relevant to landslide occurrence were constructed from a vector type spatial database using ARC/INFO GIS software. A digital map elevation model (DEM) was created using the topographic database. Slope, aspect and curvature, which are relevant to the landslide analysis, were calculated from the DEM. Soil texture. Soil material, soil drainage, soil effective thickness and topographic types were extracted from the soil database. The land-use data and the lineament were detected from IRS imagery with 5*5m resolution, using the expertise of a structural geologist, and distance from lineament was calculated by buffering. Using the detected landslide locations and the constructed spatial database, landslide analysis methods were applied and validated. For this, the calculated and extracted factors were mapped to a 5*5m grid in ARC/INFO GRID format. Next, using the weights-of-evidence method spatial relationships between the landslide location and each of the landslide- related factors such as topography, Soil and geology are found. Land cover was analyzed. The spatial relationships were used as each factors rating in the overlay analysis. Subsequently, test of conditional independence were performed for the selection of the factors to be used in landslide susceptibility mapping. The factors ratings were summed to calculate a landslide hazard index. Finally, the results of different combinations were validated using landslide location.
5.3. WEIGHT OF EVIDENCE MODEL
Weights of evidence model is used to predict a hypothesis about occurrence of an event based on combining known evidence in a study area where sufficient data are available to estimate the relative importance of each evidence by statistical methods. A weight of evidence model is a data-driven predictive model that differs from other knowledge driven predictive models. The distinction between a data-driven model and a knowledge driven model is apparent in that the weight of evidence model relies on objective assessment of input data to "estimate the relative importance of evidence by statistical means". Rather than employing knowledge driven subjectivity in populating other similar map models, Weights of evidence model uses the training data layer to make suitable adjustments to the mechanics of the model itself. The ultimate goal of this form of map modeling is to predict the likelihood of the occurrence of a particular phenomenon within a certain study area based on one or more layers of evidence. Weights of evidence modeling employ a log-linear form of Bayesian conditional probabilities that requires evidence in the form of discrete map layers and training data to prime the model. Although this method has been designed specifically for mineral potential mapping, it can also be used for other types of spatial prediction in which the goal is to predict the probability of occurrence of point objects, for example, GIS-based weights of evidence model for prediction flowing well. The method was originally developed for a nonspatial application in medical diagnosis, in which the evidence consisted of a set of symptoms and the hypothesis was of the type "this patient has disease x". For each symptom, a pair of weights was calculated, one for presence of the symptom, one for absence of the symptom. The magnitude of the weights depended on the measured association between the symptom and the pattern of disease in a large group of patients. The weights could then be used to estimate the probability that a new patient would get the disease, based on the presence or absence of symptoms. A weight of evidence was adapted in the late 1980s for mineral potential mapping with GIS. In this situation, the evidence consists of a set of exploration datasets (maps), and the hypothesis is "this location is favorable for occurrence of deposit type x". Weights are estimated from the measured association between known mineral occurrences and the values on the maps to be used as predictors. The hypothesis is then repeatedly evaluated for all possible locations on the map using the calculated weights, producing a mineral potential map in which the evidence from several map layers is combined. The method belongs to a group of methods suitable for multi-criteria decision making. The following formulation of the Bayesian probability model, known as the weights-of evidence model, was supplied to landslide susceptibility analysis as synthesized from John Mathew et al. (2007), Lee et al. (2004) and Prasada Raju et al. (1999). For a given number of cells N {D}, containing an occurrence, D and the total number of cells in the study area, N {T}, the priority probability of an occurrence is expressed by P {D} = N {D}/N {T} (Eqn 5.1) Suppose that a binary predictor pattern, B. occupying N {B} cells, occurs in the region and that a number of known landslides occur preferentially within the pattern, i.e., N {D∩B}, then the favorability of locating an occurrence given the presence of a predictor and the absence of a pattern can be expressed by the conditional probabilities: P {D|B} = P {D∩B}/P {B} = P {D}* {P {D|B}/P {B}} (Eqn 5.2) P {D|B} = P {D∩B}/P {B} = P {D}* {P {B|D}/P {B}} (Eqn 5.3) The posterior probability of an occurrence given the presence and absence of the predictor pattern is denoted by P {D׀B} and P {D׀B}, respectively; and P {B׀D} and P {B׀D} are the posterior probabilities of being inside and outside the predictor pattern B, respectively, given the presence of an occurrence D. Also, P {B} and P {B} are the prior probabilities of being inside and outside the predictor pattern. The same model can be expressed in an odds-type formulation, where the odds, O, are defined as O = P/ (1-P). Expressed as odds, equations (5.2) and (5.3) respectively, become: O {D|B} = O {D}* {P {B|D}/P {B|D}} (Eqn 5.4) O {D|B} = O {D}* {P {B|D}/P {B|D}} (Eqn 5.5) Where O {D׀B} and O {D׀B} are the posterior odds of an occurrence, given the presence and absence of a binary predictor pattern, respectively, and O {D} represents the prior odds of an occurrence. The weights for the binary predictor pattern are defined as: W+ = Loge {P {B|D}/ P {B|D}} (Eqn 5.6) W- = Loge {P {B|D}/ P {B|D}} (Eqn 5.7) Where W+ and W- are the weights of evidence when a binary predictor pattern is present and absent, respectively. Hence Loge O {D|B} = Loge O {D} + W- (Eqn 5.8) Loge O {D|B} = Loge O {D} + W- (Eqn 5.9) Suppose that are two binary predictor patterns, B1 and B2. It can be shown that the posterior probability of an occurrence, given the presence of the two-predictor patterns, is P {B1∩ B2|D} P {D} P {D| B1∩ B2} = ------------------------------------------------ (Eqn 5.10) P {B1∩ B2|D} P {D} +P {B1∩ B2|D} P {D}
If B1 and B2 are conditionally independent of each other with respect to a set of occurrences, then it indicates that the following relation is satisfied: P {B1∩ B2|D} =P {B1|D} P {B1} (Eqn 5.11) This shows equation (5.10) to be simplified as P {B1/D} P {B2/D} P {D| B1∩ B2} =P {D} ----------------------- (Eqn 5.12) P {B1} P {B2} Similarly, if more than two binary predictor patterns are used, they can be added, provided they are also conditionally independent of one another with respect to the occurrences. Thus, with Bj (j=1, 2…n) binary predictor patterns, the Loge posterior odds are Loge O {D|Bk1∩ Bk2∩B3……Bkn} =∑ Wkj = Loge O {D} (Eqn 5.13) Where the superscript k is positive (+) or negative (-) depending on whether the binary predictor pattern is present or absent, respectively. The posterior odds can be converted to posterior probabilities, based on the relationship P= O/ (1+O), to represent the favorability of locating an occurrence. Equation (13) defines the relationship of two binary patterns, B1 and B2 that are conditionally independent of each other with respect to a set of points. Algebric manipulation shows that equation (5.13) is equivalent to N {B1∩D} N {B2∩D} N {B1∩ B2∩D} = ------------------------ (Eqn 5.14) N {D} The left side of equation (5.14) is the observed number of occurrences in the overlap zones of B1 and B2. The right side is predicted number of occurrences in the overlap zone. This relationship leads to contingency table calculation for the pair-wise testing of conditional independence.
5.4 ADVANTAGES OF WEIGHT OF EVIDENCE MODEL
A weight of evidence model is relatively straightforward and can handle the problem of missing data in analysis. The only requirement is the conditional independence of input variables The quantitative methods in LHZ involve statistical, geotechnical and artificial neural network methods. Geotechnical approaches include the physical processes involved in landsliding and express the hazard in the form of safety factors. Statistical methods are based on the mathematical relationship between the observed landslides and their controlling factors. Such methods reduce the subjectivity and ensure better reproducibility of the hazard zonation processes and their outputs. The model developed for one area may not exactly give the same type of result in a different area. The statistical methods involve both bivariate as well as multivariate techniques for LHZ mapping. The commonly used multivariate statistical methods in landslide hazard assessment are linear regression, discriminant analysis and logistic regression. The bivariate statistical methods utilize the normalized landslide densities derived using the landslide occurrence in each parameter class to arrive at the hazard map. Information value method and weights of evidence modelling are two common bivariate methods applied in LHZ mapping process.
5.5 ARC-SDM
In this thesis, Arc-SDM (Spatial Data Modeller for ArcGIS) was used to derive at landslide hazard zonation map. ArcSDM is a free extension of ArcGIS, which provides tools for weights of evidence, logistic regression, fuzzy logic and neural network analysis. For the prediction of landslide susceptibility based on morphological and geological factors, the method called Weights of Evidence was chosen. It is based on the assumption that future landslides would be triggered or influenced by the same or similar controlling factors as already registered landslides. In ArcGIS Desktop, the method is accessible as part of Natural Resources Canada's ArcGIS extension ArcSDM 3.1, which is freely available from ntserv.gis.nrcan.gc.ca/sdm. In the present project, a set of seven factors was analyzed for correlation with landslide events:
The layers were converted to 10-meter rasters where necessary, and several terrain parameters were derived from the DEM using ArcGIS Spatial Analyst. In ArcSDM, each of the factors was then weighted according to its relevance for the presence of landslides. Once the weighting was completed for all factors, the weights were combined to compute the probability for landslides over the whole area on a regular grid. To convert the probability map into a hazard map and make the result more helpful for decision makers, the probability values ranging from 0 to 1 were reclassified in three hazard zones (hazard, hazard cannot be excluded, and no hazard).
CHAPTER 6 ANALYSIS AND INTERPRETATION
6.1 GENERAL
In the present study, weights of evidence modelling have been applied for landslide susceptibility mapping. Weights of evidence is a data-driven method which is basically the Bayesian approach in a log-linear form using the prior and posterior probability and is applied where sufficient data are available to estimate the relative importance of evidential themes by statistical means. This method had been initially applied to non-spatial, quantitative, medical diagnosis to combine evidence from clinical diagnosis to predict diseases. The spatial association between a set of evidential themes and a set of known landslide locations, which are expressed as the weights of evidence, is combined with the prior probability of occurrence of landslides to derive the posterior probability of occurrence of landslides, provided the evidential themes are conditionally independent with respect to the slides. This method consists of reducing each set of landslide-related factors on a map to a pattern of a few discrete states.
6.2 PREVIOUS LANDSLIDE LOCATIONS
Though it was difficult to identify the old stabilized slides directly on the Aerial Photographs, the locations were confirmed through interaction with local residents. Fifty slides (Figure-6.1) were mapped which vary in area from 0.08 to 15 ha. Most of the slides are close to the Kallar-Coonoor Road (Table-6.1). The centers of the raster cells representing the landslide locations have been converted to a point-shape file which has been considered as the sample location input.
Table 6.1 Previous Landslide Locations S NO LANDSLIDE NAME LATITUDE LONGITUDE
1 Vannarapettai-Coonoor Road 11˚ 21' 15" 76˚ 47' 00" 2 Slide near Aravakadu 11˚ 22' 05" 76˚ 46' 00" 3 Katteri road slides 11˚ 20' 00" 76˚ 47' 10v 4 Slide of the Glendale estate or Glenmore Slide 11˚ 19' 30" 76˚ 46' 40" 5 Doddakombai Slide 11˚ 16' 00" 76˚ 39' 00" 6 Aliada Slides 11˚ 17' 30" 76˚ 44' 10" 7 Karadipallam Slides 11˚ 21' 00" 76˚ 46' 30" 8 Selas Slides 11˚ 19' 50" 76˚ 44' 55" 9 Slide in Manjakombai Village 11˚ 17' 00" 76˚ 41' 30" 10 Slide of Maliikorai village 11˚23' 00" 76˚ 47' 00" 11 St Mary's Hill colony ooty 11˚ 24' 30" 76˚ 42' 30" 12 Slides in Chamraj estate 11˚ 17' 30" 76˚ 40' 00" 13 Slides on the northeastern portion of the Elk hill 11˚ 24' 00" 76˚ 42' 30" 14 Slides on ooty-Doddabetta Road 11˚ 24' 20" 76˚ 43' 30" 15 Manjanakorai Slides 11˚ 23' 00" 76˚ 41' 40" 16 Dunsdale-Somasdale slide 11˚ 27' 30" 76˚ 34' 45" 17 Slide southwest of the agriculture farm 11˚ 25' 30" 76˚ 43' 30" 18 Slide 80 m downstream of tiger hill reservoir 11˚ 33' 40 " 11˚ 23' 40" 76˚ 43' 50" 76˚ 43' 55" 19 Slides north and south of tiger hill reservoir 20 Slides on the road linking north ooty and doddabetta road junction 11˚ 25' 40" 76˚ 43' 00" 21 Slides in the govt Chinchona estate on Idhuhatti 11˚ 26' 00" 76˚ 44' 15" 22 Karimundi road slides 11˚ 26' 00" 76˚ 43' 45" 23 Slide near govt chinchona office 11˚ 25' 30" 76˚ 44' 00" 24 Twin slides 200m west of the slide in the govt Cinchona office 11˚ 35' 20" 76˚ 43' 15" 25 Slide north east of Cinchona office 11˚ 35' 45" 76˚ 44' 00" 26 Slides on the eastern flank near a graveyard of kodappamund 11˚ 25' 15 " 76˚ 43' 00" 27 Slide north east of the koddapamund graveyard 11˚ 15' 16" 76˚ 43' 00" 28 Slide in the Eucalyptus forest north of the raj bhavan upper road 11˚ 25' 45" 76˚ 43' 00" 29 Slides on the 7018 hill range -15 kms southwest of Tuneri 11˚ 26' 45" 76˚ 43' 45" 30 Potential area beyond the junction of ooty-Kundah-Manjanakarai road 11˚ 23' 15" 76˚ 42' 00" 31 Slides near exit portal of railway tunnel 15 11˚ 23' 00" 76˚ 42' 30" 32 Slides near lovedale post office 11˚ 23' 00" 76˚ 42' 30" 33 Slides located near culvert 42/79 on ooty-coonoor road 11˚ 23' 10" 76˚ 43' 40" 34 Slide at 200m west of the Cinchona office 11˚ 35' 20" 76˚ 43' 15" 35 Yellanahalli church 11˚ 22' 30" 76˚ 44' 30" 36 Anaikarai slide 11˚ 27' 30" 76˚ 43' 15" 37 Slide on the Perhimund Dam 11˚ 22' 30" 11˚ 19' 30" 76˚ 34' 20" 76˚ 47' 15" 38 Runneymade Slide 39 Hospital slide 11˚ 19' 30" 76˚ 47' 00" 40 Coonoor Slide 11˚ 20' 30" 76˚ 47' 30" 41 Aravakkadu slide 11˚ 22' 30" 76˚ 46' 30" 42 Krodumund Slide 11˚ 23' 17 " 76˚ 35' 15" 43 Porthimund Dam slide 11˚ 22' 30" 76˚ 34' 20" 44 Emerald slide 11˚ 22' 20 " 76˚ 36' 37" 45 Kadcuppa slide 11˚ 21' 25" 76˚ 35' 15" 46 Porthimund RF Slide 11˚ 20' 45" 76˚ 33' 48" 47 Tulitalai slide 11˚ 20' 32" 76˚ 40' 08" 48 Bala kola Slide 11˚ 20' 28" 76˚ 41' 07" 49 Belithala Slide 11˚ 49' 02" 76˚ 38' 42" 50 Naraidu betta Slide-I 11˚ 17' 15" 76˚ 34' 42" 51 Naraidu betta Slide-II 11˚ 17' 15" 76˚ 34' 57" 52 Naraidu betta Slide-III 11˚ 17' 17" 76˚ 35' 07" 53 Mainalaimattam 11˚ 17' 22" 76˚ 40' 02" 54 Carigmore slide 11˚ 23' 25" 76˚ 43' 10" 55 Manthada slide 11˚ 26' 25" 76˚ 43' 37" 56 Bellattimattam slide 11˚ 22' 22" 11˚ 23' 07" 76˚ 50' 30" 76˚ 23' 15" 57 Halakarai slide 58 Paiyangi slide 11˚ 26' 20" 76˚ 48' 50" 59 Slide at Gudalur 11˚ 30' 00" 76˚ 30' 00" 60 Small slides occurred on kallar-Coonoor Railway road ooty 11˚ 19' 00" 76˚ 46' 00" 61 Slide at 2km southwest of Coonoor 11˚ 20' 00" 76˚ 46' 53" 62 Slide at 2.5 km East of Aravankadu 11˚ 22' 30" 76˚ 46' 06" 63 Slide at 5 km east of Manjanakarai 11˚ 23' 20" 76˚ 46' 05" 64 Marapallam Slide 11˚ 20' 00" 76˚ 48' 55" 65 Vulnerable spot 11˚ 15' 20" 76˚ 45' 50" 66 Vulnerable spot 11˚ 20' 25" 76˚ 45' 50" 67 Vulnerable spot 11˚ 25' 30" 76˚ 45' 50" 68 Vulnerable spot 11˚ 20' 25" 76˚ 45' 50" 69 Vulnerable spot 11˚ 20' 30" 76˚ 50' 51" 70 Vulnerable spot 11˚ 15' 20" 76˚ 35' 40" 71 Vulnerable spot 11˚ 20' 25" 76˚ 35' 40" 72 Vulnerable spot 11˚ 25' 30" 76˚ 56' 10" 73 Vulnerable spot 11˚ 25' 30" 76˚ 35' 40" 74 Vulnerable spot 11˚ 15' 20" 76˚ 40' 45"
Figure 6.1 Previous Landslides Locations in the Study Area
6.3 INCIDENCE OF LANDSLIDES IN INDIA
Table 6.2 Vulnerable Areas in India REGION INCIDENCE OF LANDSLIDES Himalayas High to Very high North-Eastern Hills High Western Ghats and Nilgiris Moderate to high Cattle Cost Low Vindhyachal Low
6.4 GENERATION OF EVIDENTIAL THEMES
6.4.1 GEOMORPHOLOGY
The landscape of the Southern Indian tectonic shield is believed to have evolved through a slow geomorphic process which arose as a result of the movement of the peninsular region into the rest of the Asian mainland and the resulting geologic transformations (Figure6.2). Along with this, the peninsula also experienced an eastward tilt which changed the pattern of drainage. Hills are generally of elevations between 600 and 1000 m. However there are higher hills of 1000- 2000 m between 8 and 13 N and 18- 19 N. Peaks over 2000 m are found only in the Nilgiris. Figure 6.2 Geomorphology map
6.4.2 SOIL
The soil mainly consists of the derivatives of the ancient metamorphic rocks in India, rich in iron and manganese (Figure-6.3). There are exposed lateritic rocks along the coastal hills which appear black and are barren and mostly unfit for plant growth. Some granitic rocks are also present towards the southern parts of the district. These are a unique feature for the Western Ghats; they are, however, common in the forests of South - East Asia.
Figure 6.3 Soil Map
6.4.3 SLOPE
Slope is a very important parameter in any landslide hazard zonation mapping. If the slope is higher then there is a chance of occurrence of landslide. Slope map (Figure-6.5) has been created from the DEM generated in the PHOTOMOD software (Figure-6.4). Contour maps (Figure-6.5) are also generated. In the study area slope varies from 0° to more than 54°.
Figure 6.4 DEM of the Study Area Figure 6.5 Slope Map
6.4.4 LANDUSE
Presence of vegetation is crucial in slope stability due to better bonding of the slope material. Thus slopes with dense vegetation should be less prone to the occurrence of shallow landslides than barren slopes, while all other factors remain constant. In order to assess the contribution of different land use/land cover classes to slope destabilization, this factor has also been taken into account. Figure 6.6 Landuse Map
6.4.5 GEOLOGY
Nilgiris district is located in the plateau region. Structurally, the Nilgiris plateau belongs to continental block of peninsular India and the mountain ranges comprises of archean metamorphic rocks like Charnochite, Biotite gneiss, Magnetic quartzite, Hornblend granite along with some intrusive bodies like pegmatite,dolerite and quratz veins. Laterites are found in large quantities in the district.The laterite found over the charnochites is hard. Bauxite is the other mineral found in the district. Structurally the area is highly disturbed and is subjected to faulting. The major rivers in the district flow along the prominent fractures. The prominent fractures in the district are trending East-North, East and North-North West. Geology is not considered for our study since it is uniformly distributed in the study area.
6.4.6 DISTANCE TO ROAD
Other Evidential themes like River and Road maps were created in Arc GIS. It has been observed that many landslides occur close to the road. It is possible that the slope destabilization has been caused either by the uncontrolled or controlled blasting and widening of the roads, or by the loss of support due to removal of material from the lower portion of the slopes during road construction. In order to accommodate the effect of this anthropogenic activity, distance to road has been taken as an evidential theme.
Figure 6.7 Road Buffer Map
6.4.7 RIVER Figure 6.8 River Map
6.5 ASSIGNING RANKS BY QUALITATIVE METHOD
The final landslide zonation map was prepared from various thematic maps by applying weighted overlay analysis in GIS environment. Thus the landslide zonation map indicates the whole study area has been divided into 4 zones and the weights has been according to the table 6.3
The role of geology and structure is limited in the formation of landslides. The geology of the region has a bearing on the origin and types of soils and has little to do with landslides.
Table 6.3 Assignment of Weights and Triggering Criterion THEME RANK 1 (4*WEIGHT) RANK 2 (3*WEIGHT) RANK 3 (2*WEIGHT) RANK 4 (1*WEIGHT) Geomorphology Hills and Plateaus Piedmont zone - - Slope 35-50 deg 15-35 deg >50 deg 0-15 deg Soil Inceptisols Hill soil Alfisols Entisols Landuse Agriculture Waste land Forest Built up Land
6.6 GENERATION OF LHZ MAP BY QUALITATIVE METHODS
In order to generate the landslide prone areas a model has been developed in a GIS environment. Data in the form of thematic layers such as slope, soil and land use were input into GIS. Slope map has been derived from DEM (Digital Elevation Model) and land use map. The overlay analysis has been done. The assigning weights and criterion table 6.3 has been prepared as Look up Table (LUT) and linked with union coverage. Finally the landslide prone areas zonation was delineated (Figure-6.10) Figure 6.9 Landslide Hazard Zonation Map by Qualitative Methods
6.7 CALCULATION OF WEIGHTS USING WEIGHT OF EVIDENCE MODEL
Open Spatial Data Modeler for ArcGIS. Figure 6.10 Spatial Data Modeler Open Spatial Data Modeller--->Pre-processing-> Set analysis Parameters to set analysis Parameters.
Figure 6.11 Set Analysis Parameters Open Spatial Data Modeller--->Wofe-LR-> Calculate Theme weights to calculate theme weights Figure 6.12 Evidential Theme Weights Select all evidential themes as input. Then click "Calculate All" to calculate weights for each evidential theme. Figure 6.13 Inputs to Weight of Evidence Model At last, Open Spatial Data Modeller--->Post Processing-> Associate responses with point theme to generate a potential map Figure 6.14 Generation of Response theme Click "OK" and this will create a posterior probability map showing the zone with high probabilityof landslide occurrences. The red means high probability and blue means low probability. Figure 6.15 Landslide Hazard Zonation Map by Quantitative Methods Table 6.4 Weights of Evidence Analysis between Slope and Landslide CLASS LANDLIDE POINTS % W+ W- C C/S© 0-1 1 2.070 -2.278 0.205 -2.483 -7.771 1-3 3 0.621 -2.771 0.098 -2.869 -4.954 3-5 2 1.863 -1.755 0.095 -1.851 -5.500 5-10 3 6.625 -0.266 0.022 -0.288 -1.573 10-15 8 7.867 -0.137 0.013 -0.150 -0.888 15-35 21 18.219 0.719 -0.108 0.827 7.016 35-50 27 19.255 0.940 -0.136 1.076 9.318 >50 10 19.462 1.038 -0.145 1.183 10.293
Table 6.5 Weights of Evidence Analysis between Geomorphology and Landslide CLASS LANDSLIDE POINTS % W+ W- C C/S© Deflection Slope 27 7.039 0.861 -0.043 0.904 2.199 Highly Dissected 33 70.186 0.860 -0.858 1.718 1.842 Moderately Dissected 12 4.141 -1.195 0.105 -1.300 5.694 Undissected/Less Dissected 3 0.828 -2.880 0.151 -3.032 -6.038
Table 6.6 Weights of Evidence Analysis between Soil and Landslide CLASS LANDSLIDE POINTS % W+ W- C C/S© Alfisols 19 4.141 -1.195 0.105 -1.300 -5.694 Entisols 0 0.000 - 0.000 0.000 - Forest 4 0.828 -2.880 0.151 -3.032 -6.038 Hill soil 12 12.836 -0.137 0.022 -0.158 -1.164 Inceptisols 40 78.675 0.689 -1.042 0.731 15.581
Table 6.7 Weights of Evidence Analysis between Landuse and Landslide CLASS LANDSLIDE POINTS % W+ W- C C/S© Agriculture 18 3.934 -1.631 0.184 -1.816 -7.758 Built-up Land 37 83.437 0.210 -0.671 0.881 7.198 Forest 20 12.629 0.213 -0.027 0.240 7.198 Waste Land 0 0.000 - 0.003 0.000 - Water 0 0.000 - 0.003 0.000 -
Table 6.8 Weights of Evidence Analysis between Distance to Road and Landslide CLASS LANDSLIDE POINTS % W+ W- C C/S© 500 28 37.333 -0.977 1.345 -2.322 -37.476
CHAPTER-7 RESULTS AND DISCUSSION
7.1 LANDSLIDE HAZARD ZONATION BY QUALITATIVE METHOD
Based on the above studies, two landslide hazard zonation maps has been prepared using Remote Sensing and GIS techniques. The entire study area has been divided into four zones of varying susceptibility to landslides viz low, moderate, high, and very high. Using this technique the areas which are vulnerable to landslides are predicted and suitable mitigation measures are proposed. The first landslide zonation map (Figure-6.10) was prepared from various thematic maps by applying different analysis in a GIS environment. Landslide hazard zonation map was prepared by integrating the effect of various triggering factors. The percentage of the different risk zones of the study area is given below. Table 7.1 Percentage of Area under each Zone by Qualitative method
DIFFERENT CATEGORIES
Percentage of area Zone of low susceptibility to landslides 37% Zone of moderate susceptibility to landslides 51% Zone of high susceptibility to landslides 5% Zone of very high susceptibility to landslides 7%
The risk zones of high and very high are incident mainly on forest, agriculture and tree plantation areas. These landslide zones occur in most of the study area due to changes in landuse/ landcover and indiscriminate deforestation in the forest area. To achieve the landslide zonation, the different factors were grouped according to their relative importance and land susceptibility values (LSV). Landslides are being essentially gravity-type hence; the degree of slope was accorded the prime importance. Taking all the factors into consideration and with an intimate knowledge of the Nilgiri landslides, an LSV of 40 was assigned to slope. The thickness of the soil was considered next in importance, as all the slides were soil slides of varying thickness. An LSV of 35 was assigned to this factor. Under similar topographic conditions with similar thickness of soil and type of drainage, the susceptibility to landslides is accelerated by human environment. Hence the land use practices adopted was also considered and an LSV of 25 was assigned to this.
7.2 LANDSLIDE HAZARD ZONATION BY QUANTITATIVE METHOD
The present study demonstrates the application of weights of evidence modeling for landslide susceptibility mapping in part of Nilgiris, which is prone to frequent occurrence of landslides. Remote sensing and GIS have been useful in data preparation and at integration stages. Positive and negative weights and contrast values have been calculated for various classes of parameters used in the study. Slope saturation might be the reason for this phenomenon. Amongst the different land use/land cover categories, scrub land area has shown highest contrast and this area is more prone to landslides due to lack of cohesion of slope material, as in case of dense forest areas. An agricultural area also shows high contrast values. This could be because of the fact that cultivation has been done on stabilized, old landslides, thereby giving high positive weights in such areas. The slope category of 35-50° is found to be significant contributors with respect to landslides in the study area. The 35-60° slope category is the most unstable in the study area, where most of the colluvial accumulations are found around 35° slope angle. On the other hand, slopes gentler than 30° are found to be stable in the area. It has also been observed that steep slopes (> 50°) stand out with rock exposures and are mostly stable, provided they are devoid of any geological discontinuities.
Table 7.2 Percentage of Area under each Zone by Quantitative method DIFFERENT CATEGORIES Percentage of area Zone of low susceptibility to landslides 41.182%
Zone of moderate susceptibility to landslides 47.226% Zone of high susceptibility to landslides 7.126% Zone of very high susceptibility to landslides 3.463%
The classified posterior probability map depicts 17.182% of the total study area in the low, 11.226% in the moderate 30.126% in the high and 41.463% very high hazard class. Thus weights of evidence modeling can be utilized for estimating the conditional probability of land sliding on a cell-by-cell basis for an area, given the presence or absence of various independent variables which influence slope stability, towards the LHZ mapping.
CHAPTER 8 CONCLUSIONS
8.1 CONCLUSIONS
This present study brings out a definite relationship between the Photogrammetry and GIS techniques, which play a significant role in landslide zonation mapping. Landslide identification, which is a crucial parameter for any regional landslide hazard assessment, can be very well done particularly with aerial photographs. Coupled with aerial photos, GIS is an excellent tool to display the spatial distribution of landslides along with their attributes. However, the landslide map so prepared should be validated with ground checks.
Management of the landslides disasters can be successful only when detailed knowledge is obtained about the expected frequency, character and magnitude of the mass movement in an area. The zonation of landslide hazard must be the basis for any landslide mitigation strategy and should supply planners and decision-makers with adequate and understandable information.
While dealing with landslide-hazard mitigation, the hazard planner is concerned mainly about the final outcome i.e., the zonation map. In preparing the landslide zonation map, synthesized and weighed the data pertaining to geology, slope morphometry, distribution of soils and land use pattern have been analyzed. The role of geology and structure is limited in the formation of landslides. The geology of the region has a bearing on the origin and types of soils and has little to do with landslides.
8.2 FUTURE RECCOMENDATIONS
The mission like Cartosat-2 by Government of India is very promising as it is capable to produce high-resolution images.
Quickbird-2 satellite with 0.61 m resolution in PAN mode can be effectively employed for landslide application. Terra SAR-X, which will ultimately replace the need for aerial photography, as it can give 2 m contour interval and 1 m planimetric spatial data. More research has to be undertaken in use of SAR in landslide application and the integrated study involving SAR Polarimetry, SAR Interferometry and optical data can be much more meaningful in zoning and monitoring landslides
REFERENCES
1. Abbasi, I.A., (2003) 'Slope failure and landslide mechanism, Muree area, north Pakistan', Geological Bulletin University of Peshwar, Vol.35. 2. Antonio zanutta, Paolo baldi, and Gabriele bitelli., (2006) 'Qualitative and Quantitative photogrammetric techniques for multi-temporal landslide analysis', Annals of Geophysics, Vol.95, 1067-1080. 3. Arianna pesci, Paolo baldi, and Alessandro bedin., (2004) 'Digital Elevation Models for landslide evolution monitoring: application on two areas located in the Reno river valley (Italy)', Annals of Geophysics, Vol.47, 1339-1352. 4. Barlow, j., (2003) 'Detecting translational landslide scars using segmentation of Landsat ETM+ and DEM data in the northern cascade mountains, British Columbia', Canadian Journal of Remote sensing, Vol.29, No.4, 510-517. 5. Biswajeet pradhan, Jasmi ab talib and Saro lee., (1999) 'Probabilistic and statistical Landslide hazard mapping using GIS and remote sensing at Cameron Highland, Malaysia'. 6. Eric a saczuk and James s. Gardner., (2002) 'Modeling landslide hazards in the Kullu Valley, India using GIS and remote sensing', IEEE Transactions on geosciences and remote sensing, 1148-1156. 7. Haufmann .v, Lieb g.k., (2003) 'Mapping the kinematics of the Blaubach landslide (Austria) using Digital photogrammetry', Geophysical Research abstracts, Vol.5. 8. John Mathew, Jha, V.K. And Rawat, G.S., (2007) 'Weight of evidence modeling for landslide hazard zonation mapping in part of Bhagirathi valley, uttarakhand', Current science, Vol.92, No.5, 628-638. 9. Karsli.F and Yalcin .M, (1991) 'Landslide assessment by using digital phhotogrammetric techniques', International association of Engineering Geology, Vol.43, 27-29. 10. Kavitha Muthu and Maria Petrou., (2007) 'Landslide-hazard mapping using an expert system and a GIS', IEEE Transactions on geosciences and remote sensing, Vol.45 No.2, 522-531. 11. Ken Tsutsui, (2007) 'Detection and volume estimation of large-scale landslides based on elevation-change analysis using DEMs extracted from high-resolution satellite stereo imagery', IEEE Transactions on geosciences and remote sensing, Vol.45, No.6, 1681-1695. 12. Lee, S. and Choi, J., (2004) 'Landslide susceptibility mapping using GIS and weight-of-evidence model', International Journal of Geographical information science, Vol.18, No.8, 789-814. 13. Peter .V. Gorsevski., (2001) 'Statistical modeling of landslide hazard using GIS', Proceedings of the seventh federal interagency sedimentation conference, 103-109. 14. Pvsp Prasada Raju and Saibaba,J., (1999) 'Landslide hazard zonation mapping using remote sensing and geographic information system techniques- A case study of Pithoragarh Area, U.P', IEEE Transactions on geosciences and remote sensing, 577-579. 15. Saha A.K and Arora M.K, (2005) 'GIS-based route planning in landslide-prone areas', International journal of geographical information science, Vol.19, No.10, 1149-1175. 16. Salah Sadek, Ramzi Ramadan, and Hani Naghi., (2000) 'A GIS-Based landslide hazard framework for road repair and maintenance', Journal of urban planning and development, 1-17. 17. Yang-Lang Chang, (2007) 'Multisource data fusion for landslide classification using generalized positive Boolean functions', IEEE Transactions on geosciences and remote sensing, Vol.45, No.6, 1697-1708. 18. Yang Hong., (2007) 'An experimental global prediction system for rainfall-triggered landslides using satellite remote sensing and geospatial datasets', IEEE Transactions on geosciences and remote sensing, Vol.45, No.6, 1671-1680. 19. Yung-An Yang and Pai-Hui-Hsu, (2002) 'Application of remote sensing and GIS in landslide detection', International Journal of Remote sensing, 559-572. 20. Zaitchik, B.F, and Van Es, H.M., (2003) 'Applying a GIS slope-stability model to site-specific landslide prevention in Honduras', Journal of soil and water conservation, Vol.58, No.1, 45-53. Articles |
Buy Art Reproductions Auto Transport - Tips For Transporting Your Vehicle TI 85 Vibrations - Matrix Iteration Method Program Property market in Dubai: Will the prices keep climbing? Home of 2bucks an Ads Are You Ready For eBay Success? A Womans Funny Article My Site is Worth $7,738 |
||||||
 |
 |
||||||
Print this page