establishing precipitation thresholds for landslide …characterisation using gis-based modelling”...

Establishing Precipitation Thresholds for Landslide Initiation along with Slope Characterisation using

GIS- based Modelling

Surabhi Kuthari January, 2007

Establishing Precipitation Thresholds for Landslide Initiation along with Slope Characterisation using

GIS-based Modelling

by

Surabhi Kuthari

Thesis submitted to the International Institute for Geo-information Science and Earth Observation in partial fulfilment of the requirements for the degree of Master of Science in Geo-information Science

and Earth Observation, Specialisation: Geo-Hazards Thesis Assessment Board Supervisors Chairman: Prof. Victor Jetton (ITC) Dr. Cees J. van Westen (ITC)

Dr. Cees J. van Westen (ITC) Dr. P. K. Champati ray (IIRS) Prof. R. C. Lakhera (IIRS) Dr. P.K. Champati ray (IIRS)

External Expert: Prof. L. N. Sharma (PEC) INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE AND EARTH OBSERVATION

ENSCHEDE, THE NETHERLANDS AND

INDIAN INSTITUTE OF REMOTE SENSING (NRSA) DEHRADUN, INDIA.

I certify that although I may have conferred with others in preparing for this thesis, and drawn upon a range of sources cited in this work, the content of this thesis report is my original work. Signed……….

Disclaimer This document describes work undertaken as part of a programme of study at the International Institute for Geo-information Science and Earth Observation. All views and opinions expressed therein remain the sole responsibility of the author, and do not necessarily represent those of the institute.

Abstract Every year during intense rainfall periods, several incidences of landslides are reported from different parts of the National highway 58, which is the only motorable route connecting Badrinath, an important Hindu pilgrimage centre and other hill cities to the rest of the India. Slope failures cause traffic disruption leaving the pilgrims and inhabitants stranded for hours. Landslides and its secondary consequences such as landslide dams and subsequent flash floods often turn into major disasters. It is therefore, imperative to assess and predict the landslide hazard in the Himalayan terrain. However, the spatio-temporal prediction of rainfall triggered landslides is complicated by the multifarious interactions of the varied causes and triggers. Moreover, the accessibility of the terrain and availability of relevant data is usually problematic. The motive of the proposed research is to characterize slope failures in a complex terrain with data restraints for effective landslide hazard assessment. The present study aims to understand precipitation as a triggering mechanism and establish (spatial) precipitation thresholds for landslide initiation, along with the characterization of slopes using GIS- based models and validate the results. The proposed study is being carried out in the Alaknanda river catchment in Garhwal Himalayas, India. Geologically, the area is transected by Garhwal Lesser Himalaya and the Central Crystallines, which are separated along the Main Central Thrust. The catchment receives heavy precipitation between July and September. Landslides here are an outcome of the intrinsic geology, adverse natural topography, i.e., steep slopes in talus accumulation, weathered rocks and soils, and man-made modification of these fragile slopes. These inherently unstable slopes frequently fail during rainstorms, often with catastrophic consequences. Statistically, the critical rainfall amounts can be established, when a large data set on rainfall and landslide occurrence is available. Thus, an analytical approach is adopted to establish the relationship between rainfall magnitudes and slope failure initiation. The daily data recorded at the rain gauges in Joshimath, Badrinath (from 1987 to 2006) and Pipalkoti (from 2004 to 06) is analyzed and compared with the corresponding historical landslide records. A comprehensive landslide database since 1987 was organized. The present study identifies precipitation as the main mechanism for slope failure initiation in a part of Garhwal Himalayas and defines a lower-bound precipitation threshold, based on 72-hours precipitation and 15 days prior antecedent precipitation, if the daily rainfall is above a specified limit. The evaluation of the thresholds, based on the landslides observed in July and August, shows a high probability of landslide occurrence when the lower-bounds of the threshold are exceeded, under other considerations. The thresholds can be further improved by taking into account the spatial variability of other stability influencing parameters to aid in effective landslide hazard assessment in the Garhwal Himalayas. The prediction rates can be significantly improved with the availability of a well distributed network of weather stations and better recording of the initiation time of slope failures. In Garhwal Himalayas, where the geology is complex and varied, and accessibility of the terrain is problematic, the adopted modeling approach is a viable solution. The model takes full advantage of GIS technology to quantify topographic attributes, such as slope, aspect and drainage and based on these limited input parameters it attempts to define the stability of slopes. The modeling technique proves useful to identify and map the hill-slope stability. The slope stability index computations are primarily based upon the Digital Elevation Model. In this study Cartosat-1 data is used to derive a high resolution DEM of the study area. The GCPs were collected in the field using Differential GPS technique, processed in Leica’s SKIPRO software and later used to generate the DEM using Leica’s Photogrammetry Suite. Landslide inventory was prepared during reconnaissance survey followed by a detailed field work. Laboratory tests, field measurements and particle size analysis were carried out to determine the geotechnical and hydraulic properties. The parameter uncertainty is accounted for as uniform probability distributions between derived limits. The spatially distributed stability index maps represent the broad classes of instability well in the region which correlates well with the field

i

observation. The model has a success rate of 74.4% with a Digital Elevation Model of 15 m resolution, at identifying the slope failure initiation points.

ii

Acknowledgements The thesis on “Establishing precipitation thresholds for landslide initiation along with Slope Characterisation using GIS-based Modelling” is the part of my Geo-Hazards course, jointly organized by ITC, The Netherlands and Indian Institute of Remote Sensing (IIRS), India. First of all, I would like to acknowledge ITC and IIRS for giving me this opportunity to do this course. I would like to express my sincere appreciation and thanks to my supervisors Dr. Cees J. van Westen, Associate Professor, Department of Earth System Analysis, ITC, and Dr. P.K. Champati Ray, Scientist (SF), Geosciences Division, IIRS for their guidance, encouragement, comments, suggestions and constant support throughout my research work. I would like to express my sincere gratitude to Dr. V.K. Dadhwal, Dean IIRS for providing all the necessary support during the entire course at IIRS. My sincere thank goes to Dr. V. Hari Prasad, Programme Co-ordinator (Geo-Hazards) for his constant support and inspiration through out the course. I am also thankful to Mr. I. C. Das, Course Co-ordinator, for his immense support and cooperation during the entire project phase Thanks are also due to Prof. R.C. Lakhera, Head, Geosciences Division, IIRS, for allowing me to pursue this research work in his department. I am also grateful to Dr. Paul M. van Djik, Programme Director and Drs. Michiel Damen, Programme Coordinator, for their kind cooperation and providing all support at ITC. A special thanks to all faculty of Water Resource Division and Geosciences division of IIRS and faculty of Department of Earth System Analysis, ITC. I am grateful to Mr. Ashutosh Bhardwaj (WRD), Mr. Praveen Thakur (WRD), Dr. S.P. Agarwal (WRD) and Dr. R. S. Chatterjee (GSD) for their advice. I would like to acknowledge the support provided by the officials and staff members of Border Roads Organisation (B.R.O) and Central Water Commission (C.W.C.). I would like to express my sincere gratitude to Col. Jaswant Singh, Col. R.R. Patil and Major Nitin for providing me all necessary information. Words are not enough to express my gratitude to Mr. Prashant Kavishwar and Mr. Ajay Katuri, for their valuable suggestions in conceptualizing the methodology. Mr. Ashutosh Bhardwaj, Mr. Lesslie A., Ms Dipti for their kind guidance on the various aspects of DEM generation. I am grateful to Mr. Vivek Singh, for sharing with me his expertise during the fieldwork. Sincere thanks are due to the System Maintenance staff at IIRS and especially to Mr. Bhaskar and Mr. Avdesh, for their help. I express my thanks to all my friends at IIRS and ITC. My special thanks go to Ms. Anandita Sengupta, Ms. Chandrama Dey, Ms. Sreyasi Maiti, Ms. Nidhi and Mr. Prashob Raj for supporting me all the time. Last but not the least I express my honest and sincere gratitude to my family for their kind support, inspiration and encouragement for the completion of my thesis successfully. I would like to thank my Mom, Dad and Brother for always being there.

iii

Table of contents 1. INTRODUCTION............................................................................................................................1

1.1. Problem Statement ...................................................................................................................1 1.2. Aims and Objectives ................................................................................................................2

1.2.1. Specific Objectives.........................................................................................................2 1.3. Research Questions ..................................................................................................................2 1.4. Overview of the adopted methodology ....................................................................................2

2. LITERATURE REVIEW.................................................................................................................4 2.1. Slope failures............................................................................................................................4 2.2. Factors affecting slope stability................................................................................................4 2.3. Rainfall as a triggering agent....................................................................................................5 2.4. Concept of Precipitation thresholds .........................................................................................6 2.5. Modelling Slope Stability.........................................................................................................7

2.5.1. Slope Stability Models ...................................................................................................8 2.5.2. Hydrological models ......................................................................................................8 2.5.3. Remote sensing and GIS-based modelling of rainfall induced landslides .....................9

2.6. SINMAP...................................................................................................................................9 3. STUDY AREA...............................................................................................................................11

3.1. Himalayas...............................................................................................................................11 3.2. Alaknanda river catchment.....................................................................................................12

3.2.1. Geology ........................................................................................................................13 3.2.2. Geomorphology............................................................................................................13 3.2.3. Drainage .......................................................................................................................14 3.2.4. Soil ...............................................................................................................................14 3.2.5. Climate .........................................................................................................................15 3.2.6. Vegetation ....................................................................................................................15 3.2.7. Road network and Settlements .....................................................................................15

3.3. Landslides in the Study area...................................................................................................16 4. Precipitation Thresholds.................................................................................................................19

4.1. Rainfall data ...........................................................................................................................19 4.2. Landslide data.........................................................................................................................21 4.3. Database Organisation............................................................................................................22 4.4. Data analysis...........................................................................................................................24 4.5. Assessment of the threshold during Monsoon........................................................................34

4.5.1. Assessment of the thresholds for July and August, 2002 .............................................35 4.5.2. Assessment of the threshold for July and August, 2003...............................................36 4.5.3. Assessment of the thresholds for July and August, 2004 .............................................36 4.5.4. Assessment of the thresholds for July and August, 2005 .............................................37

5. Generation of Digital Elevation Model (DEM) for Modelling ......................................................39 5.1. GCP collection .......................................................................................................................39 5.2. Post-processing of DGPS data ...............................................................................................40

5.2.1. DEM generation ...........................................................................................................41 6. Application of a GIS-based Physical model for Slope Stability Assessment.................................46

iv

6.1. Digital Elevation Model (DEM).............................................................................................49 6.2. Model parameters ...................................................................................................................49

6.2.1. Ratio of Transmissivity to Effective recharge (T/R ratio)............................................49 6.2.2. Soil strength parameters ...............................................................................................50

6.3. Landslide point map ...............................................................................................................50 6.4. Processing of DEM ................................................................................................................51 6.5. Model Run..............................................................................................................................52 6.6. Slope Stability Index ..............................................................................................................55

7. Results and Discussion...................................................................................................................57 7.1. Results of the analysis ............................................................................................................57

7.1.1. Rainfall as a prime triggering mechanism....................................................................57 7.1.2. Precipitation thresholds for landslide initiation ...........................................................57 7.1.3. Digital Elevation Model ...............................................................................................57 7.1.4. Slope characterisation ..................................................................................................58

7.2. Related Issues .........................................................................................................................58 7.2.1. Precipitation Thresholds...............................................................................................58 7.2.2. Generation of DEM......................................................................................................59 7.2.3. GIS-based modelling....................................................................................................59

7.3. Conclusion..............................................................................................................................60 8. REFERENCES...............................................................................................................................61

v

List of figures Figure: 1-1. Flow chart of the Methodology adopted ..............................................................................3 Figure: 2-1. Changes in the factor of safety with time (Popescu, 2005) ..................................................5 Figure: 2-2. Different hydrological triggers leading to slope instability (Beek, 2002)............................6 Figure: 3-1. Alaknanda river catchment and its location .......................................................................12 Figure: 3-2. Geological map of the Study area (Virdi, 1986) ................................................................13 Figure: 3-3. Drainage map showing the main course of the river and the tributaries ............................14 Figure: 3-4. Average monthly distribution of rainfall between 1976 and 2005.....................................15 Figure: 3-5. Road map showing the NH-58 ...........................................................................................16 Figure: 3-6. Landslides in the Study Area.............................................................................................17 Figure: 3-7. Landslides in the Study Area (2) ........................................................................................17 Figure: 4-1. Image showing the geographic location and elevation of the three rain gauges in the study

area ....................................................................................................................................19 Figure: 4-2. Distribution of average, minimum, and maximum of monthly rainfall at Joshimath rain

gauge between 1987 and 2005 ..........................................................................................20 Figure: 4-3. The distribution of rainfall in Joshimath over the years between 1987 and 2005..............20Figure: 4-4. Scatter plot to show the relationship between the rainfalls recorded at Pipalkoti and

Joshimath in August 2004 and August 2005.....................................................................21 Figure: 4-5. The annual landslide distribution between 1987 and 2006 ................................................22 Figure: 4-6. Graph showing the distribution of single and multiple landslide events and days.............23Figure: 4-7. The landslide distribution coincides with the average monthly rainfall at Joshimath rain

gauge observed between 1987 and 2005...........................................................................25 Figure: 4-8. Distribution of landslides and monthly rainfall over the years. .........................................28 Figure: 4-9. The scatter plot based on the daily and prior 3-day rainfall on the landslide days.............29Figure: 4-10. The scatter plot based on the daily and prior 15-day rainfall on the landslide days........30Figure: 4-11. The scatter plot based on the daily and prior 30-day rainfall on the landslide days.........30Figure: 4-12. Scatter plot based on the 3-day total and 15-day prior antecedent rainfall. .....................32 Figure: 4-13. Scatter plot based on the 3-day total and 15-day prior antecedent rainfall. ....................33 Figure: 4-14. Minimum probable threshold for man-modified slopes shown by the green line visually

identified on the scatter plot..............................................................................................34 Figure: 4-15. Daily distribution of landslide incidences and rainfall in July and August 2002.............35Figure: 4-16. Graph showing the daily variation in the thresholds and landslide events for July and

August, 2002. ....................................................................................................................35 Figure: 4-17. Daily distribution of landslide incidences and rainfall in July and August 2003............36Figure: 4-18. Graph showing the daily variation in the threshold and landslide events for July and

August, 2003. ....................................................................................................................36 Figure: 4-19. Daily distribution of landslide incidences and rainfall in July and August 2004............37Figure: 4-20. Graph showing the daily variation in the threshold and landslide events for July and

August, 2004. ....................................................................................................................37 Figure: 4-21. Daily distribution of landslide incidences and rainfall in July and August 2004.............38Figure: 4-22. Graph showing the daily variation in the threshold and landslide events for July and

August, 2005. ....................................................................................................................38 Figure: 5-1. Cartosat-1 image showing locations of probable GCPs. ....................................................40 Figure: 5-2. The post-processing of Ground Control Points using Leica’s Ski-Pro software................41

vi

Figure: 5-3. Generation of DEM and Ortho image in LPS ....................................................................42 Figure: 5-4. The raster DEMs. (5.4.1) 10m resolution DEM, (5.4.2)15m resolution DEM, 5.4.3) 20m

resolution DEM.................................................................................................................43 Figure: 5-5. Ortho-rectified Image generated using the raster DEMs...................................................45 Figure: 6-1. Infinite slope stability model ..............................................................................................46 Figure: 6-2. The concept of specific catchment area as applied in SINMAP (Pack et al., 1998b) .......47Figure: 6-3. Landslide initiation points on Cartosat-1 Ortho-image. The National highway is also

shown. ...............................................................................................................................51 Figure: 6-4. The D∞ method for multiple flow directions followed from (Pack et al., 1998b) .............51Figure: 6-5. Slope Stability index map from 15m DEM (deterministic)................................................52 Figure: 6-6. Slope Area plot (15 m DEM_deterministic).......................................................................53 Figure: 6-7. Slope Stability index map from 20m DEM (deterministic)................................................53 Figure: 6-8. Slope Area plot (20m DEM_deterministic)........................................................................53 Figure: 6-9. Slope Stability index from 15 m DEM (probabilistic) .......................................................54 Figure: 6-10. Slope Area plot (15 m DEM_Probabilistic) .....................................................................54 Figure 6-11 Slope Stability index from 20 m DEM (probabilistic) .......................................................55 Figure 6-12: Slope Area plot (20 m DEM_Probabilistic) ......................................................................55

vii

List of tables Table: 3-1. The annual damage due to Heavy rains, landslide and flash floods in Himalayas (2002) .12 Table: 4-1. Statistics showing the classification of landslide events and number of slides in each

category ..............................................................................................................................24 Table: 4-2. The landslide incidences in 1990 and associated daily rainfall. .........................................28 Table: 5-1. Summary of the triangulation results...................................................................................42 Table: 6-1. The stability class definitions .................................................................................................49 Table: 6-2. Geotechnical properties .......................................................................................................50

viii

Establishing Precipitation Thresholds for Landslide Initiation along with Slope Characterisation using GIS-based Modelling

__________________________________________________________________________________

1. INTRODUCTION

1.1. Problem Statement

Every year during intense rainfall periods, several incidences of landslides and related casualties are reported from different parts of the Rishikesh-Badrinath National highway 58, in the state of Uttaranchal, India. This road is the only motorable route connecting Badrinath, an important Hindu pilgrimage centre and other hill cities to the rest of the nation. Slope failures cause traffic disruption leaving the pilgrims, tourists and inhabitants stranded for hours. Breaching of the highway has deeper implications as many towns, villages and hamlets are often marooned from the rest of the country. The situation worsens for the sufferers with the onset of cold waves, which are usually experienced all through the year in the Himalayas. Landslides and their secondary hazards such as landslide dams and subsequent flash floods often turn into major disasters as was witnessed in 1970 when a 77 year-old landslide dam in the Birahiganga River, a tributary of Alaknanda, breached and caused major destruction in the downstream areas of Alaknanda. In July-August 2004 a heavy downpour in Garhwal Himalaya caused debris slides and debris avalanches where at least 25 people died and 5000 people were stranded for days without food on the Joshimath-Badrinath Road. The worst events happened in the monsoon period of 1998, during which around 400 people were killed by large landslides near Okhimath and Malpa, in the Alaknanda Catchment. Given these disastrous effects, it is therefore imperative to assess and predict the landslide hazard in the Himalayan terrain. Landslide hazard is defined as “the probability for a landslide within a given area and within a given period of time” (Varnes, 1984). Thus, ideally the identification of landslide hazard must include a spatial and a temporal component, i.e., the probability for the occurrence of landslides at a certain location in space and its occurrence in time (Terlien, 1998b) . However, the spatio-temporal prediction of rainfall triggered landslides is complicated by the multifarious interactions of the varied internal causes and triggering factors. Slope instability is not only controlled by the geology, geomorphology and hydrology but also governed by complex tectonics, geodynamics and meteo-climatic factors. Increased urbanization, accompanied by expansion of roads also creates an increasing pressure on the landscape, and leads to higher degrees of vulnerability, as well as to increased landslide occurrence due to improper road and building construction. The destruction of forests and the vegetative cover that binds the top soil has been going on at an ever-increasing pace and the conversion of forest land into agricultural and horticultural holdings also adds to the increasing landslide susceptibility of the terrain. Site investigations and mitigative measures are expensive and are generally adopted in the post-disaster phase. Moreover, the accessibility of the terrain, poor distribution of weather stations and lack of high elevation rain gauges impedes the collection of data that can contribute towards understanding the mechanism of slope failure due to rainfall triggering. Furthermore, accurate dates of landslides are seldom available due to sparse population of the region and lack of media and official reporting of such events. Thus, the motive of the proposed research is to contribute to slope stability zonation in a complex terrain with data constraints for effective landslide hazard management.

1


__________________________________________________________________________________ 1.2. Aims and Objectives

The present study aims to understand precipitation as a triggering mechanism and to establish (spatial) precipitation thresholds for landslide initiation, along with the characterization of slopes using GIS- based modelling and validate the results in a part of the Garhwal Himalayas in India.

1.2.1. Specific Objectives

Based on the main objective, the following specific objectives have been defined: To understand the preparatory and triggering factors for land sliding in the study area. To generate a relevant landslide database for rainfall threshold analysis. To understand the influence of rainfall on shallow landslide initiation and establish

precipitation thresholds, by correlating rainfall data with landslide dates. To generate a DEM from CARTOSAT Stereo data for geomorphic interpretation, and as main

input for the subsequent modelling. To assess the effectiveness of a GIS-based model for slope characterization in a complex

Himalayan terrain. To calibrate the model and validate the results.

1.3. Research Questions

The following research questions have been formulated: Is there enough historical information on landslide occurrence available to generate a landslide

database that can be used in rainfall threshold analysis? Is it possible to define precipitation thresholds for landslide occurrence in such a complex

terrain? How far can GIS-based models be effectively applied in a data scarce and heterogeneous

region like the Himalayas? Is a static slope stability model for shallow landslides such as SINMAP relevant for providing

indications on landslide hazard zones in the Himalayas? Can CARTOSAT-1 meet the requirements of a high resolution DEM for Himalayan terrain as

input in the slope stability modelling? How to parameterize the SINMAP model in a data scarce and heterogeneous region such as

the Himalayas? How far the model results validate the actual scenarios on ground?

1.4. Overview of the adopted methodology

To define the spatial precipitation thresholds for slope failure instigation and applying them in GIS to model the susceptible slopes in the Alaknanda catchment, India the following research methodology was followed:

2


__________________________________________________________________________________

Figure: 1-1. Flow chart of the Methodology adopted

3


__________________________________________________________________________________

2. LITERATURE REVIEW

2.1. Slope failures

Slope failures are the gravity-driven, down slope response of earth materials to the disequilibrium on a slope under the influence of factors related to geomorphology, topography, geology, hydrology, and land use patterns. Slope failures are a form of “gravitative transfer producing immediate and perceptible modification of the earth’s surface” (Thornbury, 1954). These are described as the down-slope movements of soil or rock masses as a result of shear failure at the boundaries of the moving mass. Amongst the earliest endeavour at classifying various mass-wasting processes mention can be made of Sharpe, who in 1938 recognized slow flowage, rapid flowage, landslides and subsidence. Varnes emphasized that classifications of mass movements must include the type of movement and type of material (Varnes, 1978); movements are thus classified as falls, flows, slides, spreads and topples and the types of materials as rock, debris and earth. Sliding usually occurs in coherent masses along distinct shear surfaces. Falls are sudden slope movements. While the other slope movements occur in relatively dry conditions Flows occur when shallow soils and talus material become saturated during heavy precipitation and move down slope into and along pre-existing drainage ways. The two main sub-types of landslides recognized are translational slides in which the displacement is along a planar or undulating surface of rupture, and rotational slides in which movement occurs along a curved (concave) surface of rupture. However, there are instances where it is difficult to ascertain the type of movement which caused ground displacement. The complexity related to the phenomenon of landslides was answered (Cruden and Varnes, 1996) by combining the sort of movement and the type of materials and suggested that if the type of movement changes as the failure advances, then “the material should be described at the beginning of each successive movement”.

2.2. Factors affecting slope stability

The real force governing the dynamic landscape processes is the constant pull of gravity which makes all hill slopes susceptible to failure. Upon failure, the earth material moves down slope until slope stability is re-established. Besides, gravity the geology, geomorphology, hydrology and anthropogenic factors contribute largely towards destabilization of slopes. “Slope instability processes are the product of local geomorphic, hydrologic, and geologic conditions; the modification of these conditions by geo-dynamic processes, vegetation, land use practices, and human activities; and the frequency and intensity of precipitation and seismicity”(Soeters and Westen, 1996) In 1985, Varnes observed that land sliding initiates as a consequence of various processes, both natural and man-made and their interaction and thus, for effective analysis of slope instability the role of each parameter must be carefully ascertained. It is recognized that even though the terms ‘cause’ and ‘trigger’ of a landslide are often used synonymously they have different implications(Popescu, 2005; Terlien, 1998b). The landslide causes (Fig 2.1) are classified as preparatory causal factors, and triggering causal factor.

4


__________________________________________________________________________________

Figure: 2-1. Changes in the factor of safety with time (Popescu, 2005) The cause can be a short- or long-term process and vary from natural, intrinsic factors to man-induced and, but trigger is the ultimate external stimulus of a landslide, immediately preceding a slope failure such as earthquake or high intensity rain storm. Triggering factors lead to a sudden instability in a slope and usually precede the failure. These may cause a “local short- duration drop in the stability” of a slope due to an extreme rainfall event or a seismic tremor. Even man-modifications of natural slopes such as road excavations or land use, land cover changes can make a slope liable to fail or even trigger a slide immediately.

2.3. Rainfall as a triggering agent

Assessment of hillslope hydrology shows that rainfall contributes to the instability of slopes by means of infiltration or resulting decrease of soil suction by a moving wetting front or increase of pore water pressure by a rising water table ((Crozier, 1986) Dhakal & Sidle, 2004; (Hengxing et al., 2003)

5


__________________________________________________________________________________

Figure: 2-2. Different hydrological triggers leading to slope instability (Beek, 2002) The diverse influences exerted by the rain have been extensively investigated by many researchers working in the field of slope stability and landslide control (Asch et al., 1999; Borga et al., 2002; Finlay et al., 1997; Lin and Jeng, 2000). It is noted that for deep seated landslides, the swelling of groundwater table or perched water table along with rock and soil softening by rainfall infiltration influences the stability, whereas, shallow landslides are dominated by transient pore pressure in response to rainfall process, combined with water washing or soil erosion (Hengxing et al., 2003; Terlien, 1996). It is generally accepted that the rainfall triggering of landslides is controlled by hydro-geology of the slope-forming materials(Finlay et al., 1997; Tsaparas et al., 2002). (Crozier, 1986) studied the antecedent precipitation and sediment moisture as an important triggering factor. Glade verified that though rain-storm characteristics appear to dominate, antecedent conditions have a major influence on the initiation of landslides in certain regions(Glade, 1997). Sanderson et al. (1996) investigated the weather conditions (rainfall, air-temperature, wind-speed and air-humidity) prior to debris flows, rock falls and rockslides to study the influence of meteorological factors on the initiation of debris flows, rock falls, rockslides and rock mass stability. Different approaches have been presented in the literature (Crozier, 1986; Wilson and Wieczorek, 1995) to explain the relationship between rainfall and slope failures: rainfall thresholds, hydrological models and coupled methods. These investigations suggest that the rainfall trends should be observed in terms of duration and intensity in order to single out the minimum rainfall distribution configurations which would definitely aim at the possibility of fore-casting rainfall induced landslides.

2.4. Concept of Precipitation thresholds

A review of available literature emphasizes the role of precipitation as an important control on the initiation of slope failures and provides evidence that shallow translational slides and flows are often triggered by meteo-climatic events that are in excess of some threshold. Several empirical relationships have been deduced, based on the assumption that there exists a direct relationship between the occurrence of landslides and the quantity of rainfall, in terms of rainfall intensity and duration of rainstorm events, or short-term rainfall e.g. 24-h rainfall, and antecedent rainfall. The concept of “pluviometric threshold” was introduced in (Campbell, 1975) and (Starker, 1979) introduced the “duration-intensity relationship”. However, a significant breakthrough was achieved when Caine, in 1980 collected a global data set of rainfall and landslide occurrences on a variety of

6


__________________________________________________________________________________ natural slopes and empirically derived a lower-bound threshold of rainfall intensities and duration for the initiation of shallow landslides and flows. He defined the limit as I = 114.82D-0.39 (Caine, 1980) and standardized it for rainfall durations between 10minutes and 10 days. However, the threshold-curve was too low, as it was for varying areas of heterogeneous geology, geomorphology and climate. Despite this limitation, thresholds based on the intensity-duration method outlined by Caine, have been estimated by different authors(Aleotti, 2004; Corominas and Moya, 1999; Crosta, 1998; Crosta and Frattini, 2000). To regionalise the threshold, taking into account the local climate regime, the “intensity-duration” approach can be further refined by normalizing the intensity value by the mean annual precipitation (MAP). This is done by plotting the hourly rainfall intensity against cumulative rainfall measurements expressed as a percentage of MAP. An instability field is identified, which isolates the landslide events according to the season of occurrence. To obtain a possible correlation between rainfall parameters and the occurrence of soil slip phenomena and to identify the local rainfall threshold for triggering shallow landslides in terms of mean intensity, duration and mean annual precipitation (MAP) in the south Apuan Alps, Italy (Giannecchini, 2006) analysed the pluviometric data between 1975-2002 from a single rain gauge. (Chleborad, 2003; Chleborad et al., 2006) estimated a cumulative rainfall threshold (CT), P3=3.5–0.67P15, defined by rainfall amounts (in inches) during the last 3 days, P3, and the previous 15 days, P15, for slope failure initiation after analyzing historical precipitation data (1933-1997) associated with 187 wet season landslides. The concept of antecedent rainfall percentage exceedence time (ARPET) was presented by (Chowdhury and Flentje, 2002). While empirical thresholds are defined on the basis of landslide occurrences in relation with rainfall amounts, physical thresholds are based on numerical models that consider the relation between rainfall, pore-pressure and slope stability. Terlien suggested that in absence or lack of historical data, the hill-slope hydrological processes cannot be determined statistically and physically based hydrological and geo-technical models are applied to find critical rainfall amounts(Terlien, 1998b). Crosta studied a series of soil slip triggering meteorological events in northern Italy, using soil properties, triggering rainfall, and local lithological and morphometrical settings of different sites as input to an infiltration model to define site-specific thresholds in the Piedmont, Pre-Alpine and Alpine regions(Crosta, 1998). (Gabet et al., 2004) observed 3 years of daily sediment load and daily rainfall data to define relationship between monsoonal rainfall and landslides in the Annapurna region of Nepal and comprehend that two distinct thresholds; seasonal accumulation and daily total, are critical for slopes to fail. They developed a slope stability model, driven by daily rainfall data to explore the geomorphic controls (slope) on these thresholds. The presence of spatially well-distributed weather stations at reasonable distance from the sites of instability, aids significantly in establishing reliable thresholds. In Alaknanda catchment, which is relatively ungauged, limited studies have been carried out. In the catchment, an analysis of the extreme rainfall events and associated natural hazards (Joshi and Kumar, 2006) has shown that extreme rainfall events of late monsoon season are more damaging than those of early monsoon possibly due to excessive runoff from saturated soils.

2.5. Modelling Slope Stability

Numerous methods have been developed to assess the stability of the hill-slopes. Broadly these methods have been classified as direct and indirect; direct method is landslide inventory based and consists of geomorphological mapping; indirect methods include heuristic i.e., knowledge driven, statistical and deterministic approaches (Soeters and Westen, 1996; Westen et al., 1997). The Heuristic or expert-driven approach (Barredo et al., 2000) is qualititative and in this landslide influencing factors are directly related to the occurrence of failure. Indirect landslide susceptibility mapping combines the potential terrain factors influencing slope stability with a landslide inventory

7


__________________________________________________________________________________ map. The statistical analysis; bivariate and multi-variate(Begueria and Lorente, 1999), determines, for an area, the combination of factors leading to slope failure and gives quantitative predictions for areas with similar set up. Deterministic approaches are based on modelling the stability of slopes and generally provide the absolute values of landslide hazard in terms of safety factors, or the probability of failure under given conditions. However, these are generally data intensive, requiring accurate and detailed inputs for parameters.

2.5.1. Slope Stability Models A slope is said to be stable, if it meets a prescribed need for a fixed period of time with a suitable or acceptable safety factor. The factor of safety is the ratio of available strength to mobilized strength and is calculated from analyzing the stability of slopes (Mostyn and Small, 1987). Quantitative analyses of slope stability are based on limit equilibrium concepts first developed in Sweden in the early 1900s (Greenway, 1987). Limit equilibrium methods are still popular for their simplicity and non-limit equilibrium methods are thus, rarely employed. The infinite slope model is the most common method and is applied for plane slip surfaces. The method of slices is mainly used for circular failure surfaces and upon utilizing the concept of effective stress is successfully applied for other slip surfaces too. The above limiting equilibrium methods adopt the Mohr-Coulomb failure criterion.

2.5.2. Hydrological models

Physically based hydrological models are used to simulate the flow of groundwater under saturated and unsaturated conditions in natural slopes. Recently, the hydrological models are used in integration with slope stability models to facilitate precise simulation of the stability scenarios. These are classified as: Steady-State Models Theoretical models integrated in GIS have been developed to predict the changes in slope stability with topography, geology, and hydrology and landuse changes as well. These models employ the effective stress principle given by Tergazhi to simulate the hillslope hydrological processes in an infinite-slope stability analysis, which relates landslide potential to groundwater pressures in discrete landscape cells. The steady-state hydrological models assume that rainfall influences groundwater only by modulating steady or quasi-steady water table heights and that groundwater flows exclusively parallel to the slope (Iverson, 2000). They neglect the “slope-normal re-distribution” of pore water pressures associated with transient flow of rain water. However, they are widely used as they can be easily implemented for landslide hazard analysis as their data requirements can be easily met. Transient Response Models It is evident from theory and measurements that pore water pressures in hillslopes are strongly influenced by transient rainfall infiltration and that “pressure redistribution includes a large component normal to the slope” (Iverson, 2000). Transient pore pressures that develop in hillslopes in response to rainfall reflect the influence of topography, geology, and climate on slope failure potential and are thus, significant factors (Hengxing et al., 2003) Iverson used rational approximations of Richard’s equation in combination with a generalized infinite-slope stability model and Newton’s second law to assess the effects of transient rainfall on timing, rates, and locations of landslides (Iverson, 2000). The rainfall triggered shallow landslides have been modelled on the basis of Iverson’s model by a number of researchers(Sharma and Nakagawa, 2005; Tsai and Yang, 2006).

8


__________________________________________________________________________________ 2.5.3. Remote sensing and GIS-based modelling of rainfall induced landslides

Landslides are “natural multi-dimensional, dynamic geomorphological systems” (Brunsden, 1999) and although a complete simulation of the mechanisms and processes is not yet achieved, recent advances in the acquisition of remote sensing data with stereoscopic vision, enhanced spatial and temporal resolution; image processing techniques and Geographic information system, have enabled us to better comprehend the hazard, its causes and triggers and model the slopes with higher accuracy of prediction. Remote sensing can be applied to the monitoring of landslides over difficult to access terrains. The development of new procedures such as interferometry, permanent scatters technique(Ferreti, 2005) and their assimilation in GIS have increased the potential of remote sensing manifolds. GIS application in slope stability analysis is not just limited to the spatial and non-spatial database generation and manipulation but it also assists in modelling, as different physically based models can be readily integrated in it. A coupled approach incorporated in GIS, combining hydrological and slope stability parameters, is commonly adopted to facilitate the simulation of slope stability. Deterministic models which were traditionally used to calculate the stability of individual slopes, with the advent of GIS and its data handling capabilities, have also been successfully applied over larger areas such as catchments (Montgomery and Dietrich, 1994; Pack et al., 1998a)and road corridors (Hammond et al., 1992b). While most of these models are integrated within a GIS environment, some utilise GIS technique during pre- and post-processing and visualisation of results. Innumerous models with built-in GIS such as, CHASM(Wilkinson et al., 2002), CHILD(Tucker et al., 1999), LISA(Hammond et al., 1992a), SHALSTAB(Montgomery and Dietrich, 1994), Geo-Slope(Geo-Slope, 1994), TRIGRS(Baum et al., 2002), SMORPH, Soil-Screen, dSLAM and so on, have been developed and effectively applied to landslide studies. CHASM is a hydrological-slope stability model developed by Bristol University, links a finite difference hydrological model with Bishop’s rotational slope stability model. A PC-based integrated dynamic slope stability model, IDSSM (Dhakal and Sidle, 2004), a modified version of dSLAM, examines the influence of different rain storm characteristics on landslide initiation. Geo-SLOPE is a software package based upon limit equilibrium theories and can give solutions for different slopes i.e. infinite, slices, etc. It takes in inputs for soil properties and the output gives the factor of safety for all surfaces along with the minimum safety factor foe potential slip surface. TRIGRS, acronym for Transient rainfall infiltration and grid based regional slope stability, is a FORTRAN program for computing transient response of pore-pressure due to rainfall infiltration and consequent changes in the safety factor. It is based on the Iverson’s method and is modified by including the solutions for additional basal boundary conditions and a runoff-routing scheme (Baum et al, 2002). Even though the model has been successfully implemented and was much preferred, it could not be applied to the study area owing to its data requirements. SHALSTAB model developed by Montgomery and Dietrich provides an estimate of the spatial distribution of the critical rainfall amounts, which is defined as the minimum steady state rainfall predicted to cause instability.

2.6. SINMAP

The Stability Index Mapping (SINMAP) is a coupled infinite slope stability model with a steady-state hydrology model. SINMAP takes full advantage of the fact that debris flow source areas are, in general, strongly controlled by surface topography through shallow subsurface flow convergence, increased soil saturation, increased pore water pressures, and reduction in soil/regolith shear strength. Thus, it is appropriate for shallow translational sliding controlled by shallow Groundwater flow convergence (Acharya, 2003; Pack et al., 1998a; Zaitchik and Es, 2003; Zaitchik et al., 2003). The model has been successfully applied by (Otteman, 2001) to model debris flow susceptibility for the bent creek experimental forest near Asheville, North Carolina with 83% success rate at determining areas critical for a 125-250 mm/day rainfall.

9


__________________________________________________________________________________ In Nepal Himalayas, where field data deficient conditions are comparable with those in India, (Acharya, 2003) analyzed the slope stability deterministically and compared their model results with SINMAP and found a correlation of 0.9862 between the two. The model has been applied to an area affected by Hurricane Mitch(Zaitchik et al., 2003); the model assigned a probability of failure of < 0.975 to 52% of identified landslide systems and a probability of failure <0.5 to 75% of identified systems. They generated a distributed probability map (DPM) for slope failure and used it to identify the spatial scale of variability captured by the model. Ripley’s K-function for distribution of a spatial point process was applied to mapped landslides and to simulated data based on the DPM. Lan has used SINMAP, modified by adding the hydrodynamic pressure for simulating the groundwater seepage force, to prepare the landslide susceptibility maps for different rainfall conditions (Lan et al., 2004).

10


__________________________________________________________________________________

3. STUDY AREA

3.1. Himalayas

Himalayas (Sanskrit: “abode of snow”), the youngest orogenic mountains in the world, developed in a series of orogenies, succeeding the Eurasian-Indian plate collision and are still rising. The mountain system extends longitudinally for about 2600 km with a syntaxial arcuate bend, along the north of the Indian subcontinent. The mountains rise from the Gangetic plain up to the Qing Zang Gaoyuan plateau in Tibet, averaging 250 to 400 km of width. Laterally, the region is classed as Gangetic plain, Outer Himalayas (Siwaliks), Lesser Himalayas, Great or Higher Himalayas, and Tethyan Himalayas (Indo-Tsangpo suture zone) on the basis of their distinct and characteristic geology and topography. The Himalayas play a significant role in influencing the climate of the Indian subcontinent and exerts a major control on monsoon and rainfall patterns. Within the Himalayas climate varies depending on elevation and location. Climate ranges from subtropical in the southern foothills, with average summer temperatures of about 30° C and average winter temperatures of about 18° C; warm temperate in the Middle Himalayan valleys, with average summer temperatures of about 25° C and cooler winters; cool temperate conditions in the higher parts of the Middle Himalayas, where average summer temperatures are 15 to 18° C and winters are below freezing; to a cold alpine climate at higher elevations in the Great Himalayas, where summers are cool and winters are severe. At elevations above 4880 m, the climate is very cold with below freezing temperatures and the area is permanently glaciated. The natural vegetation is influenced by climate and elevation. The Sub-Himalayas are characterised by tropical, moist deciduous forest, although most of these have been cut for commercial timber and agricultural purposes. Species of pine, oak, rhododendron, poplar, walnut, and larch are commonly found in the Middle Himalayas at elevations between 1520 and 3660 m. Alpine vegetation occupies higher parts of the Higher Himalayas just below the snow line and includes shrubs, rhododendrons, mosses, lichens, and wildflowers such as blue poppies and edelweiss while, forests of spruce, fir, cypress, juniper, and birch are found below the timber line. The physiographic and climatic controls are observed to influence population and habitat as well. The adverse conditions tend to restrict population; however with the roads extending towards remote areas, conditions are drastically changing. Nearly forty million people are permanent dwellers in this mountain belt. The Himalayan region is characterized by an overall low economic growth rate combined with high population growth, which contributes to stagnation in the already low level of per capita gross national product. While agriculture and cattle is the prime source of income, modern industries are lacking. The hydroelectric potential of the Himalayan Rivers has been recognised, which is evident by the no. of upcoming major as well as small scale dams and barrages in the area. Fruit processing is another important industry. Tourism has also emerged as a significant growth industry in the region and according to statistics, over a million people flock each year for pilgrimage, adventure sports, and wildlife.

11


__________________________________________________________________________________ Table: 3-1. The annual damage due to Heavy rains, landslide and flash floods in Himalayas (2002)

S.no

Year Districts affected

Villages affected

Population affected

Crop area affected

Houses damaged

Human loss

Cattle loss

Estimated Loss (cr)

1 1999 202 33,158 326.12 8.45 584,823 1375 3,661 -

2 2000 200 29,964 416.24 34.79 2,736,355

3048 102,121

1020.97

3 2001 122 32,363 210.71 18.72 345,878 834 21,208 861.62

The entire Himalayan region is susceptible to various natural hazards such as, earthquake, landslides, and flash floods. The statistics for the year 1999-2001 (Table 3-1) reveal the extent of damage caused in Himalayan states due to severe rainstorms and consequent landslides and flash floods. In the Seismic zonation map of India, most of the Himalayan region is in Zone V, i.e. highly vulnerable and only a few sectors in Zone IV. Drastic changes in the land use and cover patterns along with population growth has further threatened the fragile environment of the Himalayas.

3.2. Alaknanda river catchment

The Alaknanda river catchment between Badrinath (30°44’31N and 79°29’39E) and Pipalkoti (30°26’05N and 79°25’41E) is studied to establish the rainfall threshold and GIS-based modelling in a portion of the upper catchment of the river to characterise the valley slopes. The modelling basin is bounded by the latitudes 30°26’47”N and 30°35’33”N and longitudes 79°25’30”E and 79°36’58”E and lies in Chamoli District, Uttaranchal. The area falls on the Survey of India topo-sheet no. 53N/7 and 53N/10. The study area is characterised by deep gorges and rugged mountains. The maximum elevation observed in the area is 5443m and the minimum elevation is 1077m, with respect to the mean sea level. The choice of study area is not based upon easy availability of data; rather it was realised that inspite of the damage caused in the area due to landslides, not much has been investigated in this portion of the catchment owing to complexity of terrain and data constraints.

Figure: 3-1. Alaknanda river catchment and its location

12


__________________________________________________________________________________ 3.2.1. Geology

The study area is transected by the granites and gneisses of Central Crystallines(Heim and Gansser, 1939), Quartzites and phyllites of Saknidhar Formation, and the ferruginous shales, quartzites and phyllites of Garhwal group. The calc-zone of Garhwal group is observed as we traverse from Pipalkoti to Helang. Along the Munsiari Thrust, which passes through Helang, the Munsiari Formation overrides the Lesser Himalayan sediments. The Munsiari Formation comprises garnet-bearing mica schists, deformed amphibolites, and calc-silicate lenses, quartzites, mylonitic-biotite rich gneisses and augen gneiss(Gururajan and Choudhuri, 1999). The Main Central Thrust, which was initially considered as a thrust contact separating the Lesser Himalayan sediments from the Central crystalline mass (Heim and Gansser, 1939), is now located at the boundary where the high grade metamorphics, over-thrust the low to medium grade metamorphics of Munsiari Formation. The contact can be located near Joshimath town and is moderately inclined to the north(Gairola, 1975). In Joshimath, the garnetiferous gneisses with augen texture and biotite rich schists are found. Moving towards Pandukeshwar and further north, the massive, coarse grained, sugary textured quartzites with pink garnets and tourmalines become more prominent. Near GovindGhat and Pandukeshwar, the river becomes wide and thick terraces are seen on both banks. Highly metamorphosed granitic gneisses dipping 60°-70°N are dominant rock types beyond north of Hanuman chatti. The major thrusts and faults trend in east west direction, conforming to the longitudinal extension of Himalayas.

Figure: 3-2. Geological map of the Study area (Virdi, 1986)

3.2.2. Geomorphology

The structural and lithological control on the development of landforms is demonstrated by the fact that metamorphic rocks which are dominant in the area are more resistant to erosion but weathering is highly active where joints have developed. Rugged topography is characterised with moderately

13


__________________________________________________________________________________ dissected hills. The major landforms observed are of structural, glacial, fluvial, and denudational type. The sharp crested peaks, arêtes and broad U-shaped valleys, characteristic of glacial origin are the most prominent landforms in the upper reaches. Relicts of past glaciation are seen in a fossil valley near Pandukeshwar and groove marks noted in the gneissic boulders around Joshimath town. Partly to well-developed terraces are the major fluvial landforms observed. Some terraces which are found on higher elevations, about 50-100m above the present river level, are indicative that the river Alaknanda has cut deep valleys. The mountain building activities, in the geological past, deposited talus and fluvio-glacial material at the angle of repose and in such places weathering is high. The main geomorphic units identified in the southern area include low dissected hills, moderately dissected hills, talus and scree deposits, and river terraces.

3.2.3. Drainage

The Alaknanda River, a major tributary of the river Ganges, originates from the Satopanth glacier where it is joined by river Saraswati. It is called Alaknanda after the confluence with Rishi Ganga. Its important tributaries are the Rishi Ganga, Khir Ganga, Ghrit Ganga, Nilganga, Laxman Ganga, Dhauli Ganga, Garur Ganga, and Patal Ganga. The river follows a straight course for the most of its length, flowing in North-South direction, except between Vishnuprayag and Joshimath where it moves westward. Between Joshimath and Pipalkoti the river flows in SW direction. Deep gorges have been formed due to the incision of stream. The sub-parallel to parallel drainage pattern is suggestive of a structural control.

Figure: 3-3. Drainage map showing the main course of the river and the tributaries

3.2.4. Soil

The nature and characteristics of soil varies at different locations in the study area. The soil cover is generally thin except in cultivated areas on moderate slopes that have a thicker soil cover consisting of relatively fine soil. It has been observed that the precipitous slopes are generally without soil cover. The soil found on these cliffs exists in cracks, joints and along the foliation planes. The valley slopes

14


__________________________________________________________________________________ mainly consists of the old debris material. Fine soil is found to be abundant on moderate slopes, while coarse soil in abundant on steep slopes.

3.2.5. Climate

The snow line on the southern slopes of the Great Himalayas varies from 4480 m in the eastern and central Himalayas to 5180 m in the western Himalayas. Alaknanda catchment is characterized by monsoonal climate with frequent intense rainstorms, amount ranging between up to 200-250 mm per hour (Joshi and Kumar, 2006). The amount of rainfall varies, depending on the location of the place to the windward or leeward side of the high ridges. Observation of daily rainfall data, available since 1976- 2005 received from the Joshimath rain gauge station in the study area is indicative of a mean annual rainfall of about 1137 mm and a general pattern of monthly rainfall is evident. It can be noted that 50% of annual rainfall is received between the monsoon months of July and Mid-September. The area also experiences 20% of annual rains during February and March.

AVERAGE DISTRIBUTION OF RAINFALL BETWEEN 1976-2005

53.95

101.42

130.16

66.99 72.49 81.28

226.19219.61

99.34

37.0716.51

32.53

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

MONTHS

AV

ER

AG

E M

ON

THLY

RA

INFA

LL

Figure: 3-4. Average monthly distribution of rainfall between 1976 and 2005

3.2.6. Vegetation

The Higher Himalayas are characterised by a low to moderate vegetation density. The Greater Himalayas are represented by Pinus roxburghil, Quericus incan, Q. dilatata, Q Semecarpipla, Pinus excelsa, Albies pindow, Betula utilis and Rhododendron(Tewari). However, most of the slopes in the upper reaches are apparently barren with some shrubs, rhododendrons, mosses, lichens, and wildflowers. Terrace farming is done for potatoes, pulses and barley on gentle to moderate slopes.

3.2.7. Road network and Settlements

In the study area, National Highway 58 mostly aligned along the river Alaknanda runs for about 78km from Pipalkoti to Badrinath. The highway has been made by excavations in the highly jointed friable rocks, fluvio-glacial material and talus deposits on the valley slopes. The important settlements in this area are Joshimath, Hanuman chatti, Lambagarh, Pandukeshwar, GovindGhat, Helang, Pipalkoti and the main centres of pilgrimage are Badrinath and Hemkund Sahib. The towns are generally established along the road. The expansion of this road together with rapid urbanisation has rendered these “marginally stable” (Popescu, 2005) hill slopes, apparently more susceptible to failures.

15


__________________________________________________________________________________

Figure: 3-5. Road map showing the NH-58

3.3. Landslides in the Study area

In the present study the term “landslide” is used to include various types of slope failures on natural and man-modified slopes. The slope failures observed in the basin, involving rock and/or debris demonstrated a variety of movement such as, slide, fall, and flows. However, the majority of landslides observed were noted to be shallow translational debris slides, some of which subsequently mobilized into debris flow. While, most of the landslides originate in steep colluvial slopes, some landslides originate in weathered bedrock along multiple joints (e.g. Patalganga landslide). Many of the landslides represent composite events i.e. more than one landslide has occurred at the surrounding region and some single large events. The major landslides have been discussed as follows: Lambagarh Landslide is situated 20 km from Badrinath and is a highly active debris slide zone extending for over a km. The village Lambagarh is situated above the slide zone. Several casualties are reported from this area every year. This is a natural slide in which debris material of fluvio-glacial origin and also weathered rock mass is failing. Helang Landslide, noted first in July 1970 is located near Helang Village on the left bank of the river Alaknanda. The landslide zone is located near the confluence of river Karamnasa and Alaknanda. The height of the crown and toe of the landslide from the road is about 30 m and 40 m respectively. About 250 m of the road has been affected by this landslide. Moreover the village Helang, Warosi and Dungri have also suffered extensive damages and a large area of cultivated land was washed away by this landslide. The slide is located just south of Munsiari thrust and the rocks involved are the quartzite inter-bedded with dolomitic limestone.

16


__________________________________________________________________________________

Figure: 3-6. Landslides in the Study Area

Langsi Landslide is located 62 km south of Badrinath. The crown of landslide is about 150 m above the road and the toe is about 100 m below the road. The slide face is barren but the area has thin grass cover. The rock exposed on the slide face are dolomitic limestone, schist and phyllites. Patalganga landslide is located along the Patalganga Valley at a distance of 64 km from Badrinath. The height of the crown and the depth of the toe of the landslide from the road are 120 m and 100 m respectively. The Patalganga rock slide is active along cliff of the Patalganga Valley. Rock types mainly consists of quartzites, slates and phyllites that are highly crushed, weathered and jointed developing numerous rock falls, topples and slides. The slide face is totally barren but the surrounding area is covered by thick pine forest. The lithology here comprises of crushed and weathered slate and phyllites dipping with an angle of 0-15°. Some of the landslides situated in this zone have also been caused as a result of road expansion.

Figure: 3-7. Landslides in the Study Area (2)

Tangni Landslide is located at milestone 457 on NH-58, 67 km south of Badrinath, at an altitude of 1450 m near Tangni village. The uphill and downhill slopes are 65˚ and 80˚ respectively. The height of the crown from the road is 600 m and the toe of the slide is about 100 m below the road. Some of

17


__________________________________________________________________________________ the trees are observed to be tilted towards road. The landslide is probably caused by the incision of gully, leading to removal of old debris finer material and thus making overall unstable slope. The rocks exposed on the slide face are schist, slate, dolomite and phyllites. Pakhi Landslide is located at a distance of 72 km from Badrinath, near Pakhi Village. The uphill slope is about 50˚, and the down hill slope is 75˚. The height of the crown and depth of the toe of the slide from the road are 65 m and 125 m respectively. The rock exposed is the fractured and weathered dolomite inter-bedded with Phyllites and Slate. Pipalkoti Landslide is located at a distance of 78 km from Badrinath at an altitude of 1350 m near Pipalkoti Village. The road direction and slide face direction is N 20˚. The crown is 250 m above the

road and toe of the slide is 50 m below the road. The slide face is covered with thin grass but the surrounding area is covered with Pine forest, most of which has been deforested. Alaknanda River flows in the valley but does not erode the toe of the slide. The rocks involved in the landslide are mainly Schist, Chloritic Schist and Quartzite. The presence of several thrusts and faults passing in the area has rendered the rock mass weak. The area has also witnessed two major earthquakes in 1991 with epicentre in Uttarkashi and in 1999 with epicentre at Chamoli. The highway has been widened further disturbing the slopes. Thus, landslides here are an outcome of the intrinsic geology, adverse natural topography, i.e., steep slopes underlain by talus deposits, weathered rocks and soils, and man-made modification of these fragile slopes. These inherently unstable slopes frequently fail during rainstorms, often with disastrous consequences.

18


__________________________________________________________________________________

4. Precipitation Thresholds

Precipitation thresholds for initiation of failure in slopes have been assessed in many regions of the world, based on a combination of landslide and rainfall information. Attempts have been made towards prediction of landslide initiation using intensity-duration data for rainstorms (Caine, 1980; Dai and Lee, 2001). The significance of antecedent rainfall in landslide initiation is also recognised(Glade, 1997). Statistically, the critical rainfall amounts have also been established, when an elaborate information on rainfall and landslide occurrence is available (Chleborad, 2003; Kim et al., 1991; Terlien, 1998b) The presence of well distributed networks of rain gauges and a well-maintained database on the landslide location and initiation is essential to establish reliable thresholds, by statistical and deterministic methods. However, in Garhwal Himalayas, the Alaknanda river catchment has very limited rainfall information recorded. The spatial variability of geological and anthropogenic factors further complicates the defining of a precise threshold for triggering of landslides. Thus, an analytical approach is adopted for the present study to explore a possible relationship between rainfall magnitudes and slope failure initiation, based on which a minimum threshold has been determined for Garhwal Himalaya.

4.1. Rainfall data

The daily data recorded at the rain gauge in Joshimath, Badrinath was obtained from 1976 to 2005. Similar data was obtained for Pipalkoti between 2004 and 2006 from Border Roads Organisation (BRO), Joshimath.

Figure: 4-1. ETM Image showing the geographic location and elevation of the three rain gauges in the

study area

19


__________________________________________________________________________________ It is suggested that the measurement of rainfall in landslide investigations should be site-specific to each slope failure but as this does not exist for the entire study area and the basin is largely ungauged, the rain gauge at Joshimath was taken as the main reference station. Although the spatial variability of rainfall is not considered because of data limitations, the rain gauge at Joshimath, for its location in the centre of the study area and its topographic setting, is believed to be important for the study and representative of the conditions at the site of instability. Also, based on the interaction with the local people it was realised that especially during wet seasons there is not much variation in rainfall across the area. However, the pre-monsoon showers often occur as isolated rain events which may not be distributed evenly over the catchment.

MONTHLY RAINFALL DISTRIBUTIONbetween 1987-2005

0.00

100.00

200.00

300.00

400.00

500.00

600.00

JAN

FEBMAR

APRMAY

JUN

JUL

AUGSEP

OCTNOV

DEC

MONTHS

RA

INFA

LL A

MO

UN

T

MINIMUM AVERAGE MAXIMUM

Figure: 4-2. Distribution of average, minimum, and maximum of monthly rainfall at Joshimath

rain gauge between 1987 and 2005

ANNUAL DISTRIBUTION OF RAINFALL

0.00

500.00

1000.00

1500.00

2000.00

2500.00

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

annual rainfall Figure: 4-3. The distribution of rainfall in Joshimath over the years between 1987 and 2005

The daily rainfall data from Pipalkoti was available only from August 1, 2004 onwards. It was taken as reference for the landslides that occurred in 2006.

20


__________________________________________________________________________________ Rainfall in August 2004 & August 2005

0

10

20

30

40

50

60

70

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00rainfall in Pipalkoti (mm)

rain

fall

in Jo

shim

ath

(mm

rainfall in 2005 rainfall in 2004 Figure: 4-4. Scatter plot to show the relationship between the rainfalls recorded at Pipalkoti and

Joshimath in August 2004 and August 2005. The rain gauge data from Badrinath had incomplete entries due to gauge malfunction owing to very low temperatures in the area. Thus, it was found insufficient to estimate the antecedent cumulative rainfall. Moreover, this rain gauge was maintained only between June and November and could not be used for the entire period of study. However, as the monsoon months of July and August are important for landslide instigation, it was decided to scrutinize the daily precipitation records from Badrinath rain gauge as well.

4.2. Landslide data

Historical data on landslide incidences between Pipalkoti and Badrinath was mainly obtained from the daily road damage report maintained by the Border Roads Organisation (BRO), supplemented by personal accounts and news paper reports. The report contains information about the road blockage along with the reason of the blockage. The time when the obstruction is first reported is also mentioned in the report along with the time taken to clear the hindrance. As landslides are the primary causes of blocking roads in Himalayan terrains the information on the day and location of the landslide as well as information on the extent of damage was easily extracted from the report. The initiation time of slope failure may differ from the time the road damage was first noted as the area is remote and reporting therefore, may not be accurate in case of single isolated events. Although it was difficult to precisely locate the slides based on these reports but often the time stated in them was used to verify the landslide data extracted from the road damage report. For instance, a landslide were reported on 18th September, 2005 by a leading news daily, Hindustan Times, but this slide had not been mentioned in the road damage report. Therefore, the field information various sources were carefully studied to distinguish early morning landslides from those that occurred in the later half of the day as both will be influenced by different precipitation amounts. However, the information on the time of advent was not available for all landslide incidences and the study was carried out assuming the landslide event to have occurred on the day it was reported.

21


__________________________________________________________________________________

Annual landslide distribution between 1987-2006

0

20

40

60

80

100

120

140

160

180

200

1987198

8198

9199

0199

1199

2199

3199

4199

5199

6199

7199

8199

9200

0200

1200

2200

3200

4200

5200

6

Figure: 4-5. The annual landslide distribution between 1987 and 2006

Published literature and project reports by various agencies such as GSI, CBRI (Roorkee) and WIHG (Dehradun) were studied to obtain accurate records of landslide occurrences in the study area. Local media reports were also searched thoroughly to look for landslide events. It was observed that only those damaging events had been reported, which caused casualties and/ or property loss including loss of cattle and agricultural land. Moreover, these reports often contained a vague description of the landslide location and usually the name of the nearby village was only mentioned. Although it was difficult to precisely locate the slides based on these reports but often the date stated in them was used to verify the landslide data extracted from the road damage report. These reports also proved useful to cross examine if some slides had not been reported in the BRO records. Apart from this field investigation was carried out in two stages: a reconnaissance survey followed by a detailed site study. A GPS survey was carried out to make the landslide initiation point map which was later used to validate SINMAP model results. A landslide inventory was maintained to collect information on landslide type, dimension, lithology, slope & aspect, land use, probable causes, and damage.

4.3. Database Organisation

The landslide information was compiled as mentioned above so as to get accurate distribution of slope failures in time and space. A landslide database since 1987 was organized to include information on the landslide incidences recorded between Pipalkoti and Badrinath. The total number of landslide incidences reported between 1987 and 2006 in BRO documents was 768. Most of these incidences included reactivated old landslides and a few first time events. Now, the landslide events taken up for the study include all slope failures on natural and engineered slopes, but essentially related to rainfall. Thus, landslides that were triggered by the Uttarkashi earthquake of October, 1991 and Chamoli Earthquake of March, 1999 were excluded from the study. The landslides which were mentioned in the road damage reports as most likely to have been caused by blasting in the adjoining region were again, manually filtered out.

22


__________________________________________________________________________________ As the landslides taken are situated along the highway, their locations were entered as a kilometre-stone reading on the highway. Traverses along the route were done to verify the data and ascertain its geographic location. A GPS survey was also carried out to determine the geographic coordinates of every location. It was found during field visit that the some of the old rock falls and soil slips, which were reported in the official records, could not be precisely located on ground as no associated scar or evident damage to the retaining wall was noted. Only those incidences were considered which were confirmed by local people. The information on the location, slope, and lithology were also entered using the landslide field proforma. Separately, the rainfall data from the three stations was arranged as Microsoft Excel sheets and the daily rainfall values were entered according to the dates. The monthly and annual rainfall was calculated. Computations were done using Excel to get the 3-days, 15-days and 30-days rainfall values (prior to the day of failure). As the rainfall data from Joshimath was not available for the year 2006, all those landslide events that occurred between Vishnuprayag (which is located beyond Joshimath, en route Badrinath) and Badrinath were excluded from the database and only those were included that were distributed between Pipalkoti and Joshimath. The corresponding daily, 3-days, 15-days and 30-days rainfall for each landslide day were entered to complete the required database. Landslide data was then re-arranged for all rainy days, i.e. having at least over 1 mm of daily rainfall. It was noted that only 1947 days experienced over 1mm of rainfall including 259 landslide days, out of which 131 days had a single landslide event, while multiple events were reported on the rest of the days.

single event7%

multiple events7%

no landslide86%

Figure: 4-6. Graph showing the distribution of single and multiple landslide events and days

Thus, 609 landslide incidences were extracted from the database and were classified into five categories as per the number of occurrences per day

i. Single event ii. Two events

iii. Three to five events iv. Six to nine events, and v. More than ten events in a day.

The distinction was made on the basis of slope failures recorded in a day.

23


__________________________________________________________________________________ Table: 4-1. Statistics showing the classification of landslide events and number of slides in each

category

Landslide incidences Landslide days

Single event 131 131 Two events 112 56 Three to five events 183 52 Six to nine events 107 15 More than ten events in a day 76 5

Although, the precise timing of a landslide incidence was not available, the consideration of the cumulative precipitation corresponding to immediate preceding 3 days, including the rainfall on the day of landslide occurrence and 15 days rainfall, prior to the 3 days rainfall reduced the effect of error in assessment ((Chleborad, 2003; Chleborad et al., 2006). The 3 days cumulative rainfall was considered to assess the immediate rainfall in previous 72 hours and 15 days prior rainfall was considered to assess the influence of antecedent rainfall on initiation of landslides. Thus, a comprehensive landslide database was generated to understand the influence of rainfall on the commencement of failure in a variety of setups in the catchment.

4.4. Data analysis

A threshold, minimum or maximum is defined as the limits within which a process is most likely to occur. The minimum thresholds are generally established for precipitation induced slope failures to delineate the limit below which they are most unlikely to be triggered, whereas, the maximum thresholds are given to identify the limit exceeding which there is a 100% probability of landslide occurrence. The landslide-triggering rainfall thresholds separate events that resulted from those which failed to triggered landslides and can be defined on an empirical or on physical bases. Empirical thresholds are generally established on the basis of either of the following parameters, namely rainfall intensity, duration, cumulative precipitation and the antecedent rainfall or their combinations. In the present study the precipitation thresholds for landslide initiation are defined based on the relationship between rainfall magnitude reached, over a specified time, at failure and the cumulative antecedent rainfall i.e., the rainfall accumulation over a specified number of days prior to the day on which a slope failed.

24


__________________________________________________________________________________

MONTHWISE LANDSLIDE DISRIBUTION BETWEEN 1987-2005

0

50

100

150

200

250

300

350

JAN FEB MARAPRMAY JUN JUL AUGSEP OCTNOV DEC

No.

of L

ands

lides

0.00

50.00

100.00

150.00

200.00

250.00

300.00

350.00

AV

ER

AG

E M

ON

THLY

RA

INFA

LL

Landslide monthly average rainfall

Figure: 4-7. The landslide distribution coincides with the average monthly rainfall at Joshimath rain gauge observed between 1987 and 2005.

The database was first studied to make certain that the landslide occurrences followed a pattern with the rainfall. The landslide incidences and monthly rainfall were compiled to confirm that maximum landslides occurred during the wet seasons (Fig 4.7 and 4.8). Based on the statistics for the years between 1987 and 2006, it was observed that in all 511 landslide incidences had occurred only in July and August, which is about 67% of the total landslides reported. A rise in landslide activity was observed in the months of July but major landsliding starts as the monsoon progresses. Maximum landsliding occurs in August, when the daily precipitation is high and the slope forming materials are also saturated by the antecedent rainfall during July. The situation continues up to mid-September and considerable landslide incidences are reported during this time. Significant landslide activity is also noted in February and March, which can again be correlated with the spring showers and snow melt. A year-wise comparison (Fig 4.8) was performed between 1987 and 2005 to see the relation of the slope failures with the monthly precipitation. In the graphs the months are denoted by the x-axis, the primary y-axis refers to the no. of landslide incidences recorded per month, and the secondary y-axis represents the monthly rainfall in that particular year. It is to be borne in that the scale of the axes differs in each graph.

1987

0

2

4

6

8

10

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

MO

NTH

LY R

AIN

FALL

landslide incidences monthly rainfall

1988

0

20

40

60

80

100

120

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

050

100150

200250

300350

MO

NTH

LY R

AIN

FALL


25


__________________________________________________________________________________

1989

0

5

10

15

20

25

30

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

300

350

MO

NTH

LY R

AIN

FALL


1990

0

24

68

1012

14

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50100

150200

250300

350

MO

NTH

LY R

AIN

FALL


1991

0

1

2

3

4

5

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

300

MO

NTH

LY R

AIN

FALL


1992

0

1

2

3

4

5

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

MO

NTH

LY R

AIN

FALL


1993

0

1

2

3

4

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

MO

NTH

LY R

AIN

FALL


1994

0

1

2

3

4

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

020406080100120140160180

MO

NTH

LY R

AIN

FALL


1995

0

1

2

3

4

5

6

7

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

050100150200250300350400

MO

NTH

LY R

AIN

FALL


1996

0

1

2

JAN

FEBMAR

APRMAY

JUN

JUL

AUGSEP

OCTNOV

DEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

300

MO

NTH

LY R

AIN

FALL


26


__________________________________________________________________________________

1997

0

1

2

3

4

5

6

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

020406080100120140160180200

MO

NTH

LY R

AIN

FALL


1998

02468

1012141618

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

100

200

300

400

500

600

MO

NTH

LY R

AIN

FALL


1999

0

1

2

3

4

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

050100150200250300350400

MO

NTH

LY R

AIN

FALL


2000

0

5

10

15

20

25

JAN

FEBMAR

APRMAY

JUN

JUL

AUGSEP

OCTNOV

DEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

300

MO

NTH

LY R

AIN

FALL


2001

012

345

678

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

300

350

MO

NTH

LY R

AIN

FALL


2002

02468

10121416

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

MO

NTH

LY R

AIN

FALL


2003

02468

101214161820

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

050100150200250300350400

MO

NTH

LY R

AIN

FALL


2004

0102030405060708090

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

050100150200250300350400450

MO

NTH

LY R

AIN

FALL


27


__________________________________________________________________________________

2005

02468

1012141618

JAN

FEBMAR

APRMAY

JUN

JULAUG

SEPOCT

NOVDEC

LAN

DSL

IDE

INCI

DEN

CES

0

50

100

150

200

250

300

MO

NTH

LY R

AIN

FALL


Figure: 4-8. Distribution of landslides and monthly rainfall over the years.

Overall, maximum landsliding was observed in July and August. An exception was noted in the year 1990, when seventeen incidences of landslides were reported in February and March, while only 3 incidences were reported in July and August. A cross-check with the rainfall data revealed that no major rain event had occurred in the monsoon months of 1990, whereas, significant rainfall events had occurred the same year in February and March. Although, the landslides that occurred in 1990 are an exception to the general assumption that July and August months are the most crucial for slopes to fail, nevertheless, the role of precipitation is highlighted.

Table: 4-2. The landslide incidences in 1990 and associated daily rainfall.

Landslide day No. of slides Daily rainfall (mm) 13-Feb-90 2 35 28-Feb-90 3 34 21-Mar-90 1 25 22-Mar-90 9 97 19-Jul-90 1 8

20-Aug-90 1 26 Also for the year 1996, there was no landslide reported in the monsoon months. The daily rainfall in the aforementioned period exceeded the normal limits on three instances; 28.2 mm on 10th July, 48.6 mm on 13th July and 36.8 mm on 5th August, but no incidence was reported. Even though, the chance that landslides were not reported cannot be totally ruled out, an explanation was offered that on the above dates even with high daily rainfall values the precipitation thresholds (see equations 4.1- 4.4) were not exceeded. The results of comparison between the average monthly rainfall distribution and landslide incidences recorded over the months between 1987 and 2005 (see Fig 4.7) substantiate that precipitation is the prime triggering mechanism responsible for the failure of the inherently vulnerable slopes. The high number of landslide events in July and August is also indicative that in Garhwal Himalayas, the monsoon months are most critical for slope failure initiation than the remaining days of the year. To establish the thresholds for landslide initiation in the study area the database was scrutinised to identify the rainfall magnitudes over different durations to separate combinations of daily and antecedent rainfall that triggered landslides and then establish the probable landslide triggering thresholds. A review of methodologies proposed in the literature shows that the number of antecedent days must be carefully selected (Terlien, 1998a). Thus, various antecedent rainfall intervals (3-, 15-, and 30-day prior) were considered to calculate the cumulative antecedent rainfall which influenced

28


__________________________________________________________________________________ landslide initiation the most. The database was re-arranged to observe the number of landslides occurring in a day along with the daily, 3-day, 15-day, and 30-day rainfall A comparative analysis was performed to study the relationships between daily and prior 3-day rainfall (Fig 4.9), daily and prior 15-day rainfall (Fig 4.10), daily and prior 30-day rainfall (Fig 4.11), over a period of 20 years. The minimum thresholds for each combination of landslide triggering rainfall cumulative are shown by the green line which is used to demarcate the lower bound precipitation below which no slope failure initiates. These thresholds are identified visually, on the scatter plots between the daily rainfall and the cumulative antecedent rain. Daily / Prior 3-day Rainfall The relationship between the landslide occurrences with the daily and the 3 days prior rainfall (see Fig 4.9) is defined by the equation AT1 = R1 + 1.4696R3 - 14.645 …………………………………….. (4.1) Where, R1 is the daily rainfall measured on the landslide day, R3 is the 3 days rainfall prior to the landslide day, and AT1 is the minimum probable threshold required for a single or more landslide event to occur.

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

160.00

0.00 50.00 100.00 150.00 200.003 DAYS

DA

ILY

1 event 2 events 3-5 events 6-9 events >10 events

Figure: 4-9. The scatter plot based on the daily and prior 3-day rainfall on the landslide days However, as observed the initiation of landslides is not clearly defined mainly due to lack of antecedent rainfall influence, daily and 3-days rainfall mainly represents the triggering rainfall than the antecedent moisture condition.

29


__________________________________________________________________________________

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

160.00

0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 400.00 450.0015 DAYS

DA

ILY


Figure: 4-10. The scatter plot based on the daily and prior 15-day rainfall on the landslide days The minimum threshold defined on the basis of the relationship between daily and 15 days prior rainfall total (see Fig 4.10) is given by the following equation BT1 = R1 + 0.548R15 - 35.032 …………………………………….. (4.2) Where, R1 is the daily rainfall measured on the landslide day, R15 is the 15 days rainfall prior to the landslide day, and BT1 is the minimum probable threshold required for a single or more landslide event to occur. The threshold BT1 implies that the daily rainfall contributes more than the prior 15 days precipitation cumulative in the beginning of the rainy season. At least 50 mm of cumulative rainfall is required with daily rainfall of 25 mm or above to initiate slide, thereafter the antecedent conditions play an important role. Although the lower threshold is well defined for 1-2 events, for higher number of events, the threshold is not well defined; particularly the upper limit for daily rainfall is unreasonably high if storm events corresponding to 10 events or more are considered.

0.00

20.00

40.00

60.00

80.00

100.00

120.00

140.00

160.00

0.00 100.00 200.00 300.00 400.00 500.00 600.0030 DAYS

DA

ILY


Figure: 4-11. The scatter plot based on the daily and prior 30-day rainfall on the landslide days

30


__________________________________________________________________________________ The minimum threshold defined on the basis of the relationship between daily and 30 days prior rainfall total (see Fig 4.11) is given by the following equation CT1 = R1 + 0.491R30 – 51.293 …………………………………….. (4.3) Where, R1 is the daily rainfall measured on the landslide day, R30 is the 15 days rainfall prior to the landslide day, and CT1 is the minimum probable threshold required for a single or more landslide event to occur. The threshold CT1 implies that the daily rainfall is more significant than the prior 30 days precipitation cumulative in the beginning of the rainy season. At least 60 mm of cumulative rainfall is required with daily rainfall of 25 mm or above to initiate slide, thereafter the antecedent conditions play an important role. Although the lower threshold is well defined for 1-2 events, for higher number of events, the threshold is not well defined. Based on the above analysis, the influence of 1-3 days rainfall is clearly seen as significant for landsliding in the initial period when antecedent rainfall is around 50-60 mm. It is also observed that most of the slides were not reported on the same day as rainfall but on the following day. For example, on 17th August 1988, 26 landslides were reported, when the same day’s gauge records show 1.2 mm of daily rainfall. Investigation of the rainfall data revealed that on the preceding day, 59.3mm of rainfall was recorded and it was responsible for triggering the landslides. The role of antecedent rainfall is somewhat subdued, owing to the fact that the majority of the reported slides are related to road expansion activities (Langsi, Marwari, and Vishnuprayag) and are influenced by the 1-3 days rainfall. Moreover, it is seen that most of the slope failures are observed in the glacio-fluvial colluvium (Lambagarh slide zone) and some in weathered rocks such as slates and phyllites (Patalganga). In order to improve the lower threshold for different storm events, a scatter plot (Fig 22) was included in the analysis to show the relationship between 3-day total (including the daily rainfall) and the prior 15-day rainfall (preceding those 3-days). As explained before, this was done to include rainfall in preceding 72 hours reducing the error due to time of reporting and influence of immediate preceding triggering storm event. It was observed that with a 3-day rainfall cumulative, which includes the previous day’s rainfall along with the daily rainfall, the critical conditions prevailing at the time of failure are better defined. The influence of the antecedent rainfall is difficult to quantify as it depends on several factors such as the hydro-geology of the slope forming materials and is subject to a lot of uncertainty. However, the 15 days period seems to be logical to represent the antecedent conditions as for many years it was observed that 50-60 mm rainfall was a prerequisite for landslides to initiate by a single day storm event with 25 mm of rainfall, thereafter the 15 days prior rises to 80-90 mm in the middle of July, when daily rainfall of even 10 mm can cause slides. To avoid ambiguity, the landslide days for which the daily rainfall was less than 1mm were also excluded while constructing the scatter plot between the 3 day cumulative (including daily rainfall) and prior 15 days. The validity of this assumption was verified with the landslide data of 2004, which shows only 4 slides to occur when the daily rainfall was less than 1 mm, which may be attributed to reporting on the following day or variability in rainfall measurement and event location.

31


__________________________________________________________________________________

0.00

50.00

100.00

150.00

200.00

250.00

300.00

350.00

0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 400.00 450.0015 DAYS ANTECEDENT

3 D

AY

TO

TA

L


Figure: 4-12. Scatter plot based on the 3-day total and 15-day prior antecedent rainfall. (Note: The lower bound precipitation threshold represented by the green coloured line was visually identified on the scatter plot and was checked for various storm events starting from 1987 till 2006. ) The minimum probable threshold index (see Fig4.12) based on the analysis of the 3 day rainfall total and the prior 15 days accretion associated with the slope failures reported between 1987 and 2006, is defined by the equation given as follows, T1 = R3 + 1.5351R15

– 82 …………………………………….. (4.4) Where, R3 is the 3 day rainfall cumulative including the rainfall on the landslide day, R15 is the 15 days antecedent rainfall prior to the 3 day total, and T1 is the minimum probable threshold required for a single or more landslide event to occur. The analysis is carried further with the 3-day rainfall amounts th12at occurred immediately prior to the landslide events and antecedent 15-day precipitation that occurred prior to the 3-day total. The approximate lower bounds of landslide-triggering precipitation thresholds for the single and multiple events were again visually identified on the scatter plot and are illustrated in Fig 4.13.

32


__________________________________________________________________________________

0.00

50.00

100.00

150.00

200.00

250.00

300.00

350.00

0.00 50.00 100.00 150.00 200.00 250.00 300.00 350.00 400.00 450.0015 DAYS ANTECEDENT

3 D

AY

TO

TAL


Figure: 4-13. Scatter plot based on the 3-day total and 15-day prior antecedent rainfall. (Note: The lower bound precipitation thresholds for single and multiple events, shown by separate lines, were visually identified on the scatter plot.) Fig4.13 The analysis based on the 3-day and prior 15 days rainfall cumulative resulted in five distinct minimum precipitation thresholds causing single event, two, three to five, six to nine, and more than ten landslide events recorded in a day. However, the threshold for single and 2-events are very close, therefore, clubbed together in the analysis. The minimum probable threshold for the various landslide events are discussed as follows:

i. Minimum probable threshold, T2, for two landslide incidences on a day, is defined by the following equation, T2 = R3 + 1.5351 R15

– 90 ………………………………………….... (4.5)

ii. Minimum probable threshold, T3, for three to five landslide incidences on a day T3 = R3 + 1.5351 R15

– 125 ………………………………………….. (4.6)

iii. Minimum probable threshold, T6, for six to nine landslide incidences on a day T6 = R3 + 1.5351 R15

– 179 ………………………………………….. (4.7)

iv. Minimum probable threshold, T10, for more than ten landslides on a day T10 = R3 + 1.5351 R15

– 328.62 ………………………….………….. (4.8) The precipitation thresholds thus defined, are inferred as an approximate lower-bound threshold above which the particular level of landslide incidences (i.e. 1-2, 3-5, 6-9 and >10 events per day) are more likely to occur on the vulnerable slopes. However, the threshold for single and 2-events are very close, therefore, clubbed together in the analysis. Precipitation threshold for Man- modified slopes The statistics were also examined to understand the slope failure mechanism in man-modified conditions. A human influence is evident on the instability of the marginally stable slopes. About 48% of the total landslide incidences reported was found to be directly caused by anthropogenic activities such as road expansion and improper drainage. The precipitation threshold required to initiate instability in man-modified slopes were established based on the 3-day and prior 15 days rainfall

33


__________________________________________________________________________________ cumulative. The following figure illustrates the lower limit below which a marginally stable man-modified slope is unlikely to fail.

0.00

50.00

100.00

150.00

200.00

250.00

300.00

350.00

0.00 100.00 200.00 300.00 400.00 500.0015 DAY ANTECEDENT

3 D

AY

TO

TAL

1 event 2 events 3-5 events 6-8 EVENTS

Figure: 4-14. Minimum probable threshold for man-modified slopes shown by the green line visually identified on the scatter plot

The minimum probable threshold, TM, required for a single or more landslide event to occur on a man-modified slope is defined by the following equation.

TM = R3 + 0.2664 R15 – 43.79 …………………………………….. (4.9)

It was observed that the threshold required for failure to advance in the engineered slopes is much below that required to bring down a natural slopes. The study shows that the influence of the antecedent rainfall is not substantial for destabilizing such slopes The rational explanation for this is that the excavation for the hill road in the study area is mostly done in weathered rock masses and colluvium, which are most likely to fail even, once disturbed without significant antecedent rainfall (Corominas and Moya, 1999).

4.5. Assessment of the threshold during Monsoon

The minimum threshold derived from the 3 day rainfall and prior fifteen days accumulation is used to understand the landslide – rainfall relationship in the Alaknanda catchment. The evaluation of the thresholds is done to verify if they can distinctly separate rainfall events that triggered slope failures from those that did not and also know the probability that a slope fails if the thresholds are exceeded. The assessment of the minimum thresholds is done for the years 2002, 2003, 2004, and 2005 based on the landslides reported in the months of July and August in each year. The July and August months are considered, as the results so far indicate that the monsoons are crucial for landslide activities. Moreover, the uncertainty regarding the spatial variability of the rainfall is more or less insignificant for the monsoon season. In the recent past, in 2004 maximum number of landslides had occurred, therefore, considered for validation of threshold. As shown before, maximum number of slides occurs in July-August, therefore, daily, 3-days and 15 days prior precipitation data of 60 days in 2004 were analysed. It was observed that 87% of days 1-2 landslide events occurred when threshold exceeded for similar events. Now considering the occurrence 1-2 events as marker of initiation of landslides, threshold for 3-5 events, 6-9 events and more than 10 events were analysed. It was observed that 95% of the days, landslides were reported when T3 had exceeded. Similarly 96% of days, landslides were reported when T10 had exceeded. It was observed that when most extreme event threshold T10 was just positive around 2.5,

34


__________________________________________________________________________________ the T6 was around 150, T3 was around 200 and T1 was around 250, which suggests that T1=250 represents the major storm event when 10 or more landslides are expected and T1 = 40 shows the initiation of about 87% of slides. Therefore, it can be concluded that the defined threshold is indicative of the triggering conditions leading to landslides in the prevailing geological conditions in the study area.

4.5.1. Assessment of the thresholds for July and August, 2002

In 2002, on two occasions, threshold had exceeded and landslides were observed during initial period of threshold rise and the extreme threshold was not exceeded showing relatively less extreme rainfall conditions for initiation for landslides in 2002.

2002

0

2

4

6

8

10

01-J

ul-0

2

08-J

ul-0

2

15-J

ul-0

2

22-J

ul-0

2

29-J

ul-0

2

05-A

ug-0

2

12-A

ug-0

2

19-A

ug-0

2

26-A

ug-0

2

land

slid

es

0.00

10.00

20.00

30.00

40.00

50.00

60.00

daily

rain

fall

slides daily

Figure: 4-15. Daily distribution of landslide incidences and rainfall in July and August 2002

2002

-300.000

-200.000

-100.000

0.000

100.000

200.000

300.000

1-Ju

l-02

8-Ju

l-02

15-J

ul-0

2

22-J

ul-0

2

29-J

ul-0

2

5-A

ug-0

2

12-A

ug-0

2

19-A

ug-0

2

26-A

ug-0

2

T1 T3 T6 T10

Figure: 4-16. Graph showing the daily variation in the thresholds and landslide events for July and August, 2002.

(Note: The thresholds for a single event, 3-5 events, 6-9 events and events>10 in a day are shown by separate lines. Similarly, the landslide events are shown in separate colours, corresponding to a particular landslide event. )

35


__________________________________________________________________________________ 4.5.2. Assessment of the threshold for July and August, 2003

In 2003, the thresholds had exceeded on three occasions and occurrences of slides were associated with rise in the threshold values and only once the extreme threshold was exceeded for 2 days only showing the relatively less extreme precipitation events in 2003.

2003

0

2

4

6

8

10

01-J

ul-0

3

08-J

ul-0

3

15-J

ul-0

3

22-J

ul-0

3

29-J

ul-0

3

05-A

ug-0

3

12-A

ug-0

3

19-A

ug-0

3

26-A

ug-0

3

land

slid

es

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

daily

rain

fall

slides daily Figure: 4-17. Daily distribution of landslide incidences and rainfall in July and August 2003

2003

-300.000

-200.000

-100.000

0.000

100.000

200.000

300.000

1-Ju

l-03

8-Ju

l-03

15-J

ul-0

3

22-J

ul-0

3

29-J

ul-0

3

5-A

ug-0

3

12-A

ug-0

3

19-A

ug-0

3

26-A

ug-0

3

T1 T3 T6 T10

Figure: 4-18. Graph showing the daily variation in the threshold and landslide events for July

and August, 2003. Note: The thresholds for a single event, 3-5 events, 6-9 events and events>10 in a day are shown by separate lines. Similarly, the landslide events are shown in separate colours, corresponding to a particular landslide event.


2004 was an exceptional year with large number of slides reported in the recent past and it shows on 3 occasion, thresholds had exceeded by large values and the extreme event threshold was also exceeded on two occasions. As observed in other cases, the steep rise in threshold values is associated with landslide events.

36


__________________________________________________________________________________ .

2004

0

2

4

6

8

10

01-J

ul-0

4

08-J

ul-0

4

15-J

ul-0

4

22-J

ul-0

4

29-J

ul-0

4

05-A

ug-0

4

12-A

ug-0

4

19-A

ug-0

4

26-A

ug-0

4

land

slid

es

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

daily

rain

fall

slides daily Figure: 4-19. Daily distribution of landslide incidences and rainfall in July and August 2004

2004

-400.000

-300.000

-200.000

-100.000

0.000

100.000

200.000

300.000

400.000

1-Ju

l-04

8-Ju

l-04

15-J

ul-0

4

22-J

ul-0

4

29-J

ul-0

4

5-A

ug-0

4

12-A

ug-0

4

19-A

ug-0

4

26-A

ug-0

4

T1 T3 T6 T10

Figure: 4-20. Graph showing the daily variation in the threshold and landslide events for July and August, 2004.

(Note: The thresholds for a single event, 3-5 events, 6-9 events and events>10 in a day are shown by separate lines. Similarly, the landslide events are shown in separate colours, corresponding to a particular landslide event.)


In 2005, as observed, there is sudden rise in the threshold in three occasions, the rise associated with maximum values during 5-10th August was associated with maximum number of slides, during which also T10 had also exceeded the limit.

37


__________________________________________________________________________________

2005

0

2

4

6

8

10

01-J

ul-0

5

08-J

ul-0

5

15-J

ul-0

5

22-J

ul-0

5

29-J

ul-0

5

05-A

ug-0

5

12-A

ug-0

5

19-A

ug-0

5

26-A

ug-0

5

land

slid

es

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

daily

rain

fall

slides daily

Figure: 4-21. Daily distribution of landslide incidences and rainfall in July and August 2004

2005

-400.000

-300.000

-200.000

-100.000

0.000

100.000

200.000

300.000

400.000

1-Ju

l-05

8-Ju

l-05

15-J

ul-0

5

22-J

ul-0

5

29-J

ul-0

5

5-A

ug-0

5

12-A

ug-0

5

19-A

ug-0

5

26-A

ug-0

5

T1 T3 T6 T10

Figure: 4-22. Graph showing the daily variation in the threshold and landslide events for July and August, 2005.

(Note: The thresholds for a single event, 3-5 events, 6-9 events and events>10 in a day are shown by separate lines. Similarly, the landslide events are shown in separate colours, corresponding to a particular landslide event.)

38


__________________________________________________________________________________

5. Generation of Digital Elevation Model (DEM) for Modelling

Acquisition of stereoscopic data for DEM generation has traditionally been done using across track sensors (e.g. SPOT) and along-track, from two different orbits e.g. (IKONOS). As a consequence a time-difference was observed between the scenes of the stereo-pair. This also leads to difficulties during GCP transfer and image matching for DEM generation. Thus, IRS-P5 (Cartosat-1) data was used for the generation of the Digital Elevation Model of the study area. The satellite has two identical panchromatic sensors kept on the platform with fore, +26o and aft, -5o tilt respectively. This system is capable to provide stereo data without any time difference between two scenes. The Ground control points (GCP), collected using Differential GPS technique were used to refine the orbital parameters. The methodology adopted is as follows:

GCP identification (Pre-field) GCP collection by Differential GPS (in Field) Post-processing of DGPS data (using Leica’s Ski-Pro software) Image Matching in Leica’s Photogrammetry suite DTM extraction using Leica’s OrthoBASE Pro

5.1. GCP collection

A minimum of three GCPs, each having an associated X, Y, and Z coordinate, are required for the purpose of ortho-rectification of an image and the GCPs should be evenly distributed to ensure precision (LPS manual). For the study, the Fore and Aft scenes were studied carefully in the pre-field stage itself to identify distinct features such as, bridges, road crossings, sharp bends, etc. for collection of ground control points. Twenty-five “sharp” points were marked for the purpose as they could be easily located on ground as well as on the image. Their accessibility on the ground was also an important consideration. The estimates of approximate co-ordinates of GCPs in WGS 84 were also taken in pre-field stage. Field survey for GPS data collection, by Differential GPS technique was carried out for seven days. It was decided to set up the reference station in BRO’s Officer’s mess in Joshimath as it lies roughly in the centre of the Cartosat-1 image and also the area is under high security. Upon confirmation of GCP location on ground, DGPS observation at rover was taken for an hour to get a Geodetic Dilution of precision (GDOP) less than 2; however this was achieved for some locations only). At times, due to improper satellite geometry, the GDOP values remained very high and observation time was increased. For getting 3 dimensional coordinates, at least four satellites above cut-off angle of 15° were required. Sometimes owing to the low no. of satellites, the observation did not start and thus, those points could not be taken. A total of 23 GCPs were finally collected.

39


__________________________________________________________________________________

Figure: 5-1. Cartosat-1 image showing locations of probable GCPs.

5.2. Post-processing of DGPS data

The field GPS data of base and rover stations were processed in DGPS mode using Leica’s Ski-Pro. The field recordings were first transferred to the system and the points were assigned as reference and rover accordingly. Single point processing (SPP) was first done for the base points and then the rover points were processed with respect to the base. The points which could not be accurately resolved were post-processed to remove ambiguities. However, only seventeen points could be determined precisely.

40


__________________________________________________________________________________

Figure: 5-2. The post-processing of Ground Control Points using Leica’s Ski-Pro software

5.2.1. DEM generation

The fore and aft scenes of Cartosat-1 data were used to generate the Digital Elevation Model in Leica’s Photogrammetry Suite with OrthoBASE Pro. The scenes were provided with Rational functions (RF) coefficients. These coefficients are used to specify the geometric model, which defines the internal characteristics (i.e. internal geometry of the camera or sensor while capturing the imagery) and external parameters (i.e. original position and orientation of the camera or sensor). The reference coordinate system is assigned a projection in UTM with spheroid and datum as WGS84. A block file is created in LPS and the two scenes added to the frame. The chipping coefficients are directly taken from the .rpc text files provided with the Cartosat-1 data. Pyramid layers, based on a binomial interpolation algorithm and a Gaussian filter, were generated to preserve image contents and save computation time. Internal orientation is done to define the pixel coordinate positions of the calibrated fiducial marks within each image of the block. External orientation is done to define the position and orientation of the perspective centre. If very precise values (i.e. less than a meter) of exterior orientation are imported, the aerial triangulation process can even be skipped. However, for the study, block triangulation method was selected. Two GCPs were assigned a horizontal control as they were located besides cliff. Another GCP taken at Hemkund Sahib Bend (30° 37' 08.52662" N and 79° 33' 41.97771" E) was allocated a vertical control as it could not be matched well on the two scenes. The orbital parameters of fore and aft scenes are refined using eleven ground control points (Fig5.3).

41


__________________________________________________________________________________

Figure: 5-3. Generation of DEM and Ortho image in LPS

The bundle block adjustment technique was applied for aerial triangulation, to establish the geometrical relationship between the two scenes, sensor model, and the ground. This technique is preferred over normal triangulation routines as with this the errors associated with image matching or GCPs are distributed and minimised considerably. A total RMSE (root mean square error) of 1.612 pixels and vertical accuracy of 7.6 m was reached by the above methods. The detailed triangulation report is given in Annexure-1.

Table: 5-1. Summary of the triangulation results

RMSE(meter) Ground X 5.6170540 Ground Y 5.9689932 Ground Z 7.6920795 Image X 1.9643565 Image Y 3.1174111

Total 1.6259

The digitally matched image pair was subsequently used in OrthoBASE Pro, to automatically extract the three-dimensional terrain information. The raster DEMs were then extracted with different cell-sizes i.e. 10m, 15m, and 20m. The DTM extraction report, containing statistical information about the accuracy and quality of the output DEM can be found in Annexure2, 3 and 4. The global accuracy was checked for each DEM. It is suggested that the RMSE value should be less than twice the original pixel size of image (LPSmanual, 2003) i.e. less than 5m for Cartosat-1. The DEMs were subsequently used for the ortho-rectification of the image.

42


__________________________________________________________________________________

Figure: 5-4. The raster DEMs. (5.4.1) 10m resolution DEM, (5.4.2)15m resolution DEM, 5.4.3)

20m resolution DEM.

43


__________________________________________________________________________________

44


__________________________________________________________________________________

Figure: 5-5. Ortho-rectified Image generated using the raster DEMs

The accuracy of the DEM generated from Cartosat-1 can be improved significantly with using well-distributed GCPs. However, the inaccessibility of the terrain poses a serious challenge towards achieving the desired results.

45


__________________________________________________________________________________

6. Application of a GIS-based Physical model for Slope Stability Assessment

With the advancement of modern day technology, the application of GIS in slope stability analyses is not just limited to the spatial and non-spatial database generation and manipulation but it also assists in modelling, as different physically based models can be readily integrated in it. However, the numerical modelling applied to precisely evaluate the pre-failure situation of the slopes and simulate the slope-failure initiation process is usually quite data intensive and often time consuming. However, the area being studied suffers from underdevelopment, making instrumentation and regular data collection very difficult. For the same reason there has been very little data collection here in the past, which again limits the possibilities of a data intensive modelling. Thus, a coupled approach incorporated in GIS, combining a simple steady-state hydrological model and slope stability parameters, is adopted to assess the stability of slopes in a portion of the Alaknanda catchment, Garhwal Himalayas. SINMAP is a coupled infinite slope stability model with steady state hydrology. In mapping terrain instability, it considers that the shallow slope failures are primarily governed by the surface topography. It recognizes that the pore pressure due to increased soil saturation in zones of subsurface flow convergence (Montgomery and Dietrich, 1994) reduces the effective normal stress, which is a function of the shearing strength. The modeling is done on the basis of the infinite plane slope stability model (Fig 6-1 ) given by (Hammond et al., 1992a)

………… (6.1) Where cr is root cohesion [N/m2], cs is soil cohesion [N/m2], θ is slope angle, ρs is wet soil density [kg/m3], ρw is the density of water [kg/m3], g is gravitational acceleration (9.81 m/s2), D the vertical soil depth [m], Dw the vertical height of the water table within the soil layer [m], and the internal friction angle of the soil.

Figure: 6-1. Infinite slope stability model

(Note: The destabilizing components of gravity and the restoring components of friction and cohesion are balanced on a failure plane parallel to the ground surface neglecting the edge effects) The above equation was subsequently modified by (Pack et al., 1998b) in mapping terrain stability by including the concept of a steady state hydrological model.

46


__________________________________________________________________________________ Accordingly, the thickness of material above failure plane is measured perpendicular to the slope and not vertically. The relation between the depth (vertical) and soil thickness (perpendicular) is given as h=D cosθ, (6.2) The relative wetness, which defines the relative depth of the perched water table within the soil layer, is related to the soil thickness as follows, w=Dw/D=hw/h (6.3) The model assumes a dimensionless cohesion (relative contribution of the cohesive forces to slope stability). C= (Cr + Cs)/ (hρsg) (6.4) Where Cr is root cohesion, Cs is soil cohesion, h is soil thickness, ρs is wet soil density, and g is the gravitational acceleration. Substituting the water to soil density ratio, r =ρw / ρs (6.5) Thus, the equation of stability factor used in the model upon including the concepts of steady state hydrology changes to

(6.6) The hydrologic model as applied in SINMAP, based on the recharge rate and the material transmissivity, attempts to estimate the depth of saturation. Pore water pressure is computed assuming a hydrologic steady state with depth of saturated soil computed sufficient to sustain a lateral discharge proportional to the upslope area per unit contour length, defined as the specific catchments area.

Figure: 6-2. The concept of specific catchment area as applied in SINMAP (Pack et al., 1998b)

The model assumes that the discharge at any point in the catchment area is in equilibrium with the effective steady state recharge. Thus, the lateral discharge can be given as, q=Ra (6.7) Where q, is the discharge, and a, is the contributing area. R, is the effective recharge for a critical period of wet season likely to trigger landslides. The other assumption is that for a uniformly distributed hydraulic conductivity, capacity of lateral flux at each point can be denoted as a function of its transmissivity and slope angle, i.e. T sinθ The relative wetness is thus, re-defined as:

(6.8) The equation for the factor of safety thus, becomes

47


__________________________________________________________________________________

(6.7) The uncertainty in the variables, such as, cohesion, angle of internal friction and the parameter R/T is allowed to vary between specified limits. The parameter uncertainty is thus accounted for in the model through the use of uniform probability distribution, given as C ~ U(C1, C2) t ~ U(t1, t2) x ~ U(x1, x2) ......................................................(6.8) Where, c is cohesion, t is the internal friction angle and x is the ratio between R/T. However, the results can be made deterministic if the parameters are measured precisely. The worst case scenario is defined when the values of cohesion and angle of internal friction are the least, and the ratio of R/T is very high. Thus, for a factor of safety more than 1e minimum stability index is thus defined as,

…………………………………. (6.9) Where, c is cohesion, t is the internal friction angle and x is the ratio between R/T. However, the results can be made deterministic if the parameters are measured precisely. The worst case scenario is defined when the values of cohesion and angle of internal friction are the least, and the ratio of R/T is very high. Thus, for a factor of safety more than 1e minimum stability index is thus defined as,

…………………………………. (6.10) The Stability Index (SI) is thus, defined as the probability for a slope to be stable, assuming a uniform distribution of parameters over the uncertainty. When the minimum factor of safety is less than 1, there exists a probability for a slope to fail. SI= Probability (FS>1) When the values of cohesion and angle of internal friction are highest and the ratio R/T is defined by its lower bound.

…………………………..… (6.11) In the case that FSmax < 1, then the slopes are stable i.e. SI = Probability (FS > 1) = 0

48


__________________________________________________________________________________ Table: 6-1. The stability class definitions

The slopes are assigned a stability index as the measure the magnitude of destabilizing factors required for instability and are accordingly classified as stable, moderately stable, quasi-stable, lower threshold, upper threshold and defended. The first three classes suggest that the slopes should not fail if critical limit of parameter range is not exceeded. The slopes falling between the lower and the upper threshold are those with a probability of failure close to 50% i.e. if probability is less then it is in the lower threshold and if surpassed then the slopes are in the upper threshold of instability. For these slopes, instability conditions occur as an outcome of the combined influence of the model parameters. In actual set ups, it is often noted that a slope might fail for reasons other than those modelled in SINMAP. A term “defended” is used to characterize such slopes which remain unstable with or without the parameters specified.

6.1. Digital Elevation Model (DEM)

The slope stability index computations are largely based upon the quality and accuracy of the Digital Elevation Model. Thus, Cartosat-1 data (resolution 2.5m) is used to derive high resolution DEM of the study area. The GCPs were collected in field using Differential GPS technique, processed in Leica’s Ski-Pro software and later used to generate DEM using Leica’s Photogrammetry Suite. The raster DEMs were then extracted with different cell-sizes i.e. 10m, 15m, and 20m. The raster DEM of resolution 10m, generated in automated mode could not be used for modelling as it had many pits and artefacts.

6.2. Model parameters

The model parameters were either measured in field or samples were taken and referred with the existing literature.

6.2.1. Ratio of Transmissivity to Effective recharge (T/R ratio)

The upper and lower bound parameters for T/R were derived from the permeability rates observed within the study area and the precipitation thresholds derived from the previous analysis. Transmissivity is defined by the equation; T=Kb, where K is the hydraulic conductivity (m/day), b is thickness of the saturated zone and T is the transmissivity (m2/day). The hydraulic conductivity for different zones was measured in field using a Double ring infiltrometer. As the instrument was not easily portable and also required at least an hour of observation time, it could only be set up for a few easily accessible locations. Thus, soil samples were taken at different places for conducting a particle size analysis to ascertain the soil type, from which permeability values published in literature were derived. The obtained values of permeability and thickness for each location was entered into excel

49


__________________________________________________________________________________ sheets for analysis. The values of transmissivity were thus, solved for different zones. The mean value of transmissivity obtained for a 5-10 m thick colluviums zone would vary between 600,000 and 1200,000 mm/hr. The effective recharge was assumed for a critical period of wet season likely to trigger landslides.

6.2.2. Soil strength parameters

The important soil strength parameters required in SINMAP are cohesion, angle of internal friction, and soil density. Although, the geology of the modelling basin is diverse the field work confirms that most of the slope failures are initiating in the fluvio-glacial detritus. Thus, the measurements were largely taken on the colluviums and talus accumulations which were associated with the failure. The soil samples were taken for laboratory testing for determination of geotechnical parameters. Texture analysis was also performed to determine the percentage of clay, silt and sand in a sample. The results of the analysis suggest that on an average, the percentage of sand was 67%, of clay 8.2% and silt 24.6%. The important geotechnical parameters associated with the major litho-types in the modelling basin are listed in the table shown below.

Table: 6-2. Geotechnical properties

Material type Cohesion, c (N/m2)

Angle of internal friction, ø(degree)

Unit weight, µ (kg/m3)

1 Colluvium 1000 35 1800 2 Gneiss 2500 40 2500 3 Quartzite 3000 50 2800 4 Gneiss schist 2000 38 2000

SINMAP uses a dimensionless cohesion, the formula for which is given in Eq (6.4). The value of cohesion as obtained from field were 1000 N/m2 for fluvio-glacial material and applying in the above formula the dimensionless cohesion for fluvio-glacial material was computed as 0.55, assuming negligible root cohesion.

6.3. Landslide point map

Landslide inventory was prepared during reconnaissance survey followed by a detailed field work. The landslide point map which was subsequently used for validation of model results was carefully prepared to identify the initiation zones. Landside initiation zones for some accessible slides, located close to the highway, were measured using hand-held GPS receiver. But for most of the slides, image interpretation techniques were applied to identify their initiation zones on the ortho-rectified Cartosat-1 scenes. The head scarp of the slope failure was mapped as points (Terlien et al., 1995) in GIS.

50


__________________________________________________________________________________

Figure: 6-3. Landslide initiation points on Cartosat-1 Ortho-image. The National highway is also

shown.

6.4. Processing of DEM

The Digital Elevation Model was imported as a Grid for ease of analysis. The topographic elevation of each pixel is stored in a matrix node within the grid. The sinks or pits, are the artefacts that arise due to error in the preparation of DEM and are eliminated first using a flooding approach, (i.e. the elevation of each cell of the sink is raised to the elevation of the lowest pour point on the periphery) followed in SINMAP. The flow direction is computed from the DEM based on a D∞ method which assumes a multiple flow direction, as illustrated by the following diagram,

Figure: 6-4. The D∞ method for multiple flow directions followed from (Pack et al., 1998b)

The contributing area or the upslope area (counted in terms of the number of grid cells) is calculated using a recursive algorithm for multiple directions.

51


__________________________________________________________________________________

6.5. Model Run

The SINMAP methodology is applied to a portion of the Alaknanda catchment of 300sq.km area, to characterise the slopes based upon their stability. The standards for gravity (= 9.81 ms-2), material (= 1800 kg/m3) and density of water (= 1000 kg/m3) are set in the beginning, as constant for the entire modelling procedure. These are usually provided as default, but can be changed depending upon the area. As no comprehensive mapping of soils or geomorphology had been done in the area, a single calibration region was used to calibrate the model parameters. However, it was observed during the field work that even though the geology of the terrain is diverse, the slope failures in modelling basin mostly initiate in the colluvium zones of fluvio-glacial origin. Based upon the results of material properties, the value of internal friction angle was taken as 35° and dimensionless cohesion was taken between 0 (assuming cohesion less scenario) and 0.55 for fluvio-glacial materials. The T/R parameter was allowed to vary between 720 m and 1440 m in the calibration region. The modelling parameters taken are considered practical for the colluviums found in the study area. The model is run with varying DEM resolutions of 15m and 20m to study the change in stability index with respect to the resolution of DEM. The stability index is computed cell by cell within the grid. The model is first applied deterministically to derive the stability index for the slopes in the catchment. As most of the slope failures in the modelling basin are associated with the colluvium on the valley slopes, the standards for colluvium as obtained from laboratory tests are taken to run the model. The model was first run for a T/ R ratio as 1440m (T/Rmax) for cohesion less surface with angle of internal friction as 30°. The following scenarios are observed:

Figure: 6-5. Slope Stability index map from 15m DEM (deterministic)

52


__________________________________________________________________________________

Figure: 6-6. Slope Area plot (15 m DEM_deterministic)

Figure: 6-7. Slope Stability index map from 20m DEM (deterministic)

Figure: 6-8. Slope Area plot (20m DEM_deterministic)

53


__________________________________________________________________________________

The probabilistic approach is later adopted to observe the changes in the stability index over a specified range of parameter uncertainty. The calibration parameters are allowed to vary between the upper and lower bounds of each of the input parameters.

Figure: 6-9. Slope Stability index from 15 m DEM (probabilistic)

Figure: 6-10. Slope Area plot (15 m DEM_Probabilistic)

54


__________________________________________________________________________________

Figure 6-11 Slope Stability index from 20 m DEM (probabilistic)

Figure 6-12: Slope Area plot (20 m DEM_Probabilistic)

6.6. Slope Stability Index

The slope stability classes are assigned based on a slope plot of landslide and non-landslide points. The Stability Index of the various slopes in the catchment is generated as a spatially distributed map. As SINMAP derives the important terrain parameters from the Digital Elevation model, the resolution of the DEM is of prime concern. The comparison of results obtained from running the model using different resolution of DEM shows that topographic attributes are best quantified using DEM of cell size 15m. This is suggestive of the fact that the errors (artefacts) in preparing the 10m DEM were high enough and as a result the topography was not well represented by the DEM. The stability index computations are thus done on the grids made from the raster DEMs of resolution 15m and 20m. However, comparison of the statistical analysis shows that the instability is best captured when the 15m DEM, is applied probabilistically. 44% of the total area is classified as stable, while 10% of the

55


__________________________________________________________________________________ area has slopes which are unlikely to fail if critical limits are not exceeded. According to the model output, about 37.5% of the area falls between the lower and upper threshold. The area coming under lower threshold is that in which the probability of failure is less than 50%, whereas slopes under upper threshold are more likely to fail. The model also identifies that 24.5 sq. km of the slopes are defended that is the model failed to account for the causes responsible for the failure, if any in these zones. These slopes are the most vulnerable zones according to the model. Comparing with the slope map it was observed that these slopes had an angle of 75º or more. However, when the model is applied deterministically, i.e. the lower bounds are assigned equal to the upper bounds; the model is unable to account for the instability. It assigns a “defended” class to 41.9% of the slopes in the catchment. 74.4% of the landslides are again characterised in this defended slopes zone. The model has a success rate of 74.4% with a Digital Elevation Model of 15 m resolution, at identifying the slope failure initiation points, when applied probabilistically. Whereas, the Stability Index map generated using the 20m DEM has a success rate of 64.1% only. However, the spatially distributed stability index maps represent the broad classes of instability well in the region which correlates well with the field observation.

56


__________________________________________________________________________________

7. Results and Discussion

7.1. Results of the analysis

7.1.1. Rainfall as a prime triggering mechanism

The first objective of the present work was to correlate the landslide incidences with rainfall to ascertain that rainfall is the main mechanism behind initiation of failure in slopes. Observation of the landslide distribution over the last twenty years shows that it follows a definite trend with the distribution of rainfall in the Alaknanda catchment, India. The rainfall pattern as studied for the period between 1976 and 2006 indicates that about 50% of the annual rainfall is received in July and August, during monsoon and is accompanied with significant landslide activity. Statistics reveal that about 66% of the slides occur between July and August (Fig 4.7). An observation based on the historical data reveals that maximum landslide activity was reported in the years 1988 and 2004 when the bi-monthly average for July and August was above 50% of the annual total. The significance of antecedent rainfall is also well exhibited by the fact that the slope failures are more frequent in the month of August than July.

7.1.2. Precipitation thresholds for landslide initiation

An analytical approach is adopted to establish the relationship between rainfall magnitudes and slope failure initiation. Analysis is done based on antecedent rainfall, for different time intervals before the event. Thus, various antecedent rainfall intervals (daily, 3-, 15-, and 30-day prior) were considered along with the 72-hours precipitation and prior 15 days total, to calculate the cumulative antecedent rainfall which influenced landslide initiation the most. The minimum probable thresholds for a specified level of landslide activity (i.e. number of slides occurring on a single day) were identified. The assessment of the results is done based on the landslide data of July and August months in the years 2002 to 2005. Analysis of the graphs (Fig4.16, 4.18, 4.19, and 4.22) suggests that the slope failure of a specified level usually occurs at the instant, when the threshold exceeds the lower bound for the particular level of activity. A success rate of 92% is observed in 2004 based on the 3-day (72-hr) and previous 15 day (before the 72-hr precipitation) when the minimum threshold exceeds and the daily rainfall is over 1mm. These thresholds have been analysed with respect to initiation of landslides starting from 1988 till 2006 showing validity of this approach. Slopes disturbed by road constriction shows lower threshold thereby indicating the role of precipitation on initiation of landslides on disturbed slopes.

7.1.3. Digital Elevation Model

The Cartosat-1 images (resolution 2.5m) were used to derive a high resolution DEM and the Ortho-rectified image of the study area. The GCPs were collected in the field using Differential GPS technique, processed in Leica’s SKIPRO software and later used to generate the DEM using Leica’s Photogrammetry Suite. A total RMSE (root mean square error) of 1.612 pixels and vertical accuracy of 7.6 m was achieved by the block triangulation technique. The raster DEMs were generated through automated DTM extraction in three resolutions, 10m, 15m, and 20m. The 10m DEM was not taken as it had many spicks impression. These can be reduced

57


__________________________________________________________________________________ through manual editing. For landslide modelling, the DEM generated with cell size 15m was preferred over 20m. However, the prime objective of generating high resolution DEM, in landslide modelling, is not the absolute elevation values but its manifestation of the slopes. The DEM does capture geomorphological features, such as terraces and gullies, but in areas, where cliffs and deep gorges are found, with slopes greater than 80°, the terrain is not accurately represented. The accuracy of the DEM generated from Cartosat-1 can be improved significantly with using well-distributed GCPs. However, the inaccessibility of the terrain poses a serious challenge towards achieving the desired results.

7.1.4. Slope characterisation

A simple physical model is applied in GIS to understand the zones of instability in a portion of the Alaknanda catchment, India. The steady-state hydrological model coupled with slope stability is primarily based on topography. The model recognises instability as a resultant of the combination of the important geo-technical and hydrological parameters. Thus, spatial variability of these factors is quantified as uniform probability distributions and allowed to vary over a specified range. The slope stability classes are assigned based on a slope plot of landslide and non-landslide points. The spatially distributed stability index maps represent the broad classes of instability well in the region which correlates well with the field observation. The model has a success rate of 74.4% with a Digital Elevation Model of 15 m resolution, at identifying the slope failure initiation points, when applied probabilistically. However, when the model is applied deterministically, i.e. the lower bounds are assigned equal to the upper bounds; the model is unable to account for the instability. It assigns a “defended” class to 41.9% of the slopes in the catchment. 74.4% of the landslides are again characterised in this defended slopes zone.

7.2. Related Issues

7.2.1. Precipitation Thresholds

The presence of well distributed networks of rain gauges and a well-maintained database on

the landslide location and initiation is essential to establish reliable thresholds. It is suggested that the rainfall measurements should be made at the site of instability and represent the exact pre-failure conditions of a slope, however, the lack of rain gauges poses a major limitation of the study. The data may be questionable as the measured values at the distant rain gauges may not be an exact representative of the entire area due to differences in slope, aspect and elevation. Moreover, the data from simple rain gauges only gives information on the daily rainfall magnitudes. As a result the role of storm intensity could not be ascertained. The functioning conditions of the rain gauge used are also of concern. Installation of a spatially well distributed network of rain gauges within the basin is highly recommended. These should be preferably located near the zones of instability. Even at the present rain gauge sites, an hourly monitoring is suggested during July and August, when maximum landslide activity is expected.

The entirety of the landslide data also varies. As the landslide incidence data is mainly

extracted from the road damage reports of BRO and other sources, it contains information mainly of those incidences that caused traffic disruption. In most of the cases these are the failures along the road, only rare incidence as that of the Lambagarh slide, which occurred in 2004 on the opposite bank and destroyed the highway, were reported. Also, an uncertainty exists as to the precise time of failure initiation as the time mentioned of a road blockage might vary slightly from that of the slope failure initiation. Thus, a better record of landslide

58


__________________________________________________________________________________ initiation time is needed. Another apprehension is that a considerable number of slides might have gone unnoticed or were not reported due to inaccessibility of the terrain.

The spatial variability of geological and anthropogenic factors further complicates the defining

of a precise threshold for triggering of landslides. Thus, the considering precipitation as the only driving mechanism might be an exaggeration. The database shows that a considerable number of slides occur in the months of February and March. Although these were very well related to the precipitation distribution, however it is recommended to include the air-temperature index to understand the pre-failure conditions of the slopes. It is during this time, that the snow starts to melt and thus, the inclusion of air-temperature in these studies can give significant results in snow-bound areas.

7.2.2. Generation of DEM

The selection of GCPs should be done in the pre-field stage itself. The ideal GCPs should be

readily identifiable in the field as well as on the image. To ensure this, extraction of sharp features, such as the intersection of roads, streams, canals and so on, is suggested.

The accuracy of the Digital Elevation Model relies upon the precision of the Ground control points and how well they are matched on the two scenes of a stereo-pair. When collecting the GCPs, effort must be made to minimise the GDOP values. On many occasions high GDOP values were obtained due to improper satellite geometry. This was resolved by increasing the observation time at some locations. At many places it was observed that the satellite signal was weak and these points had to be dropped.

The ambiguity of some points could not be resolved, even in the post processing stage as the GDOP values were very high. These points had to be dropped.

The accuracy of the DEM also depends on how well the GCPs are matched on the two stereo-pair. This has to be done with pixel level accuracy.

The accuracy of Digital Elevation Model generated from Cartosat-1 Stereo data can be improved with using more accurate and well-distributed GCPs for refining the rational function coefficients.

7.2.3. GIS-based modelling

The Stability Index Mapping (SINMAP) model of (Pack et al., 1998b) is a topography based

model which relies solely on the accuracy and resolution of the DEM. The 15m and 20m resolution DEM were used for modelling the slope instability.

The 10m resolution DEM could not be used for the modelling purpose as it had many sinks and artefacts.

In areas where the terrain accessibility is limited by the absence of road network, scientific instrumentation faces a major challenge. The double ring infiltrometer used for the determination of hydraulic conductivity in field, could not be set up in many places.

For the reasons stated above, parameterization of hydrological properties could not be done for data intensive models like TRIGRS- Transient rainfall infiltration and Grid-based Regional Slope Stability(Baum et al., 2002).

The model does not give adequate weightage to the critical parameters of geology, soil and vegetation, known to influence the stability of slopes. These parameters are modelled as uniformly distributed probabilities between specified limits.

In SINMAP, the parameters can be interactively calibrated to retain the unique characteristics of the topography, rainfall, and soils of a particular study area.

The parameters can also be manually adjusted during calibration to include a high percentage of landslides in the zones of instability. This can be misleading, although attempt has been made to be unbiased.

59


__________________________________________________________________________________ It is observed that, SINMAP failed to model slopes greater than 75° and assigned them a class

called “defended slopes”. These slopes are defined by the model as highly susceptible to failure. However, the “defended slope” zone could not be modelled by the parameter range taken by the model and for these stabilising factors were required to keep them stable.

However, in complex terrains where the hydro-geological characteristics of the slope forming materials are diverse the model can be applied probabilistically.

7.3. Conclusion The present study identifies precipitation as the main mechanism for slope failure initiation in a part of Garhwal Himalayas and defines a lower-bound precipitation threshold, based on 72-hours precipitation and 15 days prior antecedent precipitation, if the daily rainfall is above a specified limit. The evaluation of the thresholds, based on the landslides observed in July and August, shows a high probability of landslide occurrence when the lower-bounds of the threshold are exceeded, under other considerations. The prediction rates can be significantly improved with the availability of a well distributed network of weather stations and better recording of the initiation time of slope failures. The thresholds can be further improved by taking into account the spatial variability of other stability influencing parameters to aid in effective landslide hazard assessment in the Garhwal Himalayas. The SINMAP model takes full advantage of GIS technology to quantify topographic attributes, such as slope, aspect and drainage and based on these limited input parameters it attempts to define the stability of slopes. In Garhwal Himalayas, where the geology is complex and varied, and accessibility of the terrain is problematic, the adopted modeling approach is a viable solution. The modeling technique proves useful to identify and map the hill-slope stability. The spatially distributed stability index maps represent the broad classes of instability well in the region which correlates well with the field observation. The model has a success rate of 74.4% with a Digital Elevation Model of 15 m resolution, at identifying the slope failure initiation points.

60


__________________________________________________________________________________

8. REFERENCES

2002, Annual Report, Natural Disaster Management Division, Ministry of Agriculture. Acharya, G., 2003, GIS Approach for Slope Stability Risk Analysis: A Case Study from Nepal: Master

dissertation thesis, Universiteit Gent, Free University of Brussels., 142 p. Aleotti, P., 2004, A warning system for rainfall-induced shallow failures: Engineering Geology, v. 73,

p. 247-265. Asch, T. W. J. V., J. Buma, and L. P. H. V. Beek, 1999, A view on some hydrological triggering

systems in landslides: Geomorphology, v. 30, p. 25-32. Barredo, J. I., A. Benavides, J. Hervas, and C. J. v. Westen, 2000, Comparing heuristic landslide

hazard assessment techniques using GIS in the Tirajana basin, Gran Canaria Island, Spain: JAG, v. 2.

Baum, R. L., W. Z. Savage, and J.W.Godt, 2002, TRIGRS—A Fortran Program for Transient Rainfall Infiltration and Grid-Based Regional Slope-Stability Analysis, USGS Open-file report 02-424.

Beek, L. P. H. v., 2002, Assessment of the influence of changes in Landuse and Climate on Landslide Activity in a Mediterranean Environment, University of Utrecht, Utrecht, The Netherlands, 363 pp p.

Begueria, S., and A. Lorente, 1999, Landslide Hazard Mapping By Multivariate Statistics: Comparison of Methods and Case Study in the Spanish Pyrenees, DAMOCLES project report.

Borga, M., G. D. Fontana, and F. Cazorzi, 2002, Analysis of topographic and climatic control on rainfall-triggered shallow landsliding using a quasi-dynamic wetness index: Journal of Hydrology, v. 268, p. 56-71.

Brunsden, D., 1999, Some geomorphological considerations for the future development of landslide models: Geomorphology, v. 30, p. 13-24.

Caine, N., 1980, Rainfall intensity-duration control of shallow landslides and debris flows: Geografiska Annaler Series A: Physical Geography, v. 62, p. 23-27.

Campbell, R. H., 1975, Soil slips, debris flows, and rainstorms in the Santa Monica Mountains and Vicinity, Southern California, US Geological Survey Professional Paper, 851: 51.

Chleborad, A. F., 2003, Preliminary Evaluation of a Precipitation Threshold for Anticipating the Occurrence of Landslides in the Seattle, Washington, U.S. Geological Survey Open-File Report 03-463.

Chleborad, A. F., R. L. Baum, and J. W. Godt, 2006, Rainfall Thresholds for Forecasting Landslides in the Seattle, Washington, Area—Exceedance and Probability.

Chowdhury, R., and P. Flentje, 2002, Uncertainties in rainfall-induced landslide hazard: Quarterly Journal of Engineering Geology and Hydrogeology, v. 35, p. 61-70.

Corominas, J., and J. Moya, 1999, Reconstructing recent landslide activity in relation to rainfall in the Llobregat River basin, Eastern Pyrences, Spain: Geomorphology, v. 30, p. 79-93.

Crosta, G., 1998, Regionalization of rainfall thresholds: an aid to landslide hazard evaluation: Environmental Geology, v. 35, p. 131-145.

Crosta, G. B., and P. Frattini, 2000, Rainfall thresholds for triggering soil slips and debris flow, Mediterranean Storms: Proceedings of the 2nd EGS Plinius Conference on Mediterranean Storms, Siena, p. 463-487.

Crozier, M. J., 1986, Landslides: Causes, Consequences and Environment: London, Croom Helm. Cruden, D. M., and D. J. Varnes, eds., 1996, Landslide Types And Processes: Landslides:

Investigation and Mitigation, Transportation Research Board Special Report.247.

Dai, F. C., and C. F. Lee, 2001, Frequency-volume relation and prediction of rainfall-induced landslides: Engineering Geology, v. 59, p. 253-266.

61


__________________________________________________________________________________ Dhakal, A. S., and R. C. Sidle, 2004, Distributed simulations of landslides for different rainfall

conditions: Hydrological processes, v. 18, p. 757-776. Ferreti, A., 2005, Permanent Scatterers technology: A powerful state of the art tool for historic and

future monitoring of landslides and other terrain instability phenomena: Proceedings, international conference on landslide risk Management.

Finlay, P. J., R. Fell, and P. K. Maguire, 1997, The relationship between the probability of landslide occurrence and rainfall: Can. Geotech journal, v. 34, p. 811-824.

Gabet, E. J., D. W. Burbank, J. K. Putkonen, B. A. Pratt-Sitaula, and T. Ojha, 2004, Rainfall thresholds for landsliding in the Himalayas of Nepal: Geomorphology, v. 63, p. 131-143.

Gairola, V. K., 1975, On the petrology and structure of the Central crystallines of the Garhwal Himalaya, Uttar Pradesh: himalyan geology, v. 5, p. 445-468.

Geo-Slope, 1994, Geo-Slope Technical Guide. Giannecchini, R., 2006, Relationship between rainfall and shallow landslides in the southern Apuan Alps (Italy): Natural Hazards and Earth System Sciences, v. 6, p. 357-364. Glade, T., 1997, The temporal and spatial occurrence of rainstorm triggered landslide events in New

Zealand, Victoria University of Wellington, 380 p. Greenway, D. R., 1987, Vegetation and slope stability, in M. G. A. a. K. S. Richards, ed., Slope

Stability. Gururajan, N. S., and B. K. Choudhuri, 1999, Ductile thrusting, metamorphism and normal faulting in

Dhauliganga Valley, Garhwal himalaya: Himalayan Geology, v. 20, p. 19-29. Hammond, C., D. Hall, S. Miller, and P. Swetik, 1992b, Level I Stability Analysis (LISA)

documentation for version 2.0., General Technical Report INT-285, Department of Agriculture, Forest Service, Intermountain Research Station., p. 190.

Heim, A., and A. Gansser, 1939, Central Himalaya, Geological observations of the swiss expedition, 1936.

Hengxing, L., Z. Chenghu, C. F. Lee, W. Sijing, and W. Faquan, 2003, Rainfall-induced landslide stability analysis in response to transient pore pressure--A case study of natural terrain landslide in Hong Kong: Science in China Ser. E Technological Sciences, v. 46, p. 52-68.

Iverson, R. M., 2000, Landslide triggering by rain infiltration: Water Resour Research, v. 36, p. 1897-1910.

Joshi, V., and K. Kumar, 2006, Extreme Rainfall Events and Associated Natural Hazards in Alaknanda Valley, Indian Himalayan Region: Journal of Mountain Science, v. 3, p. 228-236.

Kim, S. K., W.P.Hong, and Y. M. Kim, 1991, Prediction of rainfall-triggered landslides in Korea., in D. H. Bell, ed., Landslides, v. 2: Rotterdam, Balkema, p. 989-994.

Lan, H. X., C. H. Zhou, L. J. Wang, H. Y. Zhang, and R. H. Li, 2004, Landslide hazard spatial analysis and prediction using GIS in the Xiaojiang watershed, Yunnan, China: Engineering Geology, v. 76, p. 109-128.

Lin, M. L., and F. S. Jeng, 2000, Characteristics of hazards induced by extremely heavy rainfall in Central Taiwan-Typhoon Herb: Engineering Geology, v. 58, p. 191-207.

LPSmanual, 2003, Leica Photogrammetry Suite, OrthoBASE & OrthoBASE Pro User’s Guide, Leica Geosystems GIS & Mapping, LLC.

Montgomery, D. R., and M. T. Dietrich, 1994, A physically based model for the topographic control of shallow landsliding.: Water Resource Research, v. 30, p. 1153-1171.

Mostyn, G. R., and J. C. Small, 1987, Methods of Stability analysis: Soil slope Instability and Stabilisation: Rotterdam, Balkema.

Otteman, R., 2001, Using GIS to model debris flow susceptibility for the Bent creek experimental forest near Asheville, North Carolina., East Carolina University, 200 p.

Pack, R. T., D. G. Tarboton, and C. N. Goodwin, 1998a, The SINMAP Approach to Terrain Stability Mapping: 8th Congress of the International Association of Engineering Geology.

Pack, R. T., D. G. Tarboton, and C. N. Goodwin, 1998b, SINMAP: A Stability Index Approach to Terrain Stability Hazard Mapping. User's manual.

Popescu, M. E., 2005, Landslide Causal Factors and Landslide Remedial Options, p. 1-21. Sharma, R. H., and H. Nakagawa, 2005, Shallow Landslide Modeling for Heavy Rainfall Events,

Annuals of Disaster Prevention Research Institute, Kyoto University.

62


__________________________________________________________________________________ Soeters, R., and C. J. v. Westen, 1996, Slope Instability recognition, analysis, and zonation, in A.

Turner, and R. Schuster, eds., Landslides investigation and mitigation. Transportation Research Board Sp report 247, National Academy press, Washington, p. 129-177.

Starker, L., 1979, The role of extreme meteorological events in the shaping of mountain relief.: Geogr Polon, v. 41, p. 13-20.

Terlien, M. T. J., 1996, Modelling spatial and temporal variations in rainfall-triggered landslides: Ph.D thesis, ITC, Netherland, Enschede, Netherlands, 253 p.

Terlien, M. T. J., 1998b, The determination of statistical and deterministic hydrological landslide triggering thresholds.: Environmental Geology, v. 35, p. 124-130.

Terlien, M. T. J., C. J. v. Westen, and T. W. J. v. Asch, 1995, Deterministic Modeling in GIS-based landslide hazard assessment, in A. C. a. F. Guzzeti, ed., Geographical Information Systems in assessing natural hazards, Kluwer academic publishers, p. 57-77.

Tewari, S., Slope Stability Effect Of Forests In Garhwal Himalayas, India. Thornbury, w. d., 1954, Principles of Geomorphology, 573 p. Tsai, T.-L., and J.-C. Yang, 2006, Modeling of rainfall-triggered shallow landslide: Environmental

Geology. Tsaparas, I., H. Rahardjo, D. G. Toll, and E. C. Leong, 2002, Controlling parameters for rainfall-

induced landslides: Computers and Geotechnics, v. 29, p. 1-27. Tucker, G. E., N. M. Gasparini, R. L. Bras, and S. T. Lancaster, 1999, Overview of the CHILD Model

Version 2.0, Final technical report, Part I-B, U.S. Army Corps of Engineers Construction Engineering Research Laboratory (USACERL).

Varnes, D. J., 1978, Slope Movements And Types And Processes, Landslides Analysis and Control. Transportation Research Board Special Report, 176, p. 11-33.

Virdi, N. S., 1986, Lithostratigraphy and structure of the central crystallines in the alaknada and dhauliganga valleys of Garhwal , U.P., in P. S. Saklani, ed., Himalayan Thrust and Associated Rocks, p. 155-166.

Westen, C. J. v., N. Rengers, M. T. J. Terlien, and R. Soeters, 1997, Prediction of the occurrence of slope instability phenomena through GIS-based hazard zonation: Geol Rundsch, v. 86, p. 404-414.

Wilkinson, P. L., M. G. Anderson, and D. M. Lloyd, 2002, An integrated hydrological model for rain-induced

Landslide prediction: Earth Surface Processes and Landforms, v. 27, p. 1285-1297. Wilson, R. C., and G. F. Wieczorek, 1995, Rainfall thresholds for the initiation

of debris flows at La Honda, California: Environ Eng Geosc, v. 1, p. 11–27. Zaitchik, B. F., and H. M. v. Es, 2003, Applying a GIS slope-stability model to site-specific landslide

prevention in Honduras: j/f, v. 58. Zaitchik, B. F., H. M. v. Es, and P. J. Sullivan, 2003, Modeling Slope Stability in Honduras: Parameter

Sensitivity and Scale of Aggregation: Soil Science Society Amer. Journal, v. 67, p. 268–278.

63


__________________________________________________________________________________

ANNEXURE 1 Adjustment Report With OrthoBASE Output image units: pixels Output ground units: meters Output z units: meters Calculated ground x, y and z coordinates: meters meters type pid ground_x ground_y ground_z gcp 4 357120.71261951 3395849.06316676 2381.49034854 gcp 5 357974.51394118 3393432.02461502 2207.67820350 gcp 6 368365.36418093 3374046.46917109 1860.17380042 gcp 7 367113.05969779 3374951.19194127 1965.94331357 gcp 9 363214.00739586 3380421.18621369 1986.19941010 gcp 10 350934.52265225 3371240.26892141 1365.58898903 gcp 11 354932.24398819 3373564.41982427 1397.75418609 gcp 12 358398.58486536 3379280.59594940 1707.75306081 tie 13 350285.57996979 3398905.20062748 4863.83393162 tie 14 350926.45012492 3398782.61177361 4778.45066594 tie 15 351052.40997955 3398755.18792163 4773.68504368 tie 16 352460.31833282 3398437.50767734 4771.68671435 tie 17 352482.20833325 3398379.98202720 4750.29793115 tie 18 349485.12935566 3398481.97970486 4880.54064634 tie 19 349538.97900651 3398411.26024590 4839.52837868 tie 20 349513.43147745 3398266.25114650 4794.44216686 tie 21 357021.35343964 3395773.28594060 2363.00909007 tie 22 356996.61853229 3395596.40299396 2342.50404601 tie 23 357164.86232035 3395219.85520211 2307.40935475 tie 24 372478.47695893 3391655.43079677 4304.92233437 tie 25 372515.98372826 3391523.11936938 4374.44468633 tie 26 357119.16150975 3395096.59376727 2294.54455751 tie 27 366420.51657827 3392335.48520781 3858.79340160 tie 28 366228.39104685 3392344.68375727 3736.27994874 tie 29 367762.64212580 3392223.29270096 4099.23818311 tie 30 369792.41019822 3391819.39325011 4210.10190637 tie 31 369856.96042739 3391743.03211403 4216.52211543 tie 32 371192.25088777 3391539.64993819 4513.54722900 tie 33 372781.22541602 3391323.55790831 4245.40498117 tie 34 372483.53656811 3391137.12759717 4435.42990768 tie 35 372474.70215017 3390797.66724436 4640.06450831 tie 36 366336.20098680 3391934.25019861 3639.61604047 tie 37 350801.57297342 3392924.56490994 4239.42953079 tie 38 350618.02209851 3392920.36444192 4313.13239283 tie 39 350820.63376724 3392763.46009822 4284.49430786 tie 40 350651.59986502 3392460.46494898 4425.48613305 tie 41 350414.57384931 3392475.31043129 4477.62921568 tie 42 351023.60905765 3392822.25353325 4173.39871670 tie 43 350987.90157388 3392627.32675509 4226.61089648 tie 44 350926.24426471 3392451.11287938 4288.93735138 tie 45 353901.06769462 3391898.70261433 4167.61390194 tie 46 354483.12081050 3391962.03430961 4123.01013069 tie 47 348272.45313299 3392635.38366936 4699.63139861 tie 48 353260.56686391 3391314.50722962 4465.08417922 tie 49 353567.10504539 3391548.86311748 4355.21546680 tie 50 371947.70015882 3387364.79662031 4266.12911328 tie 51 371676.08300243 3387339.55736480 4304.06678163 tie 52 353062.84631666 3389979.91487666 4774.54116374 tie 53 368582.90289608 3386717.64079785 3857.49514291 tie 54 371223.15083763 3386278.07683860 4493.46851854 tie 55 357789.56410832 3388368.68532364 3628.17895009 tie 56 368823.10331204 3385888.63897286 4111.76366949 tie 57 369664.71876127 3385470.70890232 4224.17400676 tie 58 357706.62547737 3388125.87883954 3769.62935254 tie 59 346985.53939658 3387641.61839236 3831.86924522

64


__________________________________________________________________________________ tie 60 346923.91269144 3387587.75591117 3834.26655290 tie 61 346926.33195841 3387405.13573409 3904.67116024 tie 62 347104.36956129 3387303.64919817 3912.11771247 tie 63 352826.78907280 3386343.58716856 4499.42075599 tie 64 353012.93369973 3386205.06190182 4578.01428049 tie 65 353606.10020448 3386178.93337619 4400.20783173 tie 66 346857.37737810 3387004.70950638 4072.75796709 tie 67 347248.09560287 3387228.81838817 3928.96562554 tie 68 347278.81957484 3387035.10856567 3963.87142309 tie 69 354900.21115138 3384955.39399137 3640.33235841 tie 70 354580.57591454 3384904.91826996 3820.93036067 tie 71 354857.35199951 3384842.43424134 3706.14217551 tie 72 370341.31939124 3380774.03461124 3874.74120980 tie 73 351651.96713918 3384815.44809799 4188.11041288 tie 74 351550.66819267 3384387.47967254 4034.17057139 tie 75 346189.12617575 3385218.85977069 4219.06234375 tie 76 346199.41140128 3384917.75381741 4330.04574982 tie 77 351721.39030247 3384170.90857421 4055.89420944 tie 78 345417.46771848 3381222.99904701 3927.89805057 tie 79 345617.46082574 3381044.30187635 3910.38971712 tie 80 345529.11131556 3381044.89373352 3940.68830231 tie 81 345565.38777781 3380940.60015280 3952.13627997 tie 82 346309.46920954 3380970.40227169 3672.71542730 tie 83 346374.99093015 3380952.83179601 3668.25389136 tie 84 363292.86473530 3374556.61381572 3620.42289841 tie 85 363410.24492131 3374494.53372860 3634.64048939 tie 86 361684.59890161 3369457.97325190 3758.58846588 tie 87 362099.43105159 3369627.77795416 3722.90643398 Control and check point residuals:meters meters type pid residual_x residual_y residual_z gcp 4 2.79181951 9.50036676 7.56374854 gcp 5 -2.83535882 -0.05798498 -0.81159650 gcp 6 0.76718093 5.81697109 -13.49569958 gcp 7 10.53679779 7.37484127 -9.31978643 gcp 9 8.52139586 4.85781369 -5.75998990 gcp 10 -2.57714775 -7.46287859 -0.83691097 gcp 11 -0.01811181 5.11022427 -9.30811391 gcp 12 6.76076536 1.06674939 -5.09553919 Image point residuals: imgid pid residual_x residual_y 1 4 -0.0845 4.1716 1 5 1.1029 -0.2863 1 6 0.4326 1.8306 1 7 -3.9306 3.4358 1 9 -3.2195 2.4222 1 10 0.4056 -3.0863 1 11 0.6804 1.6533 1 12 -2.9808 0.8439 1 13 0.0233 0.0084 1 14 -0.1341 0.0068 1 15 -0.0665 0.0069 1 16 0.1124 0.0070 1 17 0.0733 0.0066 1 18 -0.2096 0.0084 1 19 0.0231 0.0085 1 20 -0.0830 0.0077 1 21 0.0603 0.0110 1 22 0.1926 0.0116 1 23 0.1430 0.0123 1 24 -0.6402 -0.0012 1 25 -0.0361 0.0006 1 26 0.2504 0.0129

65


__________________________________________________________________________________ 1 27 0.1582 0.0013 1 28 0.1292 0.0012 1 29 -0.0495 0.0014 1 30 -0.1207 0.0012 1 31 -0.0154 0.0014 1 32 -0.0545 0.0027 1 33 0.0408 -0.0001 1 34 -0.0420 0.0013 1 35 -0.0164 0.0033 1 36 0.0464 0.0009 1 37 0.0133 0.0053 1 38 0.0017 0.0054 1 39 -0.0119 0.0049 1 40 0.0685 0.0053 1 41 0.1090 0.0058 1 42 -0.0171 0.0048 1 43 0.0183 0.0045 1 44 0.0853 0.0046 1 45 0.0377 0.0023 1 46 -0.0434 0.0018 1 47 -0.0349 0.0059 1 48 0.0905 0.0040 1 49 0.0085 0.0031 1 50 -0.0289 0.0014 1 51 -0.0490 0.0016 1 52 0.0278 0.0064 1 53 -0.0109 0.0003 1 54 -0.0788 0.0031 1 55 0.0635 -0.0002 1 56 -0.0655 0.0008 1 57 0.0025 0.0015 1 58 0.0457 -0.0004 1 59 -0.0959 -0.0002 1 60 -0.0844 -0.0002 1 61 0.0483 0.0003 1 62 -0.0914 -0.0001 1 63 0.0127 0.0030 1 64 0.0284 0.0037 1 65 0.1439 0.0024 1 66 -0.0492 0.0004 1 67 -0.0547 0.0000 1 68 -0.0785 0.0000 1 69 0.1106 -0.0001 1 70 0.0692 -0.0002 1 71 0.0960 -0.0002 1 72 -0.0134 0.0004 1 73 -0.0126 0.0008 1 74 0.0462 0.0002 1 75 -0.1059 0.0005 1 76 -0.0978 0.0012 1 77 0.1401 0.0005 1 78 -0.0936 -0.0009 1 79 -0.0966 -0.0009 1 80 -0.2047 -0.0011 1 81 -0.0956 -0.0008 1 82 -0.0417 -0.0008 1 83 -0.0357 -0.0008 1 84 -0.0382 -0.0003 1 85 0.4317 0.0008 1 86 0.0585 -0.0005 1 87 0.2105 -0.0001 Ax=-0.0900 Ay=0.1348 Mx=0.6760 My=0.7875 2 4 -0.3161 2.1688 2 5 1.1902 -0.1796 2 6 0.2857 5.0995

66


__________________________________________________________________________________ 2 7 -3.1572 5.6429 2 9 -2.7802 3.7274 2 10 0.2784 -3.1052 2 11 0.3153 3.7630 2 12 -2.3512 1.9325 2 13 -0.0187 -0.0479 2 14 0.1208 -0.0391 2 15 0.0606 -0.0394 2 16 -0.0987 -0.0411 2 17 -0.0640 -0.0390 2 18 0.1889 -0.0471 2 19 -0.0185 -0.0459 2 20 0.0757 -0.0410 2 21 -0.0523 -0.0758 2 22 -0.1700 -0.0794 2 23 -0.1258 -0.0829 2 24 0.5675 -0.0063 2 25 0.0293 -0.0143 2 26 -0.2213 -0.0854 2 27 -0.1421 -0.0010 2 28 -0.1163 -0.0003 2 29 0.0428 -0.0035 2 30 0.1061 -0.0065 2 31 0.0124 -0.0076 2 32 0.0460 -0.0217 2 33 -0.0388 -0.0093 2 34 0.0345 -0.0175 2 35 0.0107 -0.0305 2 36 -0.0426 -0.0001 2 37 -0.0109 -0.0093 2 38 -0.0006 -0.0122 2 39 0.0115 -0.0109 2 40 -0.0599 -0.0185 2 41 -0.0958 -0.0218 2 42 0.0161 -0.0066 2 43 -0.0155 -0.0089 2 44 -0.0751 -0.0121 2 45 -0.0325 -0.0071 2 46 0.0397 -0.0049 2 47 0.0328 -0.0352 2 48 -0.0791 -0.0210 2 49 -0.0063 -0.0145 2 50 0.0241 -0.0097 2 51 0.0419 -0.0111 2 52 -0.0225 -0.0419 2 53 0.0093 0.0003 2 54 0.0680 -0.0205 2 55 -0.0556 -0.0014 2 56 0.0578 -0.0039 2 57 -0.0030 -0.0081 2 58 -0.0397 -0.0007 2 59 0.0856 0.0005 2 60 0.0754 0.0004 2 61 -0.0427 -0.0017 2 62 0.0816 -0.0003 2 63 -0.0095 -0.0226 2 64 -0.0233 -0.0277 2 65 -0.1264 -0.0184 2 66 0.0443 -0.0038 2 67 0.0489 -0.0009 2 68 0.0701 -0.0012 2 69 -0.0968 -0.0020 2 70 -0.0601 -0.0013 2 71 -0.0838 -0.0014 2 72 0.0116 0.0002

67


__________________________________________________________________________________ 2 73 0.0124 -0.0076 2 74 -0.0402 -0.0038 2 75 0.0953 -0.0078 2 76 0.0883 -0.0124 2 77 -0.1237 -0.0052 2 78 0.0835 0.0002 2 79 0.0859 0.0006 2 80 0.1822 0.0013 2 81 0.0850 -0.0000 2 82 0.0365 0.0010 2 83 0.0312 0.0009 2 84 0.0328 -0.0018 2 85 -0.3856 -0.0055 2 86 -0.0568 0.0057 2 87 -0.1923 0.0067 Ax=-0.0799 Ay=0.2154 Mx=0.5591 My=1.1197 Total unit weight RMSE = 1.6259 Image accuracy for control and check points for each scene: image id 1: pid type image_x image_y residual_x residual_y 4 gcp 4194.4403 1257.5583 -0.0845 4.1716 5 gcp 4809.6645 2115.1867 1.1029 -0.2863 6 gcp 11267.7329 8643.7263 0.4326 1.8306 7 gcp 10635.5975 8403.3733 -3.9306 3.4358 9 gcp 8394.7261 6649.2189 -3.2195 2.4222 10 gcp 4016.4537 11272.6420 0.4056 -3.0863 11 gcp 5508.1261 10028.2637 0.6804 1.6533 12 gcp 6427.5073 7520.5748 -2.9808 0.8439 RMS Errors for 8 GCPs: x: 2.1430 y: 2.5367 total: 3.3207 image id 2: pid type image_x image_y residual_x residual_y 4 gcp 4002.4520 1904.5262 -0.3161 2.1688 5 gcp 4549.7036 2723.2424 1.1902 -0.1796 6 gcp 10299.7656 9210.7448 0.2857 5.0995 7 gcp 9736.6009 8993.4156 -3.1572 5.6429 9 gcp 7741.7514 7229.2958 -2.7802 3.7274 10 gcp 3845.3891 11671.6432 0.2784 -3.1052 11 gcp 5173.0795 10444.2294 0.3153 3.7630 12 gcp 5990.4798 8018.5365 -2.3512 1.9325 RMS Errors for 8 GCPs: x: 1.7678 y: 3.6058 total: 4.0158 Summary RMSE for GCPs and CHKs (number of observations in parenthesis): Control Check Ground X: 5.6170540 (8) 0.0000000 (0) Ground Y: 5.9689932 (8) 0.0000000 (0) Ground Z: 7.6920795 (8) 0.0000000 (0) Image X: 1.9643565 (16) 0.0000000 (0) Image Y: 3.1174111 (16) 0.0000000 (0)

68


__________________________________________________________________________________

ANNEXURE-2 DTM Extraction Report Date Created: 11/28/06 Time Created: 13:12:36 DTM PROJECT INFORMATION Block File Used: dem27.blk Block File Location: d:/surabhi/finaldem/ DTM Correlation Time (seconds): 1128 Points Per Second: 241 DTM Generation Time (seconds): 168 Total Processing Time (seconds): 1296 DTM Type: DEM DTM Name: d:/surabhi/finaldem/finaldem20m.img Number of Columns: 1578 Number of Rows: 1617 Cell Width: 20.0000 meters Cell Height: 20.0000 meters Upper left DEM corner coordinates: (342722.1101, 3400224.3600) Lower right DEM corner coordinates: (374262.1101, 3367904.3600) Minimum Mass Point Elevation: 1079.8192 Maximum Mass Point Elevation: 5394.2544 Mean Mass Point Elevation: 3135.3027 Projection: UTM Spheroid: WGS 84 Datum: WGS 84 Horizontal Units: meters Vertical Units: meters Strategy Parameter Settings: Image Pair Name: banda_bandf Region Description: Default Region Name of Strategy Used: Default List All of the Strategy Parameter Values Used: Search Size: 21 x 3 Allow Adaptive Change: No Correlation Size: 7 x 7 Allow Adaptive Change: No

69


__________________________________________________________________________________ Coefficient Limit: 0.8000 Allow Adaptive Change: No Topographic Type: Rolling Hills Object Type: Open Area Use Image Band: 1 DTM Filtering: low ACCURACY INFORMATION General Mass Point Quality: Excellent % (1-0.85): 68.2718 % Good % (0.85-0.70): 26.8649 % Fair % (0.70-0.5): 0.0000 % Isolated %: 0.0000 % Suspicious %: 4.8633 % Global Accuracy: Vertical Accuracy: Total # of 3D Reference Points Used: 85 Minimum, Maximum Error: -34.6592, 37.2466 Mean Error: -1.8109 Mean Absolute Error: 6.6002 Root Mean Square Error (RMSE): 9.9708 Absolute Linear Error 90 (LE90): 14.9338 NIMA Absolute Linear Error 90: +/- 12.3014 Block GCP to DTM Vertical Accuracy Total # of GCPs Used: 10 Minimum, Maximum Error: -34.6592, 9.5757 Mean Error: -8.4264 Mean Absolute Error: 12.9707 Root Mean Square Error: 16.8215 Absolute Linear Error 90: 29.3599 NIMA Absolute Linear Error 90: +/- 17.6301 Detailed Point Accuracy Information: Pt.ID X Y Z DTM Z Residual 2 363475.0655 3382045.7539 1426.1885 1419.8886 -6.2999 3 362143.7665 3388263.5960 1774.3810 1780.9633 6.5823 4 357117.9208 3395839.5628 2373.9266 2383.5023 9.5757 5 357977.3493 3393432.0826 2208.4898 2215.0530 6.5632 6 368364.5970 3374040.6522 1873.6695 1839.0103 -34.6592

70


__________________________________________________________________________________ 7 367102.5229 3374943.8171 1975.2631 1973.9850 -1.2781 8 366704.7522 3376020.9104 2110.9257 2081.5658 -29.3599 9 363205.4860 3380416.3284 1991.9594 1989.3057 -2.6537 11 354932.2621 3373559.3096 1407.0623 1392.1285 -14.9338 12 358391.8241 3379279.5292 1712.8486 1695.0477 -17.8009 Block Tie Point to DTM Vertical Accuracy Total # of Tie Points Used: 75 Minimum, Maximum Error: -30.3064, 37.2466 Mean Error: -0.9288 Mean Absolute Error: 5.7509 Root Mean Square Error: 8.6570 Absolute Linear Error 90: 11.8687 NIMA Absolute Linear Error 90: +/- 10.6510 Detailed Point Accuracy Information: Pt.ID X Y Z DTM Z Residual 13 350285.0153 3398904.9372 4865.4322 4849.6134 -15.8188 14 350925.9073 3398782.3772 4780.0074 4781.0880 1.0806 15 351051.8724 3398754.9591 4775.2316 4763.7844 -11.4473 16 352459.8434 3398437.3428 4773.1121 4777.2479 4.1357 17 352481.7369 3398379.8203 4751.7148 4763.5834 11.8687 18 349484.5731 3398481.7031 4882.1008 4872.6106 -9.4902 19 349538.4270 3398410.9886 4841.0782 4832.7090 -8.3691 20 349512.8865 3398265.9845 4795.9731 4782.9272 -13.0459 21 357021.0653 3395773.4038 2364.0372 2368.1707 4.1335 22 356996.3415 3395596.5268 2343.5047 2350.7843 7.2795 23 357164.6160 3395219.9996 2308.3408 2316.9114 8.5706 24 372479.1059 3391656.2477 4304.3119 4313.4433 9.1313 25 372516.6290 3391523.9421 4373.8013 4369.1065 -4.6948 26 357118.9223 3395096.7413 2295.4581 2304.5826 9.1245 27 366420.8668 3392336.0630 3858.6232 3847.1959 -11.4272 28 366228.7269 3392345.2550 3736.1343 3727.9748 -8.1594 29 367763.0589 3392223.9215 4098.9602 4097.2059 -1.7544 30 369792.9316 3391820.1089 4209.6503 4207.0682 -2.5820 31 369857.4903 3391743.7528 4216.0536 4215.9207 -0.1328 32 371192.8592 3391540.4243 4512.9491 4513.7816 0.8325 33 372781.8874 3391324.3980 4244.7305 4239.8432 -4.8872 34 372484.2150 3391137.9631 4434.7139 4430.4029 -4.3110 35 372475.4197 3390798.5139 4639.2660 4638.4652 -0.8008 36 366336.5661 3391934.8412 3639.4056 3640.8631 1.4575 37 350801.4162 3392924.5551 4240.0310 4237.7678 -2.2632 38 350617.8643 3392920.3477 4313.7332 4309.8435 -3.8896 39 350820.4917 3392763.4569 4285.0616 4281.5445 -3.5171 40 350651.4826 3392460.4665 4425.9920 4417.5482 -8.4438

71


__________________________________________________________________________________ 41 350414.4514 3392475.3024 4478.1427 4470.4664 -7.6763 42 351023.4625 3392822.2561 4173.9796 4170.4552 -3.5244 43 350987.7711 3392627.3352 4227.1533 4223.2866 -3.8667 44 350926.1282 3392451.1255 4289.4446 4287.5671 -1.8774 45 353901.0752 3391898.8450 4167.9012 4167.3450 -0.5562 46 354483.1389 3391962.1956 4123.2869 4125.3513 2.0645 47 348272.2676 3392635.2903 4700.2472 4695.2409 -5.0063 48 353260.6144 3391314.6471 4465.2647 4465.5162 0.2515 49 353567.1386 3391549.0057 4355.4355 4358.4448 3.0093 50 371948.6487 3387365.7531 4264.8031 4275.2650 10.4619 51 371677.0265 3387340.5050 4302.7439 4304.2112 1.4673 52 353063.0030 3389980.0969 4774.4624 4764.4795 -9.9829 53 368583.7617 3386718.5045 3856.2537 3849.6549 -6.5989 54 371224.1744 3386279.0467 4491.9594 4508.7563 16.7969 55 357789.9241 3388369.1049 3627.7357 3625.5269 -2.2087 56 368824.0509 3385889.5406 4110.3380 4108.4782 -1.8599 57 369665.7354 3385471.6550 4222.6255 4217.4536 -5.1719 58 357707.0092 3388126.3038 3769.1312 3770.1412 1.0100 59 346985.6237 3387641.6764 3831.7722 3834.2615 2.4893 60 346923.9991 3387587.8138 3834.1627 3838.6080 4.4453 61 346926.4355 3387405.2003 3904.5272 3910.9817 6.4546 62 347104.4863 3387303.7241 3911.9469 3913.1999 1.2529 63 352827.1896 3386343.9012 4498.7506 4498.2847 -0.4658 64 353013.3550 3386205.3875 4577.3023 4575.7701 -1.5322 65 353606.5326 3386179.2821 4399.4846 4397.7697 -1.7148 66 346857.5168 3387004.7861 4072.5280 4083.5470 11.0190 67 347248.2231 3387228.9012 3928.7732 3923.6382 -5.1350 68 347278.9639 3387035.1998 3963.6397 3964.1102 0.4705 69 354900.7331 3384955.8397 3639.4247 3639.8918 0.4670 70 354581.1015 3384905.3535 3820.0076 3819.7940 -0.2136 71 354857.8847 3384842.8823 3705.2092 3707.9382 2.7290 72 370342.7118 3380775.1791 3872.3902 3909.6368 37.2466 73 351652.4245 3384815.7802 4187.2637 4193.1973 5.9336 74 351551.1450 3384387.8253 4033.2716 4035.8409 2.5693 75 346189.3808 3385218.9804 4218.5347 4222.0596 3.5249 76 346199.6941 3384917.8857 4329.4523 4327.4089 -2.0435 77 351721.8899 3384171.2684 4054.9471 4055.4876 0.5405 78 345417.9690 3381223.2477 3926.7417 3923.4223 -3.3193 79 345617.9806 3381044.5646 3909.1949 3912.0601 2.8652 80 345529.6298 3381045.1531 3939.4942 3936.5108 -2.9834 81 345565.9156 3380940.8648 3950.9209 3947.7215 -3.1995 82 346310.0039 3380970.6939 3671.5030 3673.1668 1.6638 83 346375.5288 3380953.1264 3667.0359 3667.9287 0.8928 84 363294.4751 3374557.7440 3617.3339 3604.4343 -12.8997 85 363411.8656 3374495.6701 3631.5338 3601.2274 -30.3064 86 361686.5574 3369459.2360 3754.6698 3731.3593 -23.3105

72


__________________________________________________________________________________ 87 362101.3902 3369629.0488 3719.0031 3722.5890 3.5859

73


__________________________________________________________________________________

ANNEXURE-3 DTM Extraction Report Date Created: 11/28/06 Time Created: 13:57:47 DTM PROJECT INFORMATION Block File Used: dem27.blk Block File Location: d:/surabhi/finaldem/ DTM Correlation Time (seconds): 1512 Points Per Second: 270 DTM Generation Time (seconds): 383 Total Processing Time (seconds): 1895 DTM Type: DEM DTM Name: d:/surabhi/finaldem/finaldem15m.img Number of Columns: 2108 Number of Rows: 2158 Cell Width: 15.0000 meters Cell Height: 15.0000 meters Upper left DEM corner coordinates: (342722.1339, 3400233.7935) Lower right DEM corner coordinates: (374327.1339, 3367878.7935) Minimum Mass Point Elevation: 1077.8289 Maximum Mass Point Elevation: 5443.3462 Mean Mass Point Elevation: 3120.8862 Projection: UTM Spheroid: WGS 84 Datum: WGS 84 Horizontal Units: meters Vertical Units: meters Strategy Parameter Settings: Image Pair Name: banda_bandf Region Description: Default Region Name of Strategy Used: Default List All of the Strategy Parameter Values Used: Search Size: 21 x 3 Allow Adaptive Change: No Correlation Size: 7 x 7 Allow Adaptive Change: No Coefficient Limit: 0.8000

74


__________________________________________________________________________________ Allow Adaptive Change: No Topographic Type: Rolling Hills Object Type: Open Area Use Image Band: 1 DTM Filtering: low ACCURACY INFORMATION General Mass Point Quality: Excellent % (1-0.85): 69.2017 % Good % (0.85-0.70): 25.5301 % Fair % (0.70-0.5): 0.0000 % Isolated %: 0.0000 % Suspicious %: 5.2682 % Global Accuracy: Vertical Accuracy: Total # of 3D Reference Points Used: 85 Minimum, Maximum Error: -31.0015, 25.0603 Mean Error: -2.0146 Mean Absolute Error: 5.4380 Root Mean Square Error (RMSE): 8.7657 Absolute Linear Error 90 (LE90): 16.2559 NIMA Absolute Linear Error 90: +/- 11.3162 Block GCP to DTM Vertical Accuracy Total # of GCPs Used: 10 Minimum, Maximum Error: -30.9028, 9.9619 Mean Error: -6.6945 Mean Absolute Error: 10.7215 Root Mean Square Error: 13.9847 Absolute Linear Error 90: 22.5903 NIMA Absolute Linear Error 90: +/- 14.7795 Detailed Point Accuracy Information: Pt.ID X Y Z DTM Z Residual 2 363475.0655 3382045.7539 1426.1885 1421.7676 -4.4209 3 362143.7665 3388263.5960 1774.3810 1778.1704 3.7894 4 357117.9208 3395839.5628 2373.9266 2383.8885 9.9619 5 357977.3493 3393432.0826 2208.4898 2214.8731 6.3833 6 368364.5970 3374040.6522 1873.6695 1842.7667 -30.9028 7 367102.5229 3374943.8171 1975.2631 1974.1471 -1.1160

75


__________________________________________________________________________________ 8 366704.7522 3376020.9104 2110.9257 2088.3354 -22.5903 9 363205.4860 3380416.3284 1991.9594 1987.6636 -4.2958 11 354932.2621 3373559.3096 1407.0623 1392.4740 -14.5883 12 358391.8241 3379279.5292 1712.8486 1703.6827 -9.1659 Block Tie Point to DTM Vertical Accuracy Total # of Tie Points Used: 75 Minimum, Maximum Error: -31.0015, 25.0603 Mean Error: -1.3906 Mean Absolute Error: 4.7335 Root Mean Square Error: 7.8106 Absolute Linear Error 90: 15.6638 NIMA Absolute Linear Error 90: +/- 10.2263 Detailed Point Accuracy Information: Pt.ID X Y Z DTM Z Residual 13 350285.0153 3398904.9372 4865.4322 4858.7496 -6.6826 14 350925.9073 3398782.3772 4780.0074 4784.5450 4.5376 15 351051.8724 3398754.9591 4775.2316 4758.7389 -16.4927 16 352459.8434 3398437.3428 4773.1121 4777.7234 4.6113 17 352481.7369 3398379.8203 4751.7148 4765.3271 13.6123 18 349484.5731 3398481.7031 4882.1008 4884.5803 2.4795 19 349538.4270 3398410.9886 4841.0782 4824.8223 -16.2559 20 349512.8865 3398265.9845 4795.9731 4779.2031 -16.7700 21 357021.0653 3395773.4038 2364.0372 2367.3260 3.2888 22 356996.3415 3395596.5268 2343.5047 2350.4339 6.9292 23 357164.6160 3395219.9996 2308.3408 2314.7541 6.4133 24 372479.1059 3391656.2477 4304.3119 4304.1152 -0.1968 25 372516.6290 3391523.9421 4373.8013 4374.7212 0.9199 26 357118.9223 3395096.7413 2295.4581 2300.1167 4.6587 27 366420.8668 3392336.0630 3858.6232 3848.6156 -10.0076 28 366228.7269 3392345.2550 3736.1343 3724.9175 -11.2168 29 367763.0589 3392223.9215 4098.9602 4096.7869 -2.1733 30 369792.9316 3391820.1089 4209.6503 4205.8099 -3.8403 31 369857.4903 3391743.7528 4216.0536 4214.8734 -1.1802 32 371192.8592 3391540.4243 4512.9491 4515.2836 2.3345 33 372781.8874 3391324.3980 4244.7305 4243.7129 -1.0176 34 372484.2150 3391137.9631 4434.7139 4432.1963 -2.5176 35 372475.4197 3390798.5139 4639.2660 4633.3216 -5.9443 36 366336.5661 3391934.8412 3639.4056 3639.2405 -0.1650 37 350801.4162 3392924.5551 4240.0310 4238.9953 -1.0356 38 350617.8643 3392920.3477 4313.7332 4313.9329 0.1997 39 350820.4917 3392763.4569 4285.0616 4282.1802 -2.8813 40 350651.4826 3392460.4665 4425.9920 4422.4383 -3.5537 41 350414.4514 3392475.3024 4478.1427 4475.9694 -2.1733

76


__________________________________________________________________________________ 42 351023.4625 3392822.2561 4173.9796 4173.6139 -0.3657 43 350987.7711 3392627.3352 4227.1533 4227.3251 0.1719 44 350926.1282 3392451.1255 4289.4446 4290.4622 1.0176 45 353901.0752 3391898.8450 4167.9012 4169.2772 1.3760 46 354483.1389 3391962.1956 4123.2869 4126.6462 3.3594 47 348272.2676 3392635.2903 4700.2472 4700.4416 0.1943 48 353260.6144 3391314.6471 4465.2647 4465.7872 0.5225 49 353567.1386 3391549.0057 4355.4355 4358.4589 3.0234 50 371948.6487 3387365.7531 4264.8031 4268.0687 3.2656 51 371677.0265 3387340.5050 4302.7439 4303.4744 0.7305 52 353063.0030 3389980.0969 4774.4624 4767.5977 -6.8647 53 368583.7617 3386718.5045 3856.2537 3851.3905 -4.8633 54 371224.1744 3386279.0467 4491.9594 4493.5039 1.5444 55 357789.9241 3388369.1049 3627.7357 3625.4578 -2.2778 56 368824.0509 3385889.5406 4110.3380 4102.3356 -8.0024 57 369665.7354 3385471.6550 4222.6255 4203.9800 -18.6455 58 357707.0092 3388126.3038 3769.1312 3768.5501 -0.5811 59 346985.6237 3387641.6764 3831.7722 3834.2815 2.5093 60 346923.9991 3387587.8138 3834.1627 3837.2665 3.1038 61 346926.4355 3387405.2003 3904.5272 3908.7859 4.2588 62 347104.4863 3387303.7241 3911.9469 3914.2592 2.3123 63 352827.1896 3386343.9012 4498.7506 4498.1705 -0.5801 64 353013.3550 3386205.3875 4577.3023 4577.2071 -0.0952 65 353606.5326 3386179.2821 4399.4846 4396.7888 -2.6958 66 346857.5168 3387004.7861 4072.5280 4081.1066 8.5786 67 347248.2231 3387228.9012 3928.7732 3929.9311 1.1580 68 347278.9639 3387035.1998 3963.6397 3963.9937 0.3540 69 354900.7331 3384955.8397 3639.4247 3639.1738 -0.2510 70 354581.1015 3384905.3535 3820.0076 3819.2872 -0.7205 71 354857.8847 3384842.8823 3705.2092 3706.5513 1.3420 72 370342.7118 3380775.1791 3872.3902 3897.4505 25.0603 73 351652.4245 3384815.7802 4187.2637 4190.2143 2.9507 74 351551.1450 3384387.8253 4033.2716 4034.0101 0.7385 75 346189.3808 3385218.9804 4218.5347 4222.2398 3.7051 76 346199.6941 3384917.8857 4329.4523 4329.0876 -0.3647 77 351721.8899 3384171.2684 4054.9471 4055.0428 0.0957 78 345417.9690 3381223.2477 3926.7417 3926.0593 -0.6824 79 345617.9806 3381044.5646 3909.1949 3912.5623 3.3674 80 345529.6298 3381045.1531 3939.4942 3939.2306 -0.2637 81 345565.9156 3380940.8648 3950.9209 3949.1412 -1.7798 82 346310.0039 3380970.6939 3671.5030 3672.0457 0.5427 83 346375.5288 3380953.1264 3667.0359 3667.1257 0.0898 84 363294.4751 3374557.7440 3617.3339 3601.6701 -15.6638 85 363411.8656 3374495.6701 3631.5338 3600.5324 -31.0015 86 361686.5574 3369459.2360 3754.6698 3733.9462 -20.7236 87 362101.3902 3369629.0488 3719.0031 3709.8747 -9.1284

77


__________________________________________________________________________________

ANNEXURE-4 DTM Extraction Report Date Created: 11/28/06 Time Created: 16:53:50 DTM PROJECT INFORMATION Block File Used: dem27.blk Block File Location: d:/surabhi/finaldem/ DTM Correlation Time (seconds): 2458 Points Per Second: 352 DTM Generation Time (seconds): 1470 Total Processing Time (seconds): 3928 DTM Type: DEM DTM Name: d:/surabhi/finaldem/finaldem10m.img Number of Columns: 3158 Number of Rows: 3236 Cell Width: 10.0000 meters Cell Height: 10.0000 meters Upper left DEM corner coordinates: (342720.5327, 3400231.1462) Lower right DEM corner coordinates: (374290.5327, 3367881.1462) Minimum Mass Point Elevation: 1071.9148 Maximum Mass Point Elevation: 5490.1660 Mean Mass Point Elevation: 3119.0613 Projection: UTM Spheroid: WGS 84 Datum: WGS 84 Horizontal Units: meters Vertical Units: meters Strategy Parameter Settings: Image Pair Name: banda_bandf Region Description: Default Region Name of Strategy Used: Default List All of the Strategy Parameter Values Used: Search Size: 21 x 3 Allow Adaptive Change: No Correlation Size: 7 x 7 Allow Adaptive Change: No Coefficient Limit: 0.8000

78


__________________________________________________________________________________ Allow Adaptive Change: No Topographic Type: Rolling Hills Object Type: Open Area Use Image Band: 1 DTM Filtering: low ACCURACY INFORMATION General Mass Point Quality: Excellent % (1-0.85): 69.6988 % Good % (0.85-0.70): 24.3968 % Fair % (0.70-0.5): 0.0000 % Isolated %: 0.0000 % Suspicious %: 5.9045 % Global Accuracy: Vertical Accuracy: Total # of 3D Reference Points Used: 85 Minimum, Maximum Error: -29.5533, 26.0728 Mean Error: -0.9193 Mean Absolute Error: 4.7529 Root Mean Square Error (RMSE): 7.6771 Absolute Linear Error 90 (LE90): 11.6721 NIMA Absolute Linear Error 90: +/- 9.9235 Block GCP to DTM Vertical Accuracy Total # of GCPs Used: 10 Minimum, Maximum Error: -29.5533, 9.8591 Mean Error: -7.4564 Mean Absolute Error: 11.9372 Root Mean Square Error: 14.4958 Absolute Linear Error 90: 23.7151 NIMA Absolute Linear Error 90: +/- 13.5365 Detailed Point Accuracy Information: Pt.ID X Y Z DTM Z Residual 2 363475.0655 3382045.7539 1426.1885 1422.1199 -4.0686 3 362143.7665 3388263.5960 1774.3810 1780.7096 6.3286 4 357117.9208 3395839.5628 2373.9266 2383.7857 9.8591 5 357977.3493 3393432.0826 2208.4898 2214.7061 6.2163 6 368364.5970 3374040.6522 1873.6695 1844.1162 -29.5533 7 367102.5229 3374943.8171 1975.2631 1960.0208 -15.2423

79


__________________________________________________________________________________ 8 366704.7522 3376020.9104 2110.9257 2087.2106 -23.7151 9 363205.4860 3380416.3284 1991.9594 1989.0374 -2.9220 11 354932.2621 3373559.3096 1407.0623 1397.2679 -9.7944 12 358391.8241 3379279.5292 1712.8486 1701.1765 -11.6721 Block Tie Point to DTM Vertical Accuracy Total # of Tie Points Used: 75 Minimum, Maximum Error: -27.3691, 26.0728 Mean Error: -0.0476 Mean Absolute Error: 3.7950 Root Mean Square Error: 6.2272 Absolute Linear Error 90: 9.4675 NIMA Absolute Linear Error 90: +/- 8.1267 Detailed Point Accuracy Information: Pt.ID X Y Z DTM Z Residual 13 350285.0153 3398904.9372 4865.4322 4856.9586 -8.4736 14 350925.9073 3398782.3772 4780.0074 4785.5099 5.5024 15 351051.8724 3398754.9591 4775.2316 4763.8747 -11.3569 16 352459.8434 3398437.3428 4773.1121 4778.8274 5.7153 17 352481.7369 3398379.8203 4751.7148 4753.9340 2.2192 18 349484.5731 3398481.7031 4882.1008 4874.6257 -7.4751 19 349538.4270 3398410.9886 4841.0782 4839.7437 -1.3345 20 349512.8865 3398265.9845 4795.9731 4788.0571 -7.9160 21 357021.0653 3395773.4038 2364.0372 2366.9043 2.8672 22 356996.3415 3395596.5268 2343.5047 2350.2315 6.7268 23 357164.6160 3395219.9996 2308.3408 2310.8359 2.4951 24 372479.1059 3391656.2477 4304.3119 4300.9345 -3.3774 25 372516.6290 3391523.9421 4373.8013 4370.0630 -3.7383 26 357118.9223 3395096.7413 2295.4581 2298.4793 3.0212 27 366420.8668 3392336.0630 3858.6232 3846.7211 -11.9021 28 366228.7269 3392345.2550 3736.1343 3726.6667 -9.4675 29 367763.0589 3392223.9215 4098.9602 4096.5398 -2.4204 30 369792.9316 3391820.1089 4209.6503 4207.3924 -2.2578 31 369857.4903 3391743.7528 4216.0536 4212.8333 -3.2202 32 371192.8592 3391540.4243 4512.9491 4512.3539 -0.5952 33 372781.8874 3391324.3980 4244.7305 4245.0952 0.3647 34 372484.2150 3391137.9631 4434.7139 4433.0108 -1.7031 35 372475.4197 3390798.5139 4639.2660 4630.2596 -9.0063 36 366336.5661 3391934.8412 3639.4056 3638.7078 -0.6978 37 350801.4162 3392924.5551 4240.0310 4241.6491 1.6182 38 350617.8643 3392920.3477 4313.7332 4313.4109 -0.3223 39 350820.4917 3392763.4569 4285.0616 4282.0752 -2.9863 40 350651.4826 3392460.4665 4425.9920 4425.7635 -0.2285 41 350414.4514 3392475.3024 4478.1427 4476.6017 -1.5410

80


__________________________________________________________________________________ 42 351023.4625 3392822.2561 4173.9796 4177.6017 3.6221 43 350987.7711 3392627.3352 4227.1533 4226.4999 -0.6533 44 350926.1282 3392451.1255 4289.4446 4291.1663 1.7217 45 353901.0752 3391898.8450 4167.9012 4168.2562 0.3550 46 354483.1389 3391962.1956 4123.2869 4125.9182 2.6313 47 348272.2676 3392635.2903 4700.2472 4699.6100 -0.6372 48 353260.6144 3391314.6471 4465.2647 4464.8746 -0.3901 49 353567.1386 3391549.0057 4355.4355 4358.6372 3.2017 50 371948.6487 3387365.7531 4264.8031 4271.6507 6.8477 51 371677.0265 3387340.5050 4302.7439 4302.0310 -0.7129 52 353063.0030 3389980.0969 4774.4624 4771.7280 -2.7344 53 368583.7617 3386718.5045 3856.2537 3854.9200 -1.3337 54 371224.1744 3386279.0467 4491.9594 4504.6821 12.7227 55 357789.9241 3388369.1049 3627.7357 3626.5614 -1.1743 56 368824.0509 3385889.5406 4110.3380 4111.9381 1.6001 57 369665.7354 3385471.6550 4222.6255 4223.5137 0.8882 58 357707.0092 3388126.3038 3769.1312 3766.6622 -2.4690 59 346985.6237 3387641.6764 3831.7722 3835.8167 4.0444 60 346923.9991 3387587.8138 3834.1627 3836.8314 2.6687 61 346926.4355 3387405.2003 3904.5272 3907.8921 3.3650 62 347104.4863 3387303.7241 3911.9469 3914.6723 2.7253 63 352827.1896 3386343.9012 4498.7506 4497.9561 -0.7944 64 353013.3550 3386205.3875 4577.3023 4576.1089 -1.1934 65 353606.5326 3386179.2821 4399.4846 4397.6340 -1.8506 66 346857.5168 3387004.7861 4072.5280 4085.2577 12.7297 67 347248.2231 3387228.9012 3928.7732 3930.5810 1.8079 68 347278.9639 3387035.1998 3963.6397 3965.1187 1.4790 69 354900.7331 3384955.8397 3639.4247 3640.6447 1.2200 70 354581.1015 3384905.3535 3820.0076 3819.0914 -0.9163 71 354857.8847 3384842.8823 3705.2092 3703.1802 -2.0291 72 370342.7118 3380775.1791 3872.3902 3898.4630 26.0728 73 351652.4245 3384815.7802 4187.2637 4188.7217 1.4580 74 351551.1450 3384387.8253 4033.2716 4034.2005 0.9290 75 346189.3808 3385218.9804 4218.5347 4220.2466 1.7119 76 346199.6941 3384917.8857 4329.4523 4330.3766 0.9243 77 351721.8899 3384171.2684 4054.9471 4055.1763 0.2292 78 345417.9690 3381223.2477 3926.7417 3930.3249 3.5833 79 345617.9806 3381044.5646 3909.1949 3910.5003 1.3054 80 345529.6298 3381045.1531 3939.4942 3940.6021 1.1079 81 345565.9156 3380940.8648 3950.9209 3952.1644 1.2434 82 346310.0039 3380970.6939 3671.5030 3671.8821 0.3792 83 346375.5288 3380953.1264 3667.0359 3667.4602 0.4243 84 363294.4751 3374557.7440 3617.3339 3607.5122 -9.8218 85 363411.8656 3374495.6701 3631.5338 3604.1647 -27.3691 86 361686.5574 3369459.2360 3754.6698 3755.5336 0.8638 87 362101.3902 3369629.0488 3719.0031 3725.1369 6.1338

81


__________________________________________________________________________________

82

establishing precipitation thresholds for landslide …characterisation using gis-based modelling”...

Documents