Mass Movements along the Highway with a focus on Seismically Induced Mass Movements

From the Faculty of Georesources and Materials Engineering of the RWTH Aachen University

Submitted by

Sajid Ali M.Sc.

from Okara,

in respect of the academic degree of

Doctor of Natural Sciences

approved thesis

Advisors: Univ.-Prof. Dr.rer. nat. Klaus Reicherter Univ.-Prof. Dr.rer. nat. Florian Amann

Date of the oral examination: 12.05.2020

This thesis is available in electronic format on the university library’s website

Eidesstattliche Erklärung

Hiermit versichere ich eidesstattlich, dass ich die Dissertation selbstständig verfasst und alle in Anspruch genommenen Hilfen in der Dissertation angegeben habe.

Unterschrift Ort, Datum

This work is dedicated to my loving father “Sikander Ali (1949-2007)”, who is a source of motivation and inspiration for me and many others!

Acknowledgments

Firstly, I would like to thank my supervisor, Prof. Dr. Klaus Reicherter for continuous support, inspiration, motivation, advices and answering a lot of my questions, specially “the last question”.

I am obliged to Dr. Muhammad Basharat (University of Azad Jammu and Kashmir, Muzaffarabad, Pakistan) for his guidance during field campaign and suggestions. Scientific and technical support from Prof. Dr. Florian Amann and Dr. Anja Dufresne (Chair of Engineering Geology and Hydrogeology) was essential for successful completion of the work. I am also indebted to:

 Higher Education Commission (HEC) of Pakistan and Deutscher Akademischer Austauschdienst (DAAD) for their support.  Pakistan Meteorological Department for provision of weather data.  Lt. Gen. Muhammad Afzal and Maj. Gen. Inam Haider Malik (Frontier Works Organization, Rawalpindi, Pakistan) for provision of accommodation and transportation during field work along the Highway.  Maj. Hassnain Farooq and Maj. Rameez Ali for organizational support during field work.  Mr. Peter Biermanns for valuable comments and suggestions for writing process.  To anonymous referees and Editor Paolo Tarolli (University of Padova, Legnaro, Italy) for valuable comments.  Mr. Rashid Haider for giving access to relevant publication in Geological Survey of Pakistan’s library and assistance during field work.  My NUGGED family for refreshing coffee and lunch breaks, Ms. Evelyn Bützler and Dr. Tabea Schröder for answering many questions, Mr. Christopher Weismüller for drone trainings, Mr. Wahid Abbas and Mr. Aram Fathian Baneh for comments and Ms. Jorien van der Wal for fruitful seminars.

I am grateful to my family specially my mother “Ameeran Bibi” for her continuous care, support and prayers, my son “Hussain Ali” for making my life colorful, my wife Sumaira for scarifying her weekends with unlimited support (24/7) and patience during busy routines, my brother “Abid Ali” for doing everything for me far from Pakistan.

Abstract

Landslide and its severe impact on infrastructure is talk of the town now a days. It is a complex process, which is quite difficult to understand and predict. Multiple remote sensing and site specific approaches have been employed to determine associated hazard and risk. “Regional to site specific approach” was applied in this study. Initially, regional landslide susceptibility map was prepared by using geographic information system. Potential hazardous sites identified during this stage were further investigated in detail.

This study is concerned about the (KKH), located in north Pakistan. It is an important part of the China Pakistan Economic Corridor (CPEC), connecting Arabian Sea with southwest China. It is also considered a lifeline of the -Baltistan, being only all-weather physical connection with rest of country. It passes through plateau, high plains and deeply incised gorges (~6500 m). From Thakot onwards, it runs along four rivers: Indus, Gilgit, Hunza and Khunjerab. It is situated at the junction of two plate boundaries: India and Eurasian plate. Ongoing collision between these two plates has produced different tectonic environments and seismic zones. Region has witnessed major historical and recent earthquakes e.g. Pattan 1974 (M=6.2), Astore 2002 (M=6.3) and Kashmir Earthquake (M=7.6), which completely damaged some segments of the Highway.

Landslide inventory map (LIM) was basic requirement of the entire process to prepare hazard and risk maps of the Highway. Previously published reports, articles, books and road clearance logs were consulted to prepare multi-temporal inventory. Furthermore, multi-temporal satellite imagery was examined to locate active and potential slope failures, which were further classified into size and failure mode categories. Extent and type of slope failures was then validated during multiple field campaigns. Subsequently, geological map was prepared by digitizing existing published maps. Additionally, influence of lithological units (formation/group) and individual lithologies on spatial distribution and types of landslides was analysed. Highly fragmented ultramafics, deeply weathered plutons and mechanically shattered schists and slates have high landslides density. Furthermore, degree of fragmentation and attitude of joint sets defined mode of slope failure.

The Highway passes through two climatic zones: Monsoon and semi-arid to arid zone. Former is characterized by torrential rainfalls whereas latter has low mean annual rainfall (>250 mm). Strong correlation between rainfall intensity and mass movements has been found. Besides, variations in temperature coupled with precipitation initiated rockfalls. Moreover, active faults (MBT, MMT, KSF, RF, MKT, and KF) and shear zone (KJS) characterize the KKH. Rock mass is highly jointed and sheared close to these structure. Landslide clusters were found close to these structure indicating their strong control.

Preparation of landslide susceptibility map (LSM) involved analytical hierarchy process (AHP) based semi-quantitative technique. Ten parameters (lithology, seismicity, rainfall intensity, faults, elevation, slope angle, aspect, curvature, land cover and hydrology) were used to assess susceptibility. Spatial analysis was performed to estimate effect of these parameters over landslide distribution. Each parameter was rated by using expert based AHP. Weighted overlay technique was employed to produce final map, which was further classified into four levels: low, intermediate, high and very high susceptibility. High and very high susceptibility areas were found close to active faults. Three case studies (Jijal sub-section, Raikot Bridge sub-section and Attabad sub-section) were discussed to explain the final map. To end with, Landslide density analysis (LDA) and receiver operative curve (ROC) determined accuracy of the map as 72%, which was considered reasonable for planning and mitigation.

High and very high susceptibility characterize long segment of the Besham-Chilas section, therefore was preferred for risk assessment. LIM, slope angle distribution (SAD) and satellite imagery analysis were utilized to locate rockfall and debris flow sources. Which were further exploited to assess runout by using open source and customizable software “FLOW-R”. Sites with maximum probability to reach and damage the Highway, were empirically rated for risk assessment. Modified Pierson’s Rockfall Hazard Rating System (MRHRS) rated potential rockfalls whereas semi-quantitative criteria was employed to rate debris flows. Lastly, final map was then classified into four risk levels: low, intermediate, high and very high. Immediate application of countermeasures for segments at high and very high risk were proposed.

This study encompasses important dimensions of mass movements along the Highway. But still, some other aspects like numerical and physical modeling needs to be done for understanding and evaluation of the entire process.

Kurzfassung

Bergrutsche und ihre schwerwiegenden Folgen, vor allem in Bezug auf die Zerstörung verschiedenster Infrastruktur, sind heutzutage von breitem öffentlichen Interesse. Hinter diesen Ereignissen stecken komplexe Prozesse, die schwer vorherzusagen sind. Diese Arbeit zeigt Ansätze auf, um regional und standortspezifisch mit verschiedenen Methoden, u.a. der Fernerkundung, die potentielle von Bergrutschen ausgehende Gefahr zu bewerten. Eine GIS- basierte Analyse der regionalen Anfälligkeit für Bergrutsche dient dabei als Grundlage für nachfolgende detaillierte Vor-Ort-Untersuchungen.

Gegenstand der Studie ist der Karakoram-Highway (KKH) im nördlichen Pakistan, der einen bedeutenden Bestandteil des Wirtschaftskorridors zwischen China und Pakistan („China Pakistan Economic Corridor“, CPEC) darstellt. Dieser verbindet das Arabische Meer mit dem Südwesten Chinas. Neben der wirtschaftlichen Bedeutung stellt der KKH auch die „Lebensader“ der pakistanischen Region Gilgit-Baltistan dar, deren einzige Verbindung zum Rest des Landes er ist. Der KKH führt durch (Hoch-) Ebenen und tiefe (~6500 m) Schluchten. Von der Stadt Thakot aus folgt der Highway in nördlicher Richtung vier verschiedenen Flüssen (Indus, Gilgit, Hunza und Khunjerab) und verläuft entlang der Indisch-Eurasischen Plattengrenze. Die andauernde Kollision dieser beiden Kontinentalplatten bedingt im Untersuchungsgebiet verschiedene seismische Zonen mit unterschiedlichen tektonischen Konfigurationen. Die Region war Schauplatz bedeutender historischer- und rezenter Erdbeben, wie z.B. Pattan 1974 (M=6.2), Astore 2002 (M=6.3) und Kashmir 2005 (M=7.6), die Teile des KKH jeweils komplett zerstörten.

Eine Kartierung des gesamten Bergrutsch-Bestandes der Region („Landslide inventory map“, LIM) stellt die Grundlage für die darauffolgende Erstellung von Gefahrenkarten für das Einzugsgebiet des KKH dar. Zum einen wurden dazu wurden über einen langen Zeitraum veröffentlichte Berichte, Artikel, Bücher und Protokolle über Straßensperrungen herangezogen. Zum anderen wurden Satellitenbilder verschiedenen Datums betrachtet, um aktive, bereits vollzogene oder potentielle Hangbewegungen zu lokalisieren. Die beobachteten Phänomene wurden in diesem Zuge nach Größe und Art klassifiziert. Die durch Fernerkundung erhobenen Daten und Rückschlüsse wurden im weiteren Verlauf vor Ort überprüft. Auf Grundlage eines Verschnitts bestehender geologischer Karten wurde der Einfluss verschiedener Lithologien auf die Verteilung und Art von Bergrutschen analysiert. Daraus ergibt sich, das vor allem stark zerklüftete Ultramafite, stark verwitterte Plutonite und mechanisch stark beanspruchte Schiefer anfällig für Bergrutsche sind. Dabei sind weiterhin der Grad der Zerklüftung und die Kluftorientierung entscheidend für die Art und Ausprägung des Hangrutsches.

Der KKH führt durch zwei Klimazonen: Die Monsunzone und die semiaride bis aride Zone. Die erstere ist von saisonale Starkregen geprägt, während letztere einen geringen jährlichen Niederschlag (>250 mm) aufweist. Die Niederschlagsintensität in Kombination mit den gegebenen höheren Temperaturschwankungen korreliert dabei stark mit Massenbewegungen. Der KKH kreuzt verschiedene aktive Störungen (MBT, MMT, KSF, RF, MKT, KF -> define) und Scherzonen (KJS). Da diese wiederum eine hohe Zerklüftung der durchzogenen Gesteine bedingen, zeigt eine Häufung von Bergrutschen in der unmittelbaren Umgebung den enormen Einfluss dieser Strukturen.

Die Erstellung einer Bergrutsch-Anfälligkeitskarte (Landslide susceptibility map, LSM) basiert auf der Semi-quantitativen Methode des „Analytical Hierarchy Process“ (AHP). Dabei wurden zehn Parameter herangezogen um die Bergsturz-Anfälligkeit zu beurteilen: Lithologie, Seismizität, Niederschlagsintensität, Störungszonen, Höhe über NN, Hangneigung, Expositionsrichtung, Kurvatur, Landnutzung und Hydrologie. Eine räumliche Analyse zeigt den Effekt der Parameter auf die Verteilung der Bergstürze. Die Auswirkung jedes Parameters wurde durch die Anwendung eines fachgemäßen AHP bewertet, um Anschließend mit der „Weighted Overlay-Methode“ eine finale Karte zu erstellen. Diese wiederum klassifiziert das Untersuchungsgebiet in vier Klassen der Bergsturzanfälligkeit: Gering, mittel, hoch und sehr hoch. Hohe und sehr hohe Anfälligkeit herrscht in der Nähe aktiver Störungen. Anhand von drei Fallbeispielen (Jijal Sub-Sektion, Raikot Bridge Sub-Sektion and Attabad Sub-Sektion) wird die Erstellung der finalen Karte erläutert. Anhand einer „Landslide Density Analysis (LDA) und “Receiver Operative Curve” (ROC) wird die Zuverlässigkeit der Karte auf 72% geschätzt, was zum Zweck einer adäquaten Risikobeurteilung und Schadensmilderung als ausreichend eingestuft wird.

Hohe und sehr hohe Gefahr von Bergstürzen besteht für das lange Segment Besham-Chilas, das aus diesem Grund für die Gefahrenbeurteilung herangezogen wurde. LIM, die Verteilung der Hangneigung (SAD) und Satellitenbildmaterial wurden zur Lokalisierung von Quellen von Bergstürzen und Hangrutschen genutzt. Die Lokalitäten wurden weiter genutzt, um mittels der Software FLOW-R den Auslaufbereich der Massenbewegungen zu analysieren. Die Bereiche, die die höchste Wahrscheinlichkeit aufweisen, dass der KKH von Massenbewegungen erreicht und zerstört wird, wurden einer empirischen Risikobewertung unterzogen. Eine Modifikation des “Pierson’s Rockfall Hazard Rating System” (MRHRS) wurde herangezogen, um potentielle Felsstürze zu bewerten; Semi-quantitative Kriterien dienten zur Bewertung von Murgängen. Die

finale Karte wurde in die o.g. Gefahrenstufen eingeteilt und direkte Gegenmaßnahmen für stark oder sehr stark gefährdete Bereiche angeregt.

Diese Studie trägt wichtige Ansätze zur Bewertung von Massenbewegungen am KKH bei. Trotzdem sind zu einer weiterführenden Bewertung und einem umfassenderen Verständnis der Prozesse weitere Ansätze wie numerische und physikalische Modellierungen notwendig.

Table of Contents

Acknowledgments ...... f

Abstract ...... h

Kurzfassung ...... j

List of Figures ...... V

List of Tables ...... IX

List of Equations ...... XI

1 Introduction ...... 1

The Karakoram Highway, Pakistan ...... 2

Aims and Objectives ...... 4

Regional to Site Specific Approach ...... 4

Thesis Outline ...... 6

2 Distribution and Lithological Control of Landslides ...... 7

Landslide Inventory Map ...... 7

Lithological Control of Landslides ...... 10

2.2.1 The Indian Plate ...... 11

2.2.2 The Kohistan Island Arc ...... 16

2.2.3 The Eurasian Plate ...... 20

3 Structural and climatic control of Mass Movements ...... 27

Abstract ...... 27

Introduction ...... 28

Previous Work ...... 29

Aims and Objectives ...... 30

Regional Tectonic Setting ...... 30

Climate of Study Area ...... 30

Methodology ...... 31

Fault Control of Mass Movements along the KKH ...... 35

Climatic Control of Mass Movements ...... 35

Conclusion ...... 36

4 Meteorological Variation and Temporal Distribution of Rockfalls ...... 37

Introduction ...... 37

Climatic Variations in Studied Sections ...... 40

Methodology ...... 40

Research Outcomes ...... 41

5 Landslide Susceptibility Mapping by using GIS ...... 45

Abstract ...... 45

Introduction ...... 46

General situation of the study area ...... 49

Geology along the KKH ...... 53

Seismology ...... 54

Methodology ...... 56

5.6.1 Literature Review ...... 56

5.6.2 Remote Sensing ...... 59

5.6.3 Analytical hierarchy process (AHP) ...... 60

5.6.4 Weighted Overlay Method...... 63

Results ...... 64

5.7.1 Landslides along the KKH ...... 64

5.7.2 Causative Factors and Spatial Distribution analysis ...... 64

5.7.3 Landslide Susceptibility Map ...... 68

Discussion ...... 74

5.8.1 Case study ...... 76 Conclusions ...... 82

6 Empirical Assessment of Rockfall and Debris Flow Risk along the Besham-Chilas Section of the KKH...... 83

Abstract ...... 83

Introduction ...... 83

General Situation of the Study Area ...... 88

6.3.1 Location and weather conditions ...... 88

6.3.2 Geology, geomorphology, tectonics and seismology...... 88

Identification of potential hazardous sites (Regional) ...... 89

6.4.1 Landslide Inventory ...... 89

6.4.2 Slope Angle Distribution (SAD) ...... 90

6.4.3 Material Source of Debris Flow ...... 91

6.4.4 Runout Assessment ...... 91

Detailed Mapping and Risk Assessment (Site Specific) ...... 92

6.5.1 Rockfall Hazard Rating System (RHRS) ...... 92

6.5.2 Debris-Flow Risk Assessment ...... 95

Results and Discussion ...... 95

6.6.1 Analysis of Landslide Inventory ...... 95

Risk Analysis of Rockfalls ...... 98

6.7.1 Jijal Rockfall ...... 98

6.7.2 ChoChang Rockfall ...... 98

6.7.3 Lotar Rockfall ...... 99

6.7.4 Yadgar Rockfall ...... 99

Risk Analysis of Debris Flows ...... 101

6.8.1 Serai Debris Flow ...... 101

6.8.2 Harbon Debris Flow ...... 101

Final Risk Map ...... 103

Discussion and Conclusion ...... 106

7 Conclusions and Outlook ...... 109

Conclusions ...... 109

Outlook ...... 111

8 References ...... 115

List of Figures

Figure 1.1: Overview of tectonics & precipitation of study area (After Hodges 2000)...... 3

Figure 1.2: Schematic illustration of “Regional to site specific approach”...... 5

Figure 2.1: Analysis of multi-temporal satellite imagery to find and map landslides along KKH: . 9

Figure 2.2: Inter- and intra-folial folding in gneisses of Besham Group, near MMT...... 12

Figure 2.3: Correlation showing lithological control of landslides along the KKH ...... 13

Figure 2.4: Lithological map of the KKH (Hassan Abdal – Chilas Section)...... 14

Figure 2.5: Nanga Parbat augen gneisses, exposed near Raikot Bridge...... 16

Figure 2.6: Highly fragmented pyroxenites and serpentinites of Jijal Complex...... 17

Figure 2.7: Massive and sheared amphibolites...... 18

Figure 2.8: Gabbronorite association of Chilas Complex...... 19

Figure 2.9: Gneisses of Besham Group have foliation with dip slope setting...... 21

Figure 2.10: Lithological map of the KKH (Chilas-Khunjerab Pass Section)...... 23

Figure 2.11: Slates, phyllites and quartzites of Misgar Slates...... 24

Figure 2.12: Alluvial deposits formed due to deviation in river channel during landslide damming...... 26

Figure 3.1: Overview of tectonics & precipitation of Study area ...... 29

Figure 3.2: Geology along the KKH ...... 32

Figure 3.3: Mass movements along the KKH ...... 33

Figure 3.4: Results: (a) Correlation between mass movements and monthly rainfall intensity (Hassan Abdal- Gilgit Section) (b) Correlation between distance from fault and mass movement events (c) Influence of temperature and rainfall intensity on mass movements (Gilgit-Khunejrab Section)...... 34

Figure 4.1: Location of the Karakoram Highway (KKH) (B) Overview of topography, tectonics and weather conditions along the Highway...... 38

Figure 4.2: Distribution of mean monthly rainfall through a year, along five weather stations. Pattan, Gilgit and Sost represent three sections, mentioned in Fig. 4.1...... 39

Figure 4.3: Variations in monthly freeze-thaw cycles in a year. Three weather stations represent section, displayed in Fig. 4.1...... 39

Figure 4.4: Correlation between precipitation, temperature (min. and max.) and rockfalls for four years period (August 1996 – July 2000)...... 42

Figure 4.5: Site specific correlation ...... 43

Figure 5.1: Overview of tectonics & precipitation in the study area ...... 50

Figure 5.2: Glaciers along the KKH ...... 51

Figure 5.3: (a) Overview of precipitation (mean annual rainfall) and landslides frequency along the KKH (after Khan et al., 2000, Pakistan Meterological Department 1982, 83, 96, 97, 98, 99, 2000, 14, 15, 16, Frontier Works Organization archives) (b) Correlation between landslide events and precipitation (Casagli et al., 2017)...... 52

Figure 5.4: (a) Regional location of Himalaya (b) Overview of Himalayan geology (c) Geology along the KKH ...... 55

Figure 5.5: Flow Chart showing multiple steps involved in the preparation of the susceptibility map ...... 56

Figure 5.6: Temporal distribution of landslides along the Highway: a and b represent two problematic section shown in Fig. 5.11...... 58

Figure 5.7: Types of Mass Movements along the Highway (a) Thakot to Raikot Bridge (b) Raikot Bridge to Khunjerab Pass (Locations shown in Fig. 5.6)...... 59

Figure 5.8: Frequency distribution histograms of controlling parameters ...... 66

Figure 5.9: Landslide Susceptibility Map (Abbottabad-Chilas): (a) Jijal Section (area in box “X” is shown in Fig. 5.13) (b) Dassu Section (c) Sazin Section...... 70

Figure 5.10: Landslide Susceptibility Map (Chilas-Khunjerab Pass): (a) Raikot Bridge Section. 71

Figure 5.11: Example of landslide events ...... 72

Figure 5.12: ROC based accuracy assessment of the landslide susceptibility map...... 74

Figure 5.13: Very high landslide susceptibility along the KKH near Jijal...... 77

Figure 5.14: Very high landslide susceptibility near Raikot Bridge ...... 79 Figure 5.15: High landslide susceptibility along the Highway in Hunza Valley ...... 81

Figure 6.1: Overview of study area (a) Location of the study area (b) Overview of weather conditions and tectonics (c) Geology of the study area ...... 85

Figure 6.2: Impact of rockfalls and debris flows on traffic and travelers ...... 87

Figure 6.3: Identification of potential rockfalls and debris flows...... 90

Figure 6.4: (a) An overview of Yadgar Rockfall ...... 100

Figure 6.5: Overview of Serai debris flow ...... 102

Figure 6.6: Risk map of the studied section, (a) Besham-Pattan, (b) Pattan-Dassu, (c) Dassu- Kandia Valley...... 104

Figure 6.7: Risk map of the studied section, (a) Kandia Valley-Harbon Nala, (b) Harbon Nala- Chilas...... 105

Figure 7.1: Overview of countermeasures against landslides ...... 112

Figure 7.2: Condition of the Highway, damaged by continuous erosion by ...... 113

List of Tables

Table 2.1: Types of landslides along the KKH: locations of sections in given in Fig. 3...... 10

Table 2.2: Lithological description and LAF (Landslide area frequency) of formations...... 15

Table 5.1: Previous work in some parts of the northern Pakistan...... 48

Table 5.2: Fundamental scale for pair-wise comparisons (Saaty, 1987)...... 61

Table 5.3: Random Consistency Index (RI) ...... 61

Table 5.4: Pairwise matrix and weights of factor sub-classes...... 62

Table 5.5: Pairwise matrix and weights of all controlling parameters...... 63

Table 5.6: Types of landslides along the KKH...... 64

Table 5.7: Area of Susceptibility Classes ...... 69

Table 5.8: Areas of susceptibility level of map and observed Landslides...... 73

Table 6.1: Rockfall Hazard Rating System (modified after Pierson 1993) for risk assessment. .94

Table 6.2: Structural data and kinematic analysis of rockfalls...... 96

Table 6.3: Rating criteria for debris flows risk assessment ...... 97

Table 6.4: Percentage of different classes in final risk map ...... 103

List of Equations

Equ. (2.1) 퐿퐷 = (퐿푛퐿푡 ∗ 100)/(퐴푓퐴푡 ∗ 100) ...... 11

Equ. (5.1) 퐶푅 = 퐶퐼/푅퐼 ...... 60

Equ. (5.2) 퐶퐼 = (휆푚푎푥 − 푛)푛 − 1 ...... 61

Equ. (5.3) 푆 = 푊푖 푆푖푗푊푖 ...... 63

Equ. (6.1) 푃퐷푆퐷 = 퐴푆퐷퐷푆퐷 ∗ 100 ...... 93

Equ. (6.2) 퐻푛 = (퐻 − 퐻푚푖푛)/(퐻푚푎푥 − 퐻푚푖푛) ...... 98

Introduction ______

1 Introduction

Landslide is a gravity driven mass movement, influenced by different geodynamic processes. Gradual degradation of rock mass primarily by weathering along with tectonic stresses, seismicity and climatic variations results in downslope mass movement (Vallejo and Ferrer, 2011). Landslide’s classification involves different criteria based on constituent geological material and structure, their relation slope geometry, size, age, location, causes and type of slope failure (Varnes, 1978a). Commonly used classification criteria are comprised primarily of type of slope movement and secondarily of slope material. Landslides poses serious threat to infrastructure and human activities. Understanding of basic processes and controlling parameters can help to minimize or avoid associated hazard and risk, based on both direct and indirect impacts of landslides. Former is damage to infrastructure, maintenance and mitigation costs whereas later is associated with traffic disruption, low economic activity and most importantly time loss. Population expansion has pushed to construct settlements and transportation lines in hazardous areas, increasing mortality rate and economic losses. Amount of these life losses is more in developing countries as compare to developed ones, having severe impact on business (Schuster and Fleming, 1986). Reason is unwillingness of poor developing and under-developed economies to invest in mitigation of slope hazards. These countries lack legislation regarding formulation and implementation of building code to shield population in general.

Hazard is probability of a disaster to happen whereas vulnerability is its capacity to damage exposed elements (Varnes, 1978b). Product of both hazard and vulnerability is termed as risk (Cui et al. 2013). Hazard evaluation requires comprehensive awareness about landslide process and controlling parameters and vulnerability analysis needs thorough facts about elements at risk like population, infrastructure etc. Estimation of hazard and risk has been a challenge for researchers. However, availability of data regarding landslide-controlling parameters from remote sensing sources has made it easy and simple. Landslide process is a combination of conditioning factors (geology, lithology, structures etc.) and triggering factors (slope geometry, pore pressure, climate, and state of stress). Former is also known as inherent conditions, which remains constant. Whereas triggering factors vary and slight change in their status can disturb equilibrium. A good hazard map should forecast spatial and temporal probability of incidence (Soeters and van Westen, 1996). Previously, deterministic, statistic and heuristic approaches were used for hazard and risk zonation at national, regional, medium and large scale. Basic uniformitarianism

1

Introduction ______principle “past and present are key to future” is basis of heuristic and statistical methods (black box models). Where investigation of previous events explains control of different parameters on their occurrence and helps to predict future ones. This correlation is used to predict behavior of the landslide with respect to change in existing scenario. Heuristic technique comprised of two steps: direct mapping involving acquisition of data directly from the field and expert based assessment of qualitative maps (Francipane et al., 2014). Deterministic approach, also known as white box models, depends upon equations based physical simulations and modeling.

The Karakoram Highway, Pakistan

The Karakoram Highway (KKH) is an important route connecting North Pakistan with Southwest China. It also connects Gilgit Baltistan in north to rest of the country (Fig. 1.1). Before its construction, route via Kaghan and Babusar Pass was only physical connection, which remains open for only some months during summer. In mid-sixties, Government of Pakistan started thinking about all-weather road connection between Gilgit Baltistan, a strategic area, with rest of its territory. Initially, Corps of Engineers expanded existing mule track in the shape of a road until Gilgit under “Indus Valley Road Project” (Khan et al., 2003), further widened up to eight meters and extended until Khunjerab Pass, at Chinese border with cooperation of People’s Republic of China. It passes through world’s highest mountain ranges (Karakoram and Himalaya) and marvelous landscape. Therefore, it has been termed as “eighth wonder of the world”. Since its completion in 1978, traffic disruption, life and economic losses due to slope failures is a continuous phenomenon. In 2010 and 2016, torrential rainfalls (Monsoon and Westerlies) and seismic events (Pattan Earthquake 1974, M=6.2 and Kashmir Earthquake 2005, M=7.6) completely eroded and buried the Highway at many places. Going through its some of its sections (as an example: near Raikot Bridge) is a complete nightmare. Lack of interest of Highway authorities in tackling landslides is worsening the situation. Two Pakistan army’s units have been deployed at different locations; maintaining traffic flow along the KKH. They are only responsible for clearing landslide deposit, immediately after every landslide event. Geological, tectonic, seismic, geomorphological and climatic variations make this area unique landslide study area.

The Highway passes through one of the steep, rugged and tectonically active areas of the world. It is located at the junction of two plates (Indian and Eurasian Plate) with convergence rate of ~4- 5 cm per year (Jade, 2004). Ongoing collision between these two plates resulted into a large of regional and local faults. These faults are responsible for seismicity with drastic impact on infrastructure. Seismicity and different tectonic regimes has left highly jointed and fragmented

2

Introduction ______rock mass in close vicinity of active faults. A wide range of lithologies of different ages (Quaternary to Pre-Cambrian) characterizes the Highway. It runs along Indus, Gilgit and Hunza River, which control topography of the area. Deep incision by these rivers has yielded deep gorges and steep gradient. Further, the Highway achieves elevation of more than 4600 meters at Khunjerab Pass. Topography has strong influence over climate and weather conditions. Area can easily be divided into two precipitation zones: Monsoon zone with mean annual rainfall more than 1000 mm and non-Monsoon region with semi-arid to arid conditions. Temperature exhibit wide range from warm Hassan Abdal to extremely cold Khujerab Pass, where rock mass undergoes through long freezing (>6 months) period. All discussed conditions evolved parts of the Highway into highly hazardous site with a diversity of slope failures.

Figure 1.1: Overview of tectonics & precipitation of Study area (After Hodges 2000). KKH- Karakoram Highway, MBT-Main Boundary Thrust, MMT-Main Mantle Thrust, MKT-Main Karakoram Thrust.

3

Introduction ______Aims and Objectives

 Identification and mapping of mass movements along the KKH to further investigate their geological, structural and climatic control.  Preparation of landslide susceptibility map by using remote sensing and regional approaches.  Preparation of hazard and risk maps of hazardous sections already identified during regional studies.  Recommendation of appropriate, cost-effective, remedial measures to minimize the effect of hazard associated with slope movements.  The ultimate objective of the study is the safety of the travelers and the hazard-free communication between two friendly nations through road.

Regional to Site Specific Approach

Regional to site specific approach has been employed to investigate mass movements along the KKH (Fig. 1.2). Initially, landslide susceptibly map of the entire Highway was prepared by using GIS based semi-quantitative modeling. Identification of problematic sections of the Highway was prime objective during this stage. Afterwards, during second stage, these sections were then analysed in terms of their capacity to damage the Highway, vehicles and travelers. Numerical model based open source software has assessed runout extent and probability. Hotspots identified have been surveyed in detail. Empirical risk assessment of each site has been performed, followed by preparation of risk map of the Highway. In the end, countermeasures against landslides with very high and high-risk score have been recommended.

4

Figure 1.2: Schematic illustration of “Regional to site specific approach”. FWO-Frontier Works Organization, PMD-Pakistan Meteorological Department, GSP-Geological Survey of Pakistan, DEM-Digital Elevation Model.

5

Introduction ______Thesis Outline

This thesis comprises of chapters based on published articles and expected publications, leading to inevitable repetition. Followings gives an overview of chapters and SA’s contribution.

Chapter 2: Distribution and lithological control of Landslides along the Karakoram Highway, Pakistan.

Self-assessment: SA contributed >90%.

Chapter 3: It is based on chapter published in book.

(Ali et al., 2017): Structural and Climatic Control of Mass Movements along the Karakoram Highway.

Self-assessment: SA contributed >90%.

Chapter 4: It is based on an extended abstract published in a conference proceeding.

(Ali et al., 2019): Meteorological variation and rockfall occurrence along the Karakoram Highway.

Self-assessment: SA contributed >85%.

Chapter 5: It is based on one peer review published article.

(Ali et al., 2018a): Landslide susceptibility mapping by using a geographic information system (GIS) along the China–Pakistan Economic Corridor (Karakoram Highway), Pakistan.

Self-assessment: SA contributed >90%.

Chapter 6: It is based on paper under review.

Empirical Assessment of Rockfall and Debris Flow Risk along the Karakoram Highway, Pakistan.

Self-assessment: SA contributed >90%.

Chapter 7: It covers resume of investigated sections of the Highway, recommendation of effective countermeasures and research areas, needs to be further investigated.

Self-assessment: SA contributed 100%.

6

Distribution and Lithological Control of Landslide ______

2 Distribution and Lithological Control of Landslides

Mass movement is gravity-driven geometric adjustment of landmass in response to variations in mechanical and hydrological properties. These properties predominantly depends upon lithology, further controlling spatial distribution of landslides. Generally, landslide inventory maps are prepared to find spatial relevance between factors like lithology, responsible for past occurrences. In this chapter, an effort is made to look into spatial coincidence between landslides and lithology.

The Highway traverses through Indian and Eurasian Plates and Kohistan Island Arc (KIA), resulting into a wide range of lithology (volcanic, plutonic, metamorphic and sedimentary). Pre- Cambrian to Cambrian basement and Paleozoic, Mesozoic to Tertiary rocks constitute the Indian plate, further classified into orographic units of Nanga Parbat Harmosh Massif, Himalayan crystalline schuppen zone and Himalayan fold and thrust belt. The Main Boundary Thrust (MBT) delineates crystalline zone as a fold and thrust belt. KIA is thick sequence (40 km) of metamorphosed calcalkaline plutonic, volcanic and volcano-sedimentary rocks (Bard et al., 1980). Chalt volcanic Group, Jaglot Group, Chilas Complex, Kamila Amphibolites and Jijal Complex combine to form KIA. The Southern Suture (MMT) marks its southern boundary whereas the Northern Suture (MKT) in north marks its contact with the Eurasian plate. Which is largely pelitic and metamorphosed sub-ordinate sediment, intruded by granitoids of Jurassic to Cretaceous age and Karakoram batholiths in north and south respectively (Debon et al., 1987). These variations in lithology along the KKH from Hassan Abdal in south to Khunjerab Pass in north makes it unique site.

Landslide Inventory Map

Three types of landslide maps are generally prepared: (a) landslide density maps (b) landslide inventory maps (c) landslide hazard and risk maps. Landslide density maps explain spatial concentration of landslides in a specific area (DeGraff 1985). Whereas, hazard and risk maps describe hazard and risk level, associated with landslides. It involves qualitative, semi-quantitative

7

Distribution and Lithological Control of Landslide ______and quantitative assessment of hazard (Guzzetti et al., 1999a). A landslide inventory map demonstrates locations, types and extent of landslides in an area. It can be one map with all landslides or a set of maps with different landslide types (Fausto Guzzetti et al., 2010). Inventory maps are further classified into three: (a) small scale inventory (>1:200,000) (b) medium scale inventory (1:25,000 – 1:200,000) (c) large scale inventory (<1:25,000) (Guzzetti et al., 2000). Preparation of these maps involves single or multiple steps. It can simply be based upon previous reports, published articles, news and books or interpretation of aerial photography or extensive field surveys. The choice of method entirely depends upon scope of inventory prepared. Inventory maps are basically prepared for four purposes: (a) records of landslide events and their extent and magnitude (b) preparation of susceptibility maps (c) quantification of geological and morphological parameters responsible for occurrences (d) role of mass movements in landscape development (Fausto Guzzetti et al., 2010). The scope of the map defines required scale of a map. For regional susceptibility mapping, small scale inventory maps are required, prepared after published reports, articles and books. Medium and large scale maps are generally used for detail investigation to suggest remedies and to predict ground response to extreme weather conditions and seismic events. Credibility of inventory map entirely depends upon data sets used for its preparation. Worth and abundance of reports and publications, experience and expertise of expert and resolution and accuracy of aerial photography influences quality of maps.

In our case, landslides were classified (Varnes, 1978a) and then identified using their geomorphological and physical characteristics. Broadly, landslides were grouped into two categories: (a) Shallow (b) Deep seated. Former has sliding surface just a few meters deep, mostly in soil cover or weathered bedrock and was further classified into falls, slides and flows. Whereas, latter has failure plane deeper than maximum rooting depth.

During the first stage of the project, reports from Geological Survey of Pakistan (GSP) (Fayaz et al., 1985; Khan et al., 2003, 1986), published articles (Hewitt, 2009, 2002, 2001, 1998; Hewitt et al., 2011) and road clearance logs prepared by Frontier Works Organization (FWO) were considered to prepare multi-temporal landslide inventory of the Highway. The preparation of these logs was then followed by interpretation of multi-temporal satellite imagery (Google Earth Pro) to validate landslide locations (Fig. 2.1), we had from archives at first stage.

8

Distribution and Lithological Control of Landslide ______

Figure 2.1: Analysis of multi-temporal satellite imagery to find and map landslides along KKH: Two images before and after the event (≈ 55000 m3) in amphibolites (4th of April 2016), Yellow line showing the head scarp of the rockfall (Google Earth 2019).

Bulging at the toe, fresh talus deposit and vertical cracks near head scarp were used to identify slides. Usually, they occur in preexisting landslides deposits with translation as common failure mode. Falls (rock/scree/debris) exist along steep slopes comprised of jointed rock mass with single or multi-episodic events. Highly tectonised and deformed areas with higher supply of debris in catchment area, has a large number of debris flows. Both falls and flow were identified on the basis of source areas.

During field survey, locations, extents and failure modes of 261 landslides were validated. These landslides were further classified on the basis of size and failure mechanism (Table 2.2, Varnes, 1978a). Distribution of landslides along the Highway is uneven with higher frequency and density in some areas depending upon lithology and structural pattern.

9

Distribution and Lithological Control of Landslide ______

Table 2.1: Types of landslides along the KKH: locations of sections in given in Fig. 3.

Length Shallow Landslides Deep-seated Section (km) Landslides Falls Flows Slide (a) Jijal-Dassu Section 91 37 17 19 1 (b) Sazin-Chilas Section 90 10 9 7 4 (c) Raikot Bridge Section 49 5 8 15 5 (d) Hunza Valley Section 76 13 6 15 10 (e) Sost-Khunjerab Valley Section 86 22 8 16 16 Rest of the Highway 321 2 2 14 0 Total (Hassan Abdal – Khunjerab 713 89 50 86 36 Pass) Jijal-Dassu section has total 74 landslide with rockfall as major failure mode. Whereas, Sazin- Chilas has comparatively low number of landslides (30) with reference to earlier section. Raikot Bridge Section, shorter in length (49 km) comparatively has high landslide density with slide as a main failure mechanism. Hunza Valley section has a large number of deep-seated landslides (10), which dammed river in past. Which has further resulted in the deposition of back water sediments. Presence of quaternary deposits in the valley are the product of this mass movement process. Part of the Highway between Sost and Khujerab Pass, lies in extremely cold weather zone, with long freezing period. Furthermore, a large number of freeze-thaw periods in autumn and spring has formed shattered rock mass, leading to accumulation of scree on steep and long slopes. Scree and rock fall (22) is prominent landslide type in the section. Lithological Control of Landslides

Lithology influences hydrological and mechanical properties of rock mass, further controlling geometry and stability of the slope. Furthermore, structural setting within a lithological unit also defines both the stability and type of landslide process. All parameters associated with joints (discontinuities) within each lithological unit were carefully determined. Additionally, orientations of these joints was categorized into two (dip slope and inverse slope) to further assess slope stability and failure process. Dip slope is a structural setting in which joints (discontinuities) incline along the main slope. It increases the probability of slide and plane failure. Whereas, inverse slope setting explains the joints dipping into the slope, possibly favoring toppling and rotational landslide processes. Three different analysis were performed to evaluate lithological control of landslides. Initially, correlation between percentage of landslide area frequency (LAF, Table 2.2), density and general lithological units (group/formation) was established. Afterwards, specific lithologies were plotted against particular landslide type density. Landslide density was calculated by using following equation:

10

Distribution and Lithological Control of Landslide ______

퐿푛 퐴푓 Equ. (2.1) 퐿퐷 = ( ∗ 100)/( ∗ 100) 퐿푡 퐴푡 Where: LD is landslide density, Ln is number of landslides in a formation or lithology, Lt is total number of landslides, Af is area of a formation or lithology and At is total mapped area.

Twenty six formations/groups along the KKH, were digitized from previous publications (Fayaz et al., 1985; Khan et al., 2003, 2000, 1986). Their boundaries and lateral extent was then validated during field survey. Broadly, these formations were divided into three groups: (a) Indian Plate (b) Kohistan Island Arc (KIA) (c) Eurasian Plate.

Lithological units with reference to slope failures are explained in the following paragraphs, in the order of exposure from Hassan Abdal to Khunjerab Pass in South-North orientation.

2.2.1 The Indian Plate

Abbottabad Formation (AF) AF is comprised of thick-bedded sandstones with shales, siltstones and dolomites, lying over Tanawal Formation of Pre-Cambrian age (Table 2.2). Fossil content indicates its age as Cambrian (Kazmi and Jan, 1997). Part of the formation, exposed along the KKH, consists of massive dolomites and quartzitic sandstones. Tight joints characterizes slopes in this area, making it quite stable. So, no recent landslide activity found in this formation (Table 2.2).

Samana Suk Formation (SF) It is oolitic limestones with subordinate marls and calcareous shales (Table 2.2). Thick bedded limestones are exposed along the KKH between Khota Qabar and Abbottabad (Khan et al., 2003). Slopes are steep but stable in this area.

Hazara Formation (HF)

It is predominately comprised of slates and phyllites with minor shales and limestones content (Table 2.1). Age of this lithological unit is determined as Pre-Cambrian (Crawford and Davies 1975). Rusty and highly weathered slates and phyllites are exposed along the KKH at places between Havelian, Abbottabad and Mansehra. The area is geomorphological mature with low gully erosion and with mini slope catchment. The Highway is also wide enough to avoid any damage due to small-scale debris fall.

Tanawal Formation (TaF)

Quartzose schists and quartzites are main components of the formation, exposed in Batgram area along the KKH. Grades of metamorphism varies laterally with change in distance from

11

Distribution and Lithological Control of Landslide ______

Mansehra Granite. Area is geomorphological mature, without slope failures (Table 2.2). Formation was allocated age of Pre-Cambrian.

Mansehra Granite (MG)

Augen and foliated gneisses and porphyritic granites constitutes this lithological unit. The former is exposed along the Highway between Thakot and Sharkorl whereas later is observable in Mansehra and Ahl areas (Khan et al., 2003). Boulders of different sizes are formed due to spheroidal weathering. These boulders lie at unstable steep slopes, which moves downward during rain or seismic shaking. The frequency of slope failure in this lithological unit is comparatively high (10.29 %; Table 2.2).

Besham Group (BeG)

Precambrian siliceous, green and calcareous schists and amphibolites are exposed between Thakot and Jijal have been termed as Besham Group (Treloar 1990). At few places, schists were metamorphosed into gneisses in close vicinity of the MMT and magmatic intrusions (Fig. 2.5). Furthermore, mylonites are also present at these locations (Fig. 2.2) (Kazmi and Jan, 1997). Gneisses have both dip slope and inverse slope structural setting, resulting into translational and rotation failure respectively. Whereas, schists are extremely weathered at places, resulting into soil forming terraces, which are mostly cultivated. Undercutting of these terraces by hydraulic

Figure 2.2: Inter- and intra-folial folding in gneisses of Besham Group, near MMT. For scale see 1.25 m long pole.

12

Distribution and Lithological Control of Landslide ______action of seasonal nullah triggers retrogressive slope failures at multiple locations. Hence, landslide area frequency is considerably high (11.14%, Table 2.2).

Nanga Parbat Gneisses (NPG)

In Nanga Parbat-Haramosh Massif, crystalline schists and gneisses are collectively termed as Nanga Parbat Group (Kazmi and Jan, 1997). Tahirkheli (1982) assessed age of this lithology as Pre-Cambrian. Gneisses are only exposed along the Highway (between Bunar Gah and Raikot Bridge), therefore, are discussed (Fig. 2.4). It marks northernmost part of Indian plate and has faulted contact (Raikot Fault) with KIA. Remobilization of Indian shield’s granitic rocks has resulted into metamorphosed augen type gneisses, which are highly fragmented due to shallow seismicity (Fig. 2.5). Moreover, foliation has inverse slope setting, resulting into toppling, having low frequency (Table 2.2).

Figure 2.3: Correlation showing lithological control of landslides along the KKH: Fall (boulder, debris, scree), flow (debris, mud) and slide (rock, debris, mud) density VS Major lithologies. (D)- Diorites, (Do)-Dolomites, (L)-Limestones, (S)-Slates, Phyllites etc. (G)-Gneisses, (P)-Pyroxinites, (GA)-Garnet Gabbros, (S)-Serpentinites and Pyroxinites, (A)-Amphibolites, (GN)-Gabbronorites, (GB)- Sheared Gabbronorites.

13

Figure 2.4: Lithological map of the KKH (Hassan Abdal – Chilas Section). For abbreviation, please see Table 2.2.

14

Distribution and Lithological Control of Landslide ______

Table 2.2: Lithological description and LAF (Landslide area frequency) of formations and groups along the Highway (Khan et al., 2003).

Formation/Group Lithology Age LAF (%)▼ River deposits, fluvio-glacial deposits Pleistocene to Quaternary (Q) 5.44 Holocene Oolitic limestones with subordinate marls and Samansuk F (SF) Jurassic 0 calcareous shales Abbottabad Formation Dolomites, dolomitic limestones, sandstones, slates Permo- 0 (AF) and cherty bands Carboniferous

Porphyritic mica Granites, foliated and tourmaline Mansehra Granite (MG) Cambrian 10.29 bearing in parts Black slates, brown phyllites, Groupeywacks siltstones Hazara Formation (HF) Pre-Cambrian 0 and limestones Quartzites, quartzitic schists, phyllitic slates and

Indian Plate Tanawal Formation (TaF) Pre-Cambrian 1.32 conglomerates Schists, gneisses intruded by granites, pegmatites, Besham Group (BeG) Pre-Cambrian 11.14 and amphibolite dykes. Nanga Parbat Gneisses Granitic gneisses Pre-Cambrian 0.86 (NPG) Multi-phase plutons of diorites, tonalites, granodiorites Kohistan Batholith (KoB) Creto-Tertiary 8.75 and granites with minor pegmatites and aplites Garnet gabbros, dunites with chromitite layers, garnet Middle to Early Jijal Complex (JC) 11.38 pyroxenites Jurassic Gabbronorites, pyroxene quartz diorites, gabbros, Chilas Complex (CC) layered gabbros, wherlites, dunites, pyroxenites, Late Cretaceous 1.14 anorthosites and basic dykes. Massive, homogenous, medium grained and banded, Kamila Amphibolites (KA) fine grained amphibolites intruded by granitic sheets, Late Cretaceous 17.86 aplites and pegmatites Turbidites, marbles, green schists, pyroclastics, Theilichi Formation (ThF) Late Cretaceous 2.48 phyllites, basaltic flows. Rakaposhi Volcanics and Pillow basalts, andesites and tuffs, metamorphosed into

Kohistan Island KohistanArc Island Middle Cretaceous 7.30 Sediments (RVS) amphibolites. Turbidites, including parageneisses, schists, calc- Gilgit Formation (GiF) Early Cretaceous 0 silicates and amphibolites Slates, quartzites, chlorite schists, biotite, schists, talc Chalt Schists (CS) schists, quartzite turbidites, (conglomerates) ophiolitic Early Cretaceous 11.62 mélange containing ultramafics, metabasalts. Greenstone Complex Pillow basalts, andesites, rhyolites, quartzites and Early Cretaceous 2.0 (GC) ruffs, marbles and amphibolites with minor slates Karakoram Batholiths Granodiorites, massive diorites, granites, pegmatites Creto-Tertiary 0.55 (KaB) and aplites. Khunjerab Granitic Belt Granodiorites and granites. Late Cretaceous 0.71 (KGB) Shanoz Conglomerates White, saccharoidal pebbles and cobbles embedded in Late Cretaceous to 0 (SC) arenaceous matrix Early Tertiary Yellow, white to light grey thick bedded, dolomites with Jurassic to Gujhal Dolomites (GD) 1.64 minor slates Cretaceous

Dark grey to black slates and phyllites with minor Permo- Misgar Slates (MS) 3.69 quartzites and limestones Carboniferous

Eurasian Plate Dark grey limestones, dolomitic limestones, with Permo- Kilk Formation (KF) 0 alternating bands of arenaceous slates Carboniferous Dark grey argillites, interbedded with dolomitic Permo- Gircha Formation (GirF) 0.29 limestones and quartzites Carboniferous Dark grey, thinly bedded slates, intercalated with Permo- Pasu Slates (PS) 0.54 limestone bands Carboniferous Gneisses, magmatitic gneiss, lime-silicates, marble and Pre-Cambrian to Baltit Group (BaG) 1.07 garnet staurolite schists. Early Mesozoic

15

Distribution and Lithological Control of Landslide ______

Figure 2.5: Nanga Parbat augen gneisses, exposed near Raikot Bridge.

2.2.2 The Kohistan Island Arc

Jijal Complex (JC) It is a series of mafic-ultramafic of early to middle Jurassic age, existing between southern suture in south and Pattan fault in north. Dunites, pyroxinites, serpentinites and garnet gabbros configure the entire complex. A thick mylonitic zone marks the early part of the complex (Bard et al., 1980). Due to its location at hanging wall of the MMT, it is highly fragmented (Fig. 2.6). Moreover serpentinization and weathering has resulted weak rock mass. Gully erosion is prominent in this area. Abundance of fine crushed matrix in pyroxinite and serpentinite decreases permeability and hinders infiltration leading excessive runoff, further triggering debris flows. Whereas, Dunnite has a large number of rockfall, controlled by three joint sets (Fig. 2.3). The frequency of landslides in the complex is considerable high (11.38%).

16

Distribution and Lithological Control of Landslide ______

Figure 2.6: Highly fragmented pyroxenites and serpentinites of Jijal Complex. Kamila Amphibolite (KA)

Southern part of KIA consists of amphibolite complex, known as Kamila Amphibolite. It is chiefly composed of amphibolite with minor hornblendites, hornblende gneisses, diorites and granitoids. It has been classified into two parts: banded amphibolite and massive and sheared amphibolite. Former is fine grained of volcanic origin whereas later is coarse grained of plutonic origin (Kazmi and Jan, 1997). Shearing, complex folding and imbricate thrust slicing has turned it into a structurally complex area. Ductile shearing and isoclinal folds are distributed throughout lithological unit (Yamamoto et al., 2011). Part of the Highway in NE of Pattan transects shear zones at regular intervals of ~10 km. Furthermore, uncontrolled blasting for road construction has aggravated the situation, resulting into overhangs at numerous places (Fig. 2.7). Therefore, in Monsoon and Westerlies, differential erosion based rockfalls are frequent between Kiru and Leo. Moreover, in north of Dassu, when KKH runs along right bank of the Indus River, rockfalls are common along steep slopes with discontinuities dipping into free face (reverse slope). Also, colluvium based debris flows are common in both JC and KA. Colluvium in former has more fine material as compare to latter, with more frequent debris flows, whereas latter has more debris slides. Hence, KA has the highest landslide area frequency (17.86%, Table 2.2).

17

Distribution and Lithological Control of Landslide ______

Figure 2.7: Massive and sheared amphibolites.

Chilas Complex (CC)

It is 40 km wide and 300 km long complex of ultramafic to mafic rocks of late Cretaceous age (Khan et al., 1989). It encroaches Jaglot Group in north and Kamila Amphibolites in south. It is primarily composed of gabbronorite association and secondarily of ultramafic-mafic-anorthositic composition (UMA) (Kazmi and Jan, 1997). Structural studies found its initial position at the base of island arc (Coward et al. 1986). Isoclinal folding and imbrication by thrust faulting is responsible for exposed thickness of the complex. Furthermore, it is strongly foliated and characterized by stress release joints with short persistence, leading to the formation of huge blocks (≈6m3). Most of this lithological unit lies in semi-arid to arid zone. The area is characterized by sparsely vegetated and large catchment areas (≈2.5 km2).

18

Distribution and Lithological Control of Landslide ______

Figure 2.8: Gabbronorite association of Chilas Complex. Foliation has inverse slope setting, causing toppling.

Rockfall is the major and frequent landslide process with occasional debris flows (Fig. 2.3). However, overall, landslide area frequency is low (1.14%, Table 2.2).

Kohistan Batholith (KoB)

Calc-alkaline and intrusive rocks of Cretaceous-Tertiary age between Gilgit and Gorgali Gah is termed as Kohistan Batholith. These plutons have been categorized into deformed and un- deformed (Khan et al., 2003). Deformation took place during collision located near Shyok Suture (Northern Suture). Diorite, orthogneisses, pegmatites and aplites constitute the Kohistan Batholith. Type of slope failure in plutonic bodies are influenced by its deterioration by all means of weathering. KoB is less weathered than MG, therefore, has more frequent rockfalls (8.75%) while MG has more rotational and slump failures.

Thelichi Formation (TeF)

Slates, marbles and volcanics of upper part of north dipping tectonic stack, the Jaglot Group, has been described as Thelichi Formation (Kazmi and Jan, 1997). Several small mafic dikes intruded the formation at different places (Yamamoto et al., 2011). These dikes and bedding planes are steeply dipping towards north. Shearing and fragmentation of rock mass led into formation huge

19

Distribution and Lithological Control of Landslide ______pile of talus deposit on slopes. Mobilization of these deposits by rain water results into occasionally debris slides and flows (2.48%, Table 2.2).

Gilgit Formation (GiF)

Basal part of the Jaglot Group is composed of schist and paragneisses (more than 80%), also known as Gilgit Formation (Yamamoto et al., 2011). Magmatic intrusion of Chilas Complex and Kohistan Batholith and crustal thickening caused by arc accretion metamorphism up to high grade (Khan et al., 1997). It is exposed along the Highway between Jaglot and Gilgit. At places, rock mass is highly shattered and sheared leading to accumulation of scree on steep slopes, highly susceptible to differential erosion. However, landslide frequency is very low (Table 2.2).

Chalt Schists (CS)

The lithological unit is primarily comprised of slates, schists and quartz chlorites schist with quartzite and marble bands. Some parts of lithological unit near Ghulmet area (Nagar) contains ultramafics (Khan et al., 2003). The lithological unit is characterized by an active fault, which is responsible for deformation of thinly bedded and foliated schists. These schists lie at very steep angle (≈50o – 60o) and are highly susceptible to erosion. Whereas, quartzite is comparatively strong, forming large blocks. Precipitation water initially removes and erodes fine debris, producing debris or mud flows, leaving behind coarser or large boulders. Now, these left boulders have low angle of friction or repose and fall onto the Highway. So, this process of differential erosion causes frequent rockfalls and debris slides (11.62%) during rainy season.

Rakaposhi Volcanic Group (RVC)

Rakaposhi Volcanic Group (RVC) is comprised of Andesites, basalts, tuffs, rhyolites, dacites, amphibolites, phyllites, slates, banded quartzites and greywackes of early Cretaceous. It is well exposed along the Highway, in southwest of Chalt village. Where, phyllites and quartzites are highly tectonised and fractured, highly susceptible to slope failure. Landslide area frequency is quite high (7.30%, Table 2.2).

2.2.3 The Eurasian Plate

Baltit Group (BaG)

Previously it was considered a part of the Darkot Group. Marbles, gneisses, slates, schists and lime-silicates combine to form the group (Khan et al., 2003). Slates have faulted contact with ophiolitic mélange of Chalt Schists. Farther in northeast, marble sequence is present, sandwiched between slates and gneisses (Fig. 2.9). . Age of Baltit Group was assessed as Pre-Cambrian to

20

Distribution and Lithological Control of Landslide ______early Paleozoic (Tahirkheli 1982). Foliation in gneisses has both dip and inverse slope orientation, depending upon the presence of the Highway along the side of valley (north or south). The area witnesses a number of deep–seated rockslides, which dammed Hunza River in past. Whereas, the LAF of shallow landslides is comparatively low (1.07%). Rockfalls and mud flows are common in early spring, caused by snowmelt.

Figure 2.9: Gneisses of Besham Group have foliation with dip slope setting.

Karakoram Batholith (KaB)

Intrusive rocks in north of Serat, having transitional contact with the Baltit Group are called as Karakoram Batholith. It is predominantly granites and granodiorites with some alterations. Southern part of batholith has high concentration of aplites and pegmatite dykes (Khan et al., 2003). It has faulted contact with Pasu Slates in north. The granodiorites are widely jointed leading to the formation of huge boulders. Furthermore, the shattering effect of extreme weathering conditions has weakened the rockmass, resulting into occasional rockfalls.

21

Distribution and Lithological Control of Landslide ______

Shanoz Conglomerate (SC)

White saccharoidal marble pebbles and cobbles embedded in fine grained arenaceus matrix of metamorphic origin, are called Shanoz Conglomerates (Khan et al., 2003). Exposure is in Hunza Valley along right bank of river, 4 km south of Khaiber. The area is quite stable with almost negligible landslide activity.

Pasu Slates (PS)

Dark grey to black slates with intercalated limestones and quartzites are exposed in north of Hussain Bridge (Fig. 2.8). The age of Slates was determined as Permian-Carboniferous. It has sharp and intrusive contact with granodiorite of the Kohistan Batholith in south (Kazmi and Jan, 1997) whereas faulted contact with Ghujal Dolomite in north. Extremely freezing weather conditions has resulted into shattered rock mass, leading to formation of scree slopes and famous “Pasu Cones”. These scree slopes lie at steep angle, are susceptible to erosion, influenced by precipitation and snow melt. But, the landslide frequency is relatively low (Table 2.2).

Gujhal Dolomite (GD)

A thick sequence of crystalline dolomites in north of Pasu has been termed as Gujhal Dolomite. Moreover, arenaceous slates and shale are intercalated in dolomites. Presence of marble patches within sequence indicates partial metamorphism. Three joint sets characterize the dolomite with orientation favoring less frequent rockfalls (Table 2.2).

Gircha Formation (GirF)

The formation is predominately dark grey argillite with interbedded dolomite and quartzite (Table. 2.2). It has lithological variation from iron rich beds to quartz veins. Previously, slates of Pasu and Kilik Formation were considered a part of it (Gaetani et al., 1995). Permo-Carboniferous age is assigned to this lithological unit (Khan et al., 2003). The formation is exposed along the KKH between Markhun and Gircha (Fig. 2.8). The area is characterized by wide valley with gentle topography and hence have low number of landslides.

22

Distribution and Lithological Control of Landslide ______

Figure 2.10: Lithological map of the KKH (Chilas-Khunjerab Pass Section). For abbrevations, please see Table 2.2.

23

Distribution and Lithological Control of Landslide ______

Kilk Formation (KF)

Highly tectonised, thick bedded limestones, dolomitic limestones and dolomites, seceded from slates comprises the formation. Bedding is sub-vertical to vertical in most of parts (Khan et al., 2003). However, landslide frequency is quite low (Table 2.2).

Misgar Slates (MS)

Monotonous sequence of dark grey slates exposed along the Highway in north of Sost until Khujerab Pass, is known as Misgar Slates. This formation is approximately 3500 meters thick and has calcareous intercalations and dolerite, gabbro, pegmatite, and aplite intrusions. Quartzite content increases along the Highway, towards north. These slates are highly sheared and thrusted and thickened at many places (Khan et al., 2003). Furthermore, long freezing period and repeated freeze-thaw cycles has shattered rock mass leading to scree accumulation on long and steep slopes (Fig. 2.11). Differential erosion in these scree deposits triggers mud flow and then boulder fall at later stage. The lithological unit covers a large area and contains significant number of landslides. However, landslide area frequency is comparatively moderate (Table 2.2).

Figure 2.11: Slates, phyllites and quartzites of Misgar Slates.

24

Distribution and Lithological Control of Landslide ______

Khunjerab Granitic Belt (KGB)

Medium grained granodiorites exposed near the Khunjerab Pass along the KKH (Fig. 2.8), and are termed as Khunjerab Granitic Belt (KGB). Most of the magmatism in the Karakoram is granitic except some rare places having mafic intrusions (Kazmi and Jan, 1997). It is separated from Karakoram granitic belt by the Reshun Thrust. It is thought to be result of subduction associated with KIA collision with the Eurasian Plate. Landslide activity in this lithological part is considerably low.

Quaternary (Q)

A wide range of unconsolidated Quaternary deposits are present along the Highway. Fluvial- glacial, alluvial, landslide and terraces deposits are amongst them. Between Hassan Abdal and Thakot, rivers have deeply incised reworked alluvial deposits, leading to formation of 60 to 70 m deep gullies. Area between Havelian and Hassan Abdal and between Mansehra and Shinkiari is geomorphological mature (Khan et al., 2003).

Highway section between Thakot and Sazin passes through steep slope, partially covered by glacial moraines, talus and scree deposits. Sazin onward until Gilgit, valley contains two types of quaternary deposits: large landslide deposits (>6 km2) and lake deposits formed during river damming by blocked by these landslides.

The Hunza Valley has also witnessed a large number of historical landslide damming leading to accumulation of Quaternary deposits in the area. Deposition of alluvial deposits due landslide caused deviation in river course is common in both Sazin onward sections of the KKH (Fig. 2.10).

From Shishkat to Khunjerab Pass, valley has large Alpine glaciers (Ghulkin, Gulmit, Batura and Pasu), generating glacial sediments. Semi-consolidated end moraines are lying at steep angles, along the KKH.

It contains all types of landslides, particularly debris slide. These deposits contains two types of mass movement phenomena: retrogressive failures and differential erosion. Later triggers mud flows at first stage and then boulder falls at the second stage.

25

Distribution and Lithological Control of Landslide ______

Figure 2.12: Alluvial deposits formed due to deviation in river channel during landslide damming. Red line is indicating recent slope failure, buried three motor cyclists. Small figure is showing location of alluvial deposit, formed by river deviation.

26

Structural and climatic control of Mass Movements ______

3 Structural and climatic control of Mass Movements

Ali S, Schneiderwind S, Reicherter K (2017) Structural and Climatic Control of Mass Movements along the Karakoram Highway. In: Advancing Culture of Living with Landslides. Springer International Publishing, pp 509–516. Abstract

The Karakoram Highway (KKH) connects Pakistan and China by traversing through rapidly rising mountainous area, which is the junction between the Indian and Eurasian plates including the Kohistan Island Arc. Being a plate boundary, the area is highly prone to active tectonics. The Main Mental Thrust (MMT), Main Karakoram Thrust (MKT), Main Continental Thrust (MCT) and Panjal Thrust are the major fault systems operating in the region. The area is seismically active and various major earthquakes (Muzaffarabad Oct, 2005: M=7.6, Afghanistan Oct, 2015: M=7.5) have occurred. Geology of the area primarily consists of rocks including sedimentary, metamorphic and plutonic rocks. Granite and ultramafic rocks, slates and quartzites are the dominant lithologies of the area. Alterations of these rocks result in a large amount of incompetent and weak lithologies. The KKH passes through some of the world’s deepest gorges with high relief. Glacial deposits (moraines) fill the floor of these gorges. Since its construction in 1979, it has been damaged at various locations by a number of mass movements. In our study, data of mass movement events was acquired from Frontier Works Organization (FWO), further used to prepare landslide inventory map along KKH. Moreover, this mass movement data was correlated with distance from active faults, seismic information and rainfall data aiming to quantify the regression of mass movements and individual distance to active faults in the study area. Besides, the impact of rainfalls on slope stabilities in the region was also investigated. Active faults in this area have caused brittle deformation of crystalline rocks, resulted into poor to fair rock mass close to faults with closely spaced joints, having low shear strength. As a result, distances from active faults have inverse effect on number of events. Additionally, during heavy rainfall, water seeps down into joints and fractures, decreasing shear strength and increasing pore water pressure and ultimately triggering mass movements. Mass movements in Hassan Abdal-Gilgit Section showed

27

Structural and climatic control of Mass Movements ______dependence on rainfall intensity, whereas, slope failures in Gilgit-Khunjerab Pass section are function of both temperature and rainfall intensity. Introduction

The Karakoram Highway (KKH), also known as Pamir Highway, is the highest Highway in the world. It passes through world’s highest mountain ranges (Karakoram, Himalaya, and Hindukush). Geomorphology and climate along the KKH are unique and complex. Being a junction of two tectonic plates (India and Eurasia) converging at ~4-5 cm/y (Jade 2004), the area is seismically active. Main Mantle Thrust (MMT), Main Karakoram Thrust (MKT), Main Continental Thrust (MCT), and the Panjal Thrust are the major fault systems controlling the structure of the region. Seismic activity along the KKH is demonstrated by 317 M>5 and 10 M>7 Earthquakes events (Muzaffarabad Oct, 2005: M=7.6, Afghanistan Oct, 2015: M=7.5) (Zhiquan et al., 2016). These seismic events have not only caused a large number of co-seismic disasters such as catastrophic mass movements along highway but also have developed a highly fragmented rock mass. General geology of the area primarily consists of a variety of rock mass: sedimentary, metamorphic and plutonic (Kibria and Masud, 2006). Alterations of these rocks resulted into a large amount of incompetent lithologies. Additionally, the KKH passes through deeply incised gorges having high relief and glacial deposits (moraines) in places.

Weather conditions along the KKH are arid to semi-arid and sub-tropical. Mean annual rainfall values vary from 0-15 mm to 2000 mm in places (Fig. 3.1). Furthermore, a wide range of temperature having maximum 46°C and minimum -30°C characterizes the KKH. The complexity and uniqueness of geology, geomorphology and climate has made the KKH a geohazards laboratory with a variety of related events (Fig. 3.3). Mass movements along the KKH is one of the main challenges faced by highway authorities since its construction in 1979. Rock falls, rockslides, debris falls and debris slides and mudflows (Shroder and Bishop, 1998) are commonly reported. There are several factors, which contribute to the occurrence of mass movements. Steep slope of the area is main conditioning factor for different type of hazards (Keefer 1984; Harp and Noble 1993; Wang et al. 2003; Qi et al. 2010; Xu et al. 2013). Fracturing, intense weathering and abundance of unconsolidated material including soils, terrace deposits, glacial moraines etc. and their steep contacts often facilitate landslide process (Ercanoglu, 2005; Nagarajan et al., 2000). Abundance of overhangs due to uncontrolled blasting is often triggered by episodic catastrophic events such as high-intensity rainfall, earthquakes, and rapid snowmelt in mountainous areas (Chen et al., 2006; Dadson et al., 2004; Hovius et al., 1997; Ocakoglu et

28

Structural and climatic control of Mass Movements ______al., 2002). Further, presence of streams along the KKH influences stability by toe erosion or by saturating the toe material or by both (Gökceoglu and Aksoy, 1996; Nandi and Shakoor, 2010). Previous Work

Various researchers have worked on mass movements along the KKH with different approaches. Zhiquan et al., (2016) studied and classified debris flow along the KKH into four categories on the basis of hydrological parameters and their spatial distribution. Liao et al. (2011) analysed bridges in Hunza Valley along the KKH, damaged by glacial debris flows. They assessed mechanical properties of bridges and stress exerted by debris flows and further classified bridge damages into four types. Yang et al. (2011) studied debris flow disasters along the Highway, using a comprehensive fuzzy evaluation method. Kibria and Masood (2007) worked on landslides along the KKH (Havelian to Raikot Bridge) by detailed analysis, mainly focusing on geotechnical parameters. Derbyshire et al. (2001) studied the geomorphological hazards along the KKH along a 200 km section, using gravimetric techniques and hazard survey. Riaz and Khattak (2002)

Figure 3.1: Overview of tectonics & precipitation of Study area (After Hodges 2000). KKH- Karakoram Highway, MBT-Main Boundary Thrust, MMT-Main Mantle Thrust, MKT-Main Karakoram Thrust, ◊-Locations of photos shown in Fig. 3.3.

29

Structural and climatic control of Mass Movements ______studied mass movements triggered by earthquakes that result in destruction of the KKH at various localities. Aims and Objectives

In this chapter, an attempt is made to:

 Evaluate relationship between variations in rainfall intensity, temperature and mass movement frequency.  Check the control of active faults on mass movements.

Regional Tectonic Setting

The KKH is located in a tectonically active region because of its existence on a convergent margin of two plates; Indian and Eurasian plate (Fig. 3.2). This tectonic environment has resulted in several active faults, crossing the KKH at various locations. Two important fault systems, MMT and MKT are prominent structural features of the study area. MMT marks contact between the northern part of the Indian Plate and southern part of the Kohistan-Ladakh Island Arc. MMT has exposure along the KKH near Jijal (Tahirkheli and Jan 1979). It has contact with the Nanga Parbat Haramosh massif, here named as Raikot Fault, showing right lateral and reverse displacement on the western and eastern margins of the MMT respectively. The Kohistan Fault is also exposed near Jijal, along a road cut of the KKH (DiPietro et al., 2000). It is a strike-slip fault showing oblique dextral displacement. MKT marks contacts between the southern part of Eurasian Plate (Karakoram Block) and the Northern part of Kohistan-Ladakh Island Arc and is thrusting southward (Borneman et al., 2015). It has a strike-slip sense in the west of Chitral but thrust sense in the north of Gilgit where it crosses the KKH (Coward et al., 1986). The Karakoram Fault shows a strike-slip sense with right lateral movement. There are other surfaces and blind faults associated with these major fault systems, exhibiting shallow seismicity in study area (MonaLisa et al., 2009). These active faults produced different types of discontinuities and concentration of these discontinuities along these faults is usually high and decreases or increases with distance from faults. Presence of these discontinuities decreased rock strength that ultimately affects the stability of slope. Climate of Study Area

Broadly, the KKH can be divided into two zones: a) Monsoon zone, which starts from Abbottabad to Sazin, which receives heavy precipitation near 1000 mm/year, b) Non-Monsoon, which starts

30

Structural and climatic control of Mass Movements ______after Sazin, where the Indus Valley takes a sharp bend. From Gilgit to Khunjerab Pass, arid to semi-arid conditions prevail in valleys but surrounding mountains have a high precipitation rate throughout the year. North of Dassu, along the KKH, there is decrease in precipitation. Between Abbottabad (1445.2 mm/y) and Dassu (Pattan 1313.286 mm/y), there is almost uniform rainfall. But north of Dassu, there is sudden decrease in rainfall (220 mm/y) due to change in orientation of valley (from NS to ES) as there is sharp bend in the Indus Valley. Therefore, only a limited number of mass movements occurred in this zone near Raikot Bridge, where occurrence was purely due to presence of active fault (Raikot Fault). However, after Chilas, rainfall decreases up to Gilgit (127 mm/y). North to Gilgit it increases towards Hunza (375 mm/y) (Fig. 3.1). Methodology

The Frontier Works Organization (FWO) maintains smooth and hazard free traffic flow by clearing the Highway. For mass movement events, road clearance logs have been acquired from FWO, which contains all necessary information required for analysis. These mass movement events were plotted in GIS with active faults in close proximity of the KKH. A map showing the locations of active faults prepared using previously published data. Then, a buffer analysis (multiple ring buffer of 5, 10, 20, 30 km) was performed to check proximity impact on mass movement events. Afterwards, monthly rainfall and temperature data of six stations along the KKH (Abbottabad, Pattan, Chilas, Bunji, Gilgit, Hunza) for the same period was obtained from Pakistan Metrological Department (PMD) for the same period. Correlations were performed to check rainfall and temperature influence on these occurrences for Gilgit-Khunjerab Pass section, as this part of highway passes through snowcapped valleys. However, for Highway section between Hassan Abdal-Gilgit, association between rainfall and number of mass movement was only assessed.

31

Structural and climatic control of Mass Movements ______

Figure 3.2: Geology along the KKH (compiled from: Khan and Jan, 1991; DiPietro et al., 2000; Derbyshire et al., 2001; DiPietro and Pogue, 2004; Hewitt et al., 2011). SV-Sediments and Volcanics, KaB-Karakoram Batholith, CVS-Chalt Volcanics and Sediments, KoB-Kohistan Batholith, CC-Chilas Complex, KA-Kamila Amphibolite, JC-Jijal Complex, PG-Precambrian Gneisses, PMLS-Paleozoic and Mesozoic Limestones and Sandstonees, MRS-Miocene Redstones, MMT-Main Mantle Thrust, KJS-Kamila Jal Shear zone, MKT-Main Karakoram Thrust, KF-Karakoram Fault, KSF-Kamila Strike-slip fault, IKSZ-Indus Kohistan Seismic zone, HSZ- Harman Seismic Zone, RSSZ-Raikot Sassi Seismic zone, YSZ-Yasin Seismic Zone.

32

Structural and climatic control of Mass Movements ______

Figure 3.3: Mass movements along the KKH: (a) Rockfall at ChoChang near Dassu (b) Rockfall near Bunji (c) Debris slide near Bangla Pari (d) Debris Flow near Rahimabad (e) Boulder fall near Harban Nala (f) Mudflow and active debris slide near Jutal (g) Rotational failure near Kayal (h) Snow avalanche near Rahimabad (i) Debris flow blocked one side of tunnel, near Chalt Valley, Hunza (j) Rockfall fall near Shatial.

33

Structural and climatic control of Mass Movements ______

Figure 3.4: Results: (a) Correlation between mass movements and monthly rainfall intensity (Hassan Abdal - Gilgit Section) (b) Correlation between distance from fault and mass movement events (c) Influence of temperature and rainfall intensity on mass movements (Gilgit-Khunejrab Section).

34

Structural and climatic control of Mass Movements ______Fault Control of Mass Movements along the KKH

Active faults in the study area exhibited brittle deformation, resulting into closely jointed and intensely fractured rock mass, forming block of different sizes (from a fraction of meter to several meters), enveloped by fine matrix. Therefore, quality of rock mass close to faults is poor and vice versa. Highly jointed rock mass has low shear strength but high secondary permeability, which increases chances of failure of a slope. After heavy precipitation, water needs passage to reach into openings to increase the pore pressure and to lubricate the failure plane. Buffer analysis has been performed to check an impact of faults on occurrence of mass movements along the KKH. Analysis (Fig. 3.4b) indicated inverse relation between fault distance and mass movement events. Increment in distance from the fault decreases mass movement frequency rapidly. Moreover, value of R2 is close to one indicating a certain correlation between variables. Climatic Control of Mass Movements

Rainfall is considered as one of the important triggering factors for mass movements as it changes hydrogeological parameters (Corominas and Moya, 2008). After infiltration, it lubricates the failure plane and increases pore water pressure leading to a decrease in shear strength and failure of slopes. Analysis of climatic data acquired from the PMD indicated a wide range of temperature and rainfall. The section of the KKH between Abbottabad and Sazin passes through an area which receives heavy precipitation throughout the year. According to analysis (Fig. 3.4a), a strong positive correlation between mass movements and monthly rain precipitation existed. Rise in rainfall intensities in June & July 1996 (over 150 mm/month) triggered sixteen mass movements. From the patterns in the results (Fig. 3.4a), it was speculated that time gap between mass movements and rainfall is because of time spent to seep or infiltrate into rock mass to change hydrological parameters. In this part of our research, time difference between events of rainfall and mass movements and threshold rainfall intensity has not been evaluated.

The section of the KKH passing through Hunza Valley is surrounded by snow covered peaks and glaciers. Mass movements in this area are debris and mud flow, scree and boulder falls. As per analysis of the data, two episodes of mass movements were found in Hunza Valley; one in the months of March and April, triggered by an increase in temperature followed by sudden melting of glacial ice and other in the months of July and August, when there is an increase in rainfall intensity (Fig. 3.4c). Most of the mass movement events dominated in the months of spring and summer (Fig. 3.4c), where there was a sudden increase in temperature in April. The snow melt

35

Structural and climatic control of Mass Movements ______initiated by rise in temperature coupled with increase in precipitation increased pore water pressure that triggered a large number of mass movements. Conclusion

In this study, an effort has been made to check control of climate and faults on mass movement along the KKH. Multiple rings buffer analysis has been performed to check influence of active faults. A strong control of active faults on mass movements has been found. Most of the mass movements were found near active faults and increase in distance from the active faults; their number decrease (Fig. 3.4b). Afterwards, association between mass movement events and monthly rainfall intensity of Hassan Abdal to Gilgit section has been assessed. A strong influence of rainfall intensity on mass movements events was witnessed (Fig. 3.4a). The cumulative rainfall of 530 mm in (June, July, August 1995) and 512 mm in February, March and April 1999 triggered 34 and 72 mass movements respectively (Fig. 3.4a). The section between Gilgit and Khunjerab pass is surrounded by snowcapped mountains and glaciers, therefore, correlation of average monthly temperature, average rainfall intensity with mass movement was performed. Through this correlation (Fig. 3.4c), it has been found that a rise in temperature initiated rapid melting of glacial ice which triggered mass movements in March and April. Afterwards, increase in rain precipitation triggered mass movements in the months of July and August. It is important to mention, in this chapter, number of mass movements represent number of Highway’s blockages. Multiple episodes of landslide activity at single locality means numerous mass movements events.

36

Meteorological Variation and Temporal Distribution of Rockfalls ______4 Meteorological Variation and Temporal Distribution of Rockfalls

Introduction

The Karakoram Highway (KKH) is a part of China’s “One Belt, One Road” (OBOR) initiative and is an important trade route between Pakistan and China (Fig. 1). It passes through rapidly rising and seismically active mountainous area. Different geological, tectonic and climatic environments typify the KKH. It originates from Hassan Abdal, located at Indian plate, enters the Kohistan Island Arc and then the Eurasian plate at Jijal and Chalt towns respectively. Three important regional faults (Main Mantle Thrust, Main Karakoram Thrust and Karakoram Fault) and many local tectonic features cross the Highway at different locations. The Highway passes through active seismic zones (Indus Kohistan Seismic Zone-IKSZ, Herman Seismic Zone-HSZ, Raikot-Sassi Seismic Zone-RSSZ and Yasin Seismic Zone-YSZ) which were responsible for destructive seismic events (Pattan 1974: M=6.2 and Kashmir Earthquake: M=7.6) (Ali et al., 2018a; Casagli et al., 2017). Change in altitude of the KKH from 439 m to 4600 m led into different climatic environment having wide range of precipitation (250 mm/a to 1400 mm/a) and temperature (from 45 Co in South to - 20 Co). All these existing circumstance has made study area an ideal location to investigate different aspects of landslide phenomenon.

The Highway’s stability is in question since its operation in 1979 (Ali et al., 2018a; Casagli et al., 2017). Uncontrolled blasting for its excavation triggered many landslides. Some landslide zones are still active and posing threat to travelers. Rockfall is a common landslide process along highways in steep terrain. Along the KKH, highly fractured and jointed rock mass in close vicinity to active faults is highly susceptible to rockfall occurrence. Geological and geomorphic conditions make area subject to rockfalls but are always prompted by external factors such as earthquake, extreme weather conditions (heavy rainfall and temperature) and several others (D’Amato et al., 2016). Changes in weather conditions (precipitation and temperature) influences mechanical response of rock mass by fluctuating pore water pressure, leaching, frost wedging and propagation of thermal stresses during thawing period.

37

Meteorological Variation and Temporal Distribution of Rockfalls ______

Figure 4.1:(A) location of the Karakoram Highway (KKH) (B) Overview of topography, tectonics and weather conditions along the Highway: MTF-Main Thrust Fault, STF-Secondary Thrust Fault, SSF-Strike-slip Fault, MMT-Main Mantle Thrust, MKT-Main Karakoram Thrust, MBT-Main Boundary Thrust (Ali et al., 2017, 2018). Yellow rectangles represent three sections (a, b and c) whose rainfall data, freeze-thaw cycles and analysis are shown in fig. 4.2, 4.3 and 4.4. White stars showing locations of two sites (Fig. 4.5), explained to avoid bias related to local geological variation.

38

Meteorological Variation and Temporal Distribution of Rockfalls ______

Figure 4.2: Distribution of mean monthly rainfall through a year, along five weather stations. Pattan, Gilgit and Sost represent three sections, mentioned in Fig. 4.1.

Figure 4.3: Variations in monthly freeze-thaw cycles in a year. Three weather stations represent section, displayed in Fig. 4.1.

39

Meteorological Variation and Temporal Distribution of Rockfalls ______

Previously, some researchers correlated metrological changes with landslide occurrence (D’Amato et al., 2016; Delonca et al., 2014; Douglas, 1980; Luckman, 1976; Mateos et al., 2012; Matsuoka, 1994; Perret et al., 2006; Sandersen et al., 1996; Sarro et al., 2014). In this study, an effort is made to check the control of precipitation and variations in temperature (freeze thaw) on rockfall occurrence. Climatic Variations in Studied Sections

Variations in weather conditions along the Highway, enabled to investigate temporal distribution of rockfalls in a year. Based on weather conditions, three section were selected: (a) Sost section (b) Gilgit section (c) Pattan section. Weather data for two stations was available and acquired from PMD, whereas for Sost, it was downloaded from a web source. Mean monthly rainfall data was used for an overview and daily rainfall was utilized for correlations. Minimum and maximum daily temperatures were exploited to interpret positive and negative gradient of temperature. Freeze-thaw cycles were calculated for each month in year in each section. Freeze-thaw cycle is based on minimum and maximum temperature. It is important to mention, a single day can represent only one freeze-thaw cycle. Pattan section lies in southern part of the KKH, having torrential rainfall during Monsoon and Westerlies (Fig. 4.2). Furthermore, December, January and February has mean 4, 14 and 12 freeze-thaw cycles respectively. While, Gilgit section occupies areas around Gilgit city. It is characterized by low rainfall intensity (Fig. 4.2) but with high number of freeze-thaw cycles in December, January, February and March (Fig. 4.3). Sost section constitutes high northern part of the KKH, lying in extremely cold conditions (Fig. 4.1). It has high rainfall intensity in March, April and May and large number of freeze-thaw cycles in March, April, May, October and November. Besides, a long freezing period (>55 days), when maximum temperature remains below 0 Co, characterize this section. Methodology

Rockfall inventory for five years (February 1982-August 1983, August 1996-July 2000) was prepared by using road clearance log of Frontier Works Organization (FWO) and Geological Survey of Pakistan (GSP) publications (Fayaz et al., 1985; Khan et al., 2003, 1986). Weather data for the same period was acquired from Pakistan Meteorological Department (PMD) and the National Centers for Environmental Prediction (NCEP) web source. We used rockfall events of three sections (locations in Fig. 1) for correlation. Further, we performed statistical analysis for specific sites to avoid bias related geological heterogeneity.

40

Meteorological Variation and Temporal Distribution of Rockfalls ______Research Outcomes

Due to different climatic environments, three sections (Fig. 4.1) along the KKH were selected to check any association between meteorological factors and rockfall occurrences. Section “a” is north most part of the KKH. Due to its altitude upto 4600 meters, it experiences extreme weather condition with minimum temperature -30 Co in winter. Rock mass undergoes through the longest freezing period (>55 days). During this period, when maximum temperature remains below 0 Co, frequency of rockfall events is almost negligible (Fig. 4.4a). Rockfall frequency is high after mid- April, when ice melts leading to loss of cohesion. Ice thermal wedging during positive gradient of negative temperature in mid-March and later produces some rockfalls (Fig. 4.4a). Differential erosion during ice melt and rainfall in summer is a major triggering factor in both section “a” and “b”. Results from this step validated earlier observations.

Section “b” lies in arid with low precipitation (250-500 mm/a). Temperature varies between -9.6 Co in winter and 44 Co in summer. Results showed that, a large number of repeated diurnal freeze thaw cycles and ice melt in the end of February and start of March resulted in rockfalls. Further, differential erosion and loss of cohesion are responsible (Fig. 4.4b). Expansion of joints during freezing period and then loss of cohesion after ice melt is the main reason for the catastrophe.

Section “c” lies in a climatic region which receives heavy precipitation during the months of July- August (Monsoon) and March-April (Westerlies). Here, occasionally daily rainfall exceeds 140 mm. Minimum and maximum temperature in this area is -5.3 Co in winter and 39 Co in summer respectively. Slopes along the KKH, between Jijal and Dassu, are orientated to south and are exposed to sunshine with wide range of temperature. Most of the rockfall events are concentrated in the months of February and March. Precipitation during this period, coupled by cumulative effect of diurnal freeze thaw cycles is responsible for frequent rockfalls (Fig. 4.4c). Rainfall during freezing period is responsible for expansion of joints.

41

Figure 4.4: Correlation between precipitation, temperature (min. and max.) and rockfalls for four years period (August 1996 – July 2000). Locations of sections “a, b and c” are given in Fig. 4.1B. (a) Green arrow is indicating an increase in negative minimum temperature, which caused ice thermal wedging. Rainfall coupled with cumulative diurnal freeze thaw cycles led into large number of rockfalls. Lastly, no rockfall occurred during long freezing period. (b) Snow melt in early spring led into loss of cohesion and ultimately rockfalls (green rectangles). Even when rainfall is less than 3 mm/d. (c) Black dotted circles are explaining occurrence of large number of rockfall due to precipitation during continuous diurnal freeze thaw cycles.

42

Figure 4.5: Site specific correlation (a) Rockfall near Kiru (see location in Fig. 4.1). Black circles encompassing rockfalls during rainfall with multiple freeze-thaw cycles. (b) Rockfall in Sost section (see location in 4.1). Black rectangles contain rockfall events during positive gradient of negative temperature.

43

Meteorological Variation and Temporal Distribution of Rockfalls ______To authenticate triggering mechanism, it is important to minimize any bias produced due to variations in geological, lithological conditions and slope orientation. Therefore, two sites (Fig. 4.1) were chosen to perform new correlations between meteorological factors and rockfall frequency in different months (Fig. 4.5).

Kiru rockfall site in Pattan section was chosen for site specific correlation. It is 1.5 km long zone, predominately composed of diorites and massive, sheared amphibolites. Three discontinuity sets exist with one dipping into the slope. Toppling is an important mode of slope failure. Uncontrolled blasting during construction of the Highway, produced overhangs in shattered rock mass. This sites is situated in Monsoon region where rainfall during cold weather with multiple freeze-thaw cycles triggered rockfalls (Fig. 4.5a).

Rockfall occurrences in Sost section, previously termed “Notorious Killing Zone” (Khan et al., 2003), were correlated with daily rainfall, daily maximum and minimum temperature. It is situated at 3700 meters above sea level. Highly tectonised slates with lenticular quartzites comprise the site. Highly fissile slates produced scree slope whereas quartzites are quite hard enough to form steep slopes. After long freezing period, positive gradient in maximum temperature initiated ice melt, further leading to loss in cohesion and differential erosion. Therefore, rockfalls are clustered in this particular period (Fig. 4.5). Moreover, freezing period hardly found rockfall events (Fig. 4.5).

It is necessary to mention that metrological variations are not the sole cause of rockfalls in the area. However, rockfall frequency increases or decreases in response to these changes.

44

Landslide Susceptibility Mapping by using GIS ______

5 Landslide Susceptibility Mapping by using GIS

Ali S, Biermanns P, Haider R, Reicherter K (2018) Landslide susceptibility mapping by using GIS along the China–Pakistan economic corridor (Karakoram Highway), Pakistan. Nat Hazards Earth Syst Sci Discuss 1–28. doi: 10.5194/nhess-2018-39

Abstract

The Karakoram Highway (KKH) is an important route, which connects northern Pakistan with Western China. Presence of steep slopes, active faults and seismic zones, sheared rock mass and torrential rainfall make the study area a unique geohazards laboratory. Since its construction, landslides constitute an appreciable threat, having blocked the KKH several times. Therefore, landslide susceptibility mapping was carried out in this study, to support Highway authorities in maintaining smooth and hazard-free travelling. Geological and geomorphological data were collected and processed using a geographic information system (GIS) environment. Different conditioning and triggering factors for landslide occurrences were considered for preparation of the susceptibility map. These factors include lithology, seismicity, rainfall intensity, faults, elevation, slope angle, aspect, curvature, land cover and hydrology. According to spatial and statistical analyses, active faults, seismicity and slope angle mainly control the spatial distribution of landslides. Each controlling parameter was assigned a numerical weight by utilizing the analytic hierarchy process (AHP) method. Additionally, the weighted overlay method (WOL) was employed to determine landslide susceptibility indices. As a result, the landslide susceptibility map was produced. In the map, the KKH was subdivided into four different susceptibility zones. Some sections of the Highway fall into high to very high susceptibility zones. According to results, active faults, slope gradient, seismicity and lithology have a strong influence on landslide events. Credibility of the map was validated by landslide density analysis (LDA) and receiver operator characteristics (ROC), yielding a predictive accuracy of 72% which is rated as satisfactory by previous researchers.

45

Landslide Susceptibility Mapping by using GIS ______Introduction

Landslides are a result of different geodynamic processes and represent a momentous type of geohazard, causing economic and social loss by damaging infrastructure and buildings (Vallejo and Ferrer, 2011). Landslides are mainly caused by conditioning and triggering factors. Conditioning factors include relief, lithology, geological structure, geomechanical properties and weathering, whereas precipitation, seismicity, change in temperature and static or dynamic loads are triggering factors. Variations in these factors affect the occurrence of landslides. Heterogeneity in lithology influences hydrological and mechanical characteristics of rock mass. Slope morphology (curvature) depends upon lithology and structure within it. Size and type of mass movement changes with variations in lithology and structures. Some lithologies are more permeable and allow water to infiltrate and to increase the pore water pressure. This increase in pore water pressure during rainfall events ultimately affects shear strength of the rock mass and slope stability (Bachri and Shresta, 2010; Cardinali et al., 2000). Whereas, the less pervious rock mass have low infiltration and high runoff leading to debris and mud flows (Canuti et al., 1993). Sheared and highly jointed rock masses contain shallow slope failures whereas rockfalls are concentrated in well-bedded massive rock mass. Distance from a tectonic feature has an inverse relation with rock fracturing and degree of weathering (Pradhan et al., 2010a). State of weathering and fracturing makes slopes unstable (Ruff and Czurda, 2008). Slope is an important driving parameter for slope failures in the same geological and climatic setting (Coco and Buccolini, 2015). Shear strength decreases with increase in slope. Therefore, landslide density increases with increase in steepness (Pradhan et al., 2010a). Curvature expresses the shape of the slope. If it is positive then slope will be upwardly convex and will be concave in the case of a negative value. The later has the ability to retain the water for longer time leading to increase in pore water pressure and hence in slope failures (Pradhan et al., 2010a). Assessment of risks related to landslides was a long term challenge for geologists but was significantly facilitated with the eventual availability of remote sensing data (Shahabi and Hashim, 2015). Preparation of landslide inventory maps, acquisition of geomorphological data (elevation, slope, slope curvatures, aspect), hydrological parameters and extraction of lineaments from remote sensing products is now comparatively an easier task.

Landslide susceptibility mapping is the spatial prediction of landslide occurrence by considering causes of previous events (Guzzetti et al., 1999b). It largely depends upon knowledge of slope movement and controlling factors (Yalcin, 2008). It has hitherto been carried out by many researchers in order to denominate potential landslide hazard zones through evaluation of

46

Landslide Susceptibility Mapping by using GIS ______responsible factors (Basharat et al., 2016; Komac, 2006; Lee et al., 2004, 2002; Shahabi and Hashim, 2015; Süzen and Doyuran, 2004). Preparation of their maps was based broadly on qualitative and quantitative approaches. Early research work (Nilsen et al., 1979) was largely quantitative, utilizing deterministic and statistical correlations and regression analysis of landslides and their controlling factors. In this context, safety factors, calculated on the basis of engineering parameters are adduced to imply deterministic methods. In more recent works, statistical methods are favoured, attempting to draw correlation between spatial distribution of landslides and their controlling factors. Among these are the analytical hierarchy process, bivariate and multi-variate, logistic regression neural networks, fuzzy logic etc. (Basharat et al., 2016; Guzzetti et al., 1999b; Komac, 2006; Lee et al., 2004, 2002; Shahabi and Hashim, 2015; Süzen and Doyuran, 2004). These techniques were proven to be better options for comparatively large and complex areas (Cardinali et al., 2000). Expert opinion and landslide inventories are the decisive components of qualitative approaches (Yalcin, 2008). In most cases, landslide inventories were adduced to estimate failure susceptibility based on previous hazards in locations with similar geological, geomorphological and hydrological set-ups. Some geoscientists (Ahmed et al., 2014; Ayalew et al., 2004; Basharat et al., 2016; Kamp et al., 2008; Kanwal et al., 2016; Shahabi and Hashim, 2015; Yalcin, 2008) incorporated statistical techniques (Analytical Hierarchy Approach (AHP) with Weighted Linear Combination (WLC) and Weighted Overlay Method (WOM)) into qualitative methods to provide the identified factors with a numerical weightage. The combination of AHP and Weighted Overlay Method (WOM) was termed as Multi Criteria Decision Analysis (MCDA) (Ahmed, 2015; Basharat et al., 2016; Kanwal et al., 2016). AHP is a simple and flexible method to analyse and solve complex problems (Saaty, 1987, 1990). It facilitates the estimation of influence that different factors might have on landslide development by comparing them in possible pairs in a matrix. This approach involves field experience and background knowledge of the researcher. A field campaign along the KKH in May 2016 enhanced our knowledge about factors controlling landslides events. Previous research considered MCDA as better choice because of its accuracy to predict landslide hazard (Ahmed, 2015; Basharat et al., 2016; Kamp et al., 2008; Komac, 2006; Park et al., 2013; Pourghasemi et al., 2012). MCDA (combination of AHP with qualitative approaches) was declared a better option for regional studies (Soeters and van Westen, 1996). Previous wide usage in landslide susceptibility mapping, high accuracy, simple process and flexibility according to local variation in landslide controlling parameters compelled us to choose this model. Furthermore, the use of Geographic Information Systems (GIS) facilitated the extraction of geomorphic and hydrological parameters required for susceptibility assessment. Digital Elevation Models (DEMs) are commonly processed in GIS to

47

Landslide Susceptibility Mapping by using GIS ______extract crucial parameters for susceptibility assessment such as elevation, slope, aspect, curvature, watershed etc. (Ayalew et al., 2005, 2004; Basharat et al., 2016; Ohlmacher and Davis, 2003; Rozos et al., 2011; Shahabi and Hashim, 2015; Süzen and Doyuran, 2004). Previously, landslide susceptibility maps of different fragments of northern Pakistan were prepared (Bacha et al., 2018; Basharat et al., 2016; Ahmed et al., 2014; Kamp et al., 2008; Kanwal et al., 2016; Khan et al., 2018; Rahim et al., 2018) (Table 5.1). They used single model based method except two (Ahmed et al., 2014; Bacha et al., 2018) compared performances of two different models. Ahmed et al. (2014), Kanwal et al. (2016) and Rahim et al. (2018) used a regional geological map (1:500000) to produce a landslide susceptibility map of the Upper Indus basin. Whereas, inventories were based on published rock avalanche maps (Kanwal et al., 2016), geomorphological mapping (Ahmed et al., 2014), co-seismic landslides (Basharat et al., 2016) and remote sensing along with filed mapping (Bacha et al., 2018; Khan et al., 2018). Geological, geomorphological and human induced parameters were also considered for production of susceptibility map (Table 5.1).

Table 5.1: Previous work in some parts of the northern Pakistan.

Authors Method Causative Factors Study Area

Aspect, Elevation, Faults, Lithology Land MCE and Muzaffarabad Kamp et al., 2008 Cover, Rivers, Roads, Slope, Tributaries, AHP District Aspect Relief Slope, Curvature, Aspect, Rain, WOM and Upper Indus Farooq Ahmed et al., 2014 Seismic Hazard Fuzzy logic watershed Faults, Drainage, Ndvi, Geology MCE and Aspect, Elevation, Faults, Lithology, Land Basharat et al., 2016 Tehsil Bala Kot AHP Cover, Hydrology, Roads, Slope, Curvature AHP based Shigar and Shyok Slope, Aspect, Lithology, Land Cover, Kanwal et al., 2016 heuristic Basin in Karakoram Faults, Road Network, Streams approach range Haramosh valley, Slope, Aspect, Curvature, Lithology, Land Bagrote valley and Khan et al., 2018 FR Cover, Faults, Road Network, Distance from parts of Nagar Stream, SPI, TWI valley. Aspect, Fault, Geology, Land Cover, Hunza-Nagar Bacha et al., 2018 WOT, FR, Proximity to Road, Slope, Proximity to valley Stream Slope, Aspect, Elevation, Drainage Network, AHP and Rahim et al., 2018 SPI, TWI, Lithology, Fault Lines, Rainfall, Ghizer District WLC Road Network, Land Cover, Soil Texture Elevation, Slope Angle, Aspect, Curvature, AHP and Lithology, Seismicity, Faults, Land Cover, This Study KKH (CPEC) WOM Rainfall Intensity And Distance From Streams

48

Landslide Susceptibility Mapping by using GIS ______General situation of the study area

The Karakoram Highway (KKH), a part of the China-Pakistan Economic Corridor (CPEC), connects northern Pakistan with Western China (Fig. 5.1). It passes through rapidly rising mountain ranges of the Himalaya, Karakoram and Hindu Kush forming the junction between the Indian and Eurasian plates including the Kohistan Island Arc (Derbyshire et al., 2001). The area is characterized by fractured and weathered rock masses, diverse lithologies (igneous, metamorphic, and sedimentary), high seismicity, deep gorges, high relief, arid to monsoon climate and locally high rates of tectonic activity. These conditions make the study area a unique geohazards laboratory. Starting with its construction in 1979, KKH’s stability has been endangered by a variety of geohazards.

The study area is the 840 km long (10 km buffer) Karakoram Highway (KKH), N35, located in the Karakoram Mountains, Himalaya. The area hosts some of the highest reliefs and highest peaks (Nanga Parbat: 8126 m, Rakaposhi: 7788 m) in the world (Hewitt, 1998). Goudie et al. (1984)termed the study area the steepest place on the earth where elevation drops from 7788 m to 2000 m over a horizontal distance of 10 km (Fig. 2b, 2d). From Abbotabad, the Highway leads northwards through the sub-Himalayas entering the Indus Valley at Thakot, and the Hunza Valley at Gilgit, running parallel to the eponymous rivers. From Thakot onwards, it passes through deeply incised valleys and gorges.

Weather conditions along the KKH are not uniform and are characterized by a wide range of annual mean temperatures (-5 °C to 46 °C) and precipitation (15 mm to 1500 mm). The distribution of precipitation is additionally strongly fluctuating throughout the year. During the westerlies (January, February and March) and the monsoon period (July, August), the study area receives heavy rainfall. According to meteorological data, average annual precipitation between Abbottabad and Dassu is 1444 mm. However, north of Dassu, an abrupt change from monsoon to semi-arid to arid conditions is recorded which is owed to a change in valley orientation from north-south to east-west (Fig.5.1 and Fig. 5.3). Furthermore, vertical climatic zonation exists in the Hunza Valley along the KKH. The surrounding peaks and slopes higher than 5000 m receive precipitation greater than 1000 mm per year, whereas the valley floor below is characterized by a semi-arid to arid climate (Hewitt, 1998).

49

Landslide Susceptibility Mapping by using GIS ______

Figure 5.1: Overview of tectonics & precipitation in the study area: (a) Location of the study area in the region (b) Active faults and major earthquake events in the region (USGS Earthquake Catalog, 2017) (c) Locations of the weather stations with mean annual rainfall (Pakistan Meteorological Department) and overview of tectonics and topography of the study: Box 2a and 2b represents location of Fig. 5.2. KKH=Karakoram Highway, MBT=Main Boundary Thrust, MMT=Main Mantle Thrust, MKT=Main Karakoram Thrust (modified after Ali et al., 2017).

50

Landslide Susceptibility Mapping by using GIS ______

The Karakoram and Himalaya host some of the world’s longest continental glaciers with the steepest gradient and highest glacial erosion rates (Goudie et al., 1984). The snouts of some glaciers (Batura, Ghulkin, Pasu, Gulmit, Gulkin) are close to the KKH and partly cross it (Fig. 5.2a, 5.2c). Relatively warm temperatures in the valleys results in sudden melting of ice, frequent surges, catastrophic debris flows and blockage of rivers.

Figure 5.2: a) Glaciers along the KKH (b) Relief along the KKH (c) Pasu Glacier’s snout approaching the KKH (d) Profile drawn along axis drawn in 5.2b: elevation drops from 7788 m to 2000 m over a horizontal distance of 10 km. (See location in Fig. 5.1)

51

Landslide Susceptibility Mapping by using GIS ______

Figure 5.3: (a) Overview of precipitation (mean annual rainfall) and landslides frequency along the KKH (after Khan et al., 2000, Pakistan Meterological Department 1982, 83, 96, 97, 98, 99, 2000, 14, 15, 16, Frontier Works Organization archives) (b) Correlation between landslide events and precipitation (Casagli et al., 2017).

52

Landslide Susceptibility Mapping by using GIS ______Geology along the KKH

Tectonically, the area is characterized by orogenic features that started forming with the onset of the Indo-Eurasian collision 50 Myr ago. Crustal shortening, subduction, and active faulting are still ongoing with convergence rates of ~4-5 cm/y (Jade, 2004) and uplift rates of ~7mm/y (Zeitler, 1985). Main Mantle Thrust (MMT), Kamila Jal Shear zone (KJSZ), Raikot Fault, Main Karakoram Thrust (MKT) and Karakoram Fault are important tectonic features responsible for brittle deformation along the Highway (Fig. 5.4) (Bishop et al., 2002; Burg et al., 2006; DiPietro et al., 2000; Goudie et al., 1984). Due to this brittle deformation, the rock mass is highly jointed and fractured. The general geology along the KKH consists of sedimentary, igneous and metamorphic rocks. Highly active landslide zones were identified from the multi-temporal landslide inventory of the KKH (Fig. 5.6).

Jijal-Dassu, Raikot Bridge, Hunza Valley and Khunjrab valley sections is characterised by a large number of mass movements and therefore, detailed geology is only discussed for these sections. The geology of the Jijal-Dassu section is composed of ultramafic and low to high-grade metamorphic rocks. The Mansehra granite, the Besham group, the Jijal Complex, the Kamila amphibolite and the Chilas complex are important lithological units in this section. The Besham group comprises biotitic gneisses, cataclastic gneisses and quartzite which were metamorphosed during the Himalayan orogeny ∼65 Myr ago (Ding et al., 2012). It shares a faulted contact with the Jijal Complex (Kohistan Island arc) along the northward dipping MMT (Williams, 1989). The ultramafic rocks with garnet granulites and Alpine-type metamorphic rocks between Jijal and Pattan are collectively termed as Jijal Complex (Tahirkheli and Jan, 1979). The Besham group shows signs of crushing of individual minerals and staining of quartz whereas the Jijal Complex is massive and sheared (Khan et al., 1986). Owing to the contacts of the Jijal Complex with the MMT in the north and the Pattan Fault in the south, it is highly tectonised and deformed. The Kamila amphibolite consists of sheared basic lavas and intrusive plutons (Treloar et al., 1996). It is classified into two types: garnet bearing and garnet free amphibolites. The former is massive and sheared due to the presence of Pattan and Kamila Jal shear zones (KJS), whereas the latter is banded. The garnet free amphibolite shares a sheared contact with the Chilas complex, mafic intrusions of predominantly gabbro-norites, sheared gabbro-norites and diorites (Searle et al., 1999).

The Raikot Bridge section exhibits continuous mass movement process because of its location at the seismically active western limb of the MMT known as Raikot Fault, a strike-slip fault with right lateral movement. It is marked by concentration of hot springs and a large shear zone. Granitic

53

Landslide Susceptibility Mapping by using GIS ______gneisses, quartzites, gabronorite, schists and Quaternary sediments are the main lithological units in this section. Continuous erosion by the Indus River and highly deformed rocks are responsible for landslide events.

MKT is a part of the Hunza Valley section and is responsible for many landslide events. The pre- dominant local lithologies of the Baltit Group, Chalt schists, Karakoram batholith as well as Quaternary sediments are highly tectonised and deformed. The highly deformed Misgar slates, along with the Gujhal dolomite and Kilk formation, are the main components of the Sost section. The Karakoram fault is an important tectonic feature in this section. Highly fissile and closely jointed slates are important sources of scree on steep slopes along the Highway. Intense weather conditions aggravate the situation in this section. Seismology

The Highway passes through one of the seismically most active areas in the world. The presence of active thrusts and strike-slip faults gives rise to earthquakes, anon triggering numerous landslides. The seismic activity along KKH is demonstrated by 317 M>5 and 10 M>7 recorded earthquake events (Muzaffarabad Oct, 2005: M=7.6, Afghanistan Oct, 2015: M=7.5) since 1904 (Zhiquan et al., 2016). The Highway passes through important seismic zones: the Indus Kohistan seismic zone (IKSZ), the Hamran seismic zone (HSZ), the Raikot-Sassi seismic zone (RSSZ) and the Yasin seismic zone (YSZ) (Fig. 5.4). The Jijal-Dassu section of the KKH passes through the northern part of IKSZ. IKSZ is 50 km wide and represents a highly active wedge-shaped structure containing a shallow and midcrustal zone (MonaLisa et al., 2009). The Muzaffarabad (2005, M=7.6) and Pattan earthquakes (1974, M=6.2) are recent destructive earthquakes in this seismic zone.

Sazin section of the Highway is a part of HSZ, an active seismic zone hosting recent events with magnitudes of 3 to 6.2. The active Raikot fault traverses RSSZ and is responsible for shallow seismicity of magnitudes 3 to 6.3. Both the fault and the KKH run, in direct vicinity, on the western banks of Indus River. The YSZ encompasses the region surrounding the small town of Yasin. It is characterized by earthquakes with magnitudes between 3 and 5 and focal depths of less than 50 km. The MKT is suspected to be the main source of seismicity for this seismically active region.

54

Landslide Susceptibility Mapping by using GIS ______

Figure 5.4: (a) Regional location of Himalaya (b) Overview of Himalayan geology (c) Geology along the KKH (compiled from: (Derbyshire et al., 2001; DiPietro and Pogue, 2004; Hewitt et al., 2011; Khan and Jan, 1991); Derbyshire et al., 2001; DiPietro and Pogue, 2004; DiPietro et al., 2000; Hewitt et al., 2011). Four boxes represent four sections (a-Jijal- Dassu Section, b-Raikot bridge Section, c- Hunza Valley Section, d-Khunjrab valley Section) MMT-Main Mantle Thrust, KJS-Kamila Jal Shear zone, MKT- Main Karakoram Thrust, KF- Karakoram Fault, KSF-Kamila Strike-slip fault, IKSZ-Indus Kohistan Seismic zone, HSZ- Harman Seismic Zone, RSSZ-Raikot Sassi Seismic zone, YSZ-Yasin Seismic Zone, SV-Sediments and Volcanics, KaB-Karakoram Batholith, CVS-Chalt Volcanics and Schist, KoB-Kohistan Batholith, CC-Chilas Complex, KA- Kamila Amphibolites, JC, Jijal Complex, PG-Precambrian Gneisses, PMLS-Palaeozoic and Mesozoic Limestones and Sandstones, MRS-Miocene Redstones.

55

Landslide Susceptibility Mapping by using GIS ______

Figure 5.5: Flow Chart showing multiple steps involved in the preparation of the susceptibility map: FWO-Frontier Works Organization, PMD-Pakistan Meteorological Department, GSP- Geological Survey of Pakistan, DEM-Digital Elevation Model. Methodology

The flow chart (Fig. 5.5) describes the steps and techniques involved in preparation of the susceptibility map, involving multiple techniques, literature review, field observation and remote sensing.

5.6.1 Literature Review

In the first step, existing information and data for the study area were collected from archives of the Frontier Works Organization (FWO), Geological Survey of Pakistan (GSP), Pakistan Meteorological department (PMD) and research catalogues (Khan et al., 2000). FWO is

56

Landslide Susceptibility Mapping by using GIS ______responsible for clearance and maintenance of the KKH after its potential blockage. Road maintenance log were collected and digitized. Following three important maps were prepared from data collected:

I. A multi-temporal landslide inventory map (Fig. 5.6) was prepared using GIS, based on GSP’s publications (Fayaz et al., 1985)and road maintenance logs of FWO. II. A comprising geological and seismo-tectonic map (1:50000 and 1:250000) of the area was prepared by digitizing and compiling various pre-existing maps (Khan et al. 2000). Data related to lithology, faults and seismicity has been extracted from these maps. III. An annual precipitation map of the study area was prepared by using rainfall data of six weather stations and previous map (Fig. 5.1 & Fig. 5.3) along the KKH. Landslide Inventory

A landslide inventory map is an important instrument to display the location, date of occurrence and type of mass movement (Fausto Guzzetti et al., 2010). Landslide inventory maps are prepared to define and record extent of mass movements in different regions, to investigate an impact of lithology, geological structures (fault, fold etc.) on types, distribution and occurrence of landslides, to use for preparation of landslide susceptibility mapping and to analyse geomorphic evolution of an area. Preparation of these maps involves multiple techniques based on satellite imagery, field interpretations and compilation of previous publications (Guzzetti et al., 2000; van Westen et al., 2006).

In this study, we used real time data of landslide occurrences acquired from road clearance logs of Frontier Works Organization (FWO) for different periods (1982, 1983, 1996, 1997, 1998, 1999, 2000, 2014, 2015, 2016), publications of Geological Survey of Pakistan (GSP) (Fayaz et al., 1985; Khan et al., 2003, 1986), a research article (Hewitt, 1998), Google Earth imagery and field surveys to prepare multi-temporal landslide inventory along the Highway (Fig. 5.6). Polygon outlines for clearly visible landslides on satellite imagery (based on data of FWO and GSP) were drawn. This landslide inventory map was then validated during field campaign. Spatial analysis and validation of the final susceptibility map were performed by using these polygons. Due to small scale of the inventory map, visibility of polygons was extremely low. Therefore, these polygons were then converted and displayed as points in the inventory map (Fig. 5.7).

57

Figure 5.6: Temporal distribution of landslides along the Highway: a and b represent two problematic section shown in Fig. 5.11.

58

Landslide Susceptibility Mapping by using GIS ______

Figure 5.7: Types of Mass Movements along the Highway (a) Thakot to Raikot Bridge (b) Raikot Bridge to Khunjerab Pass (Locations shown in Fig. 5.6).

Field Observation

Locations of landslides, lithological contacts and faults were validated and supplemented during a field visit. In addition to locations, types, size, failure mechanisms and structural control of landslides were determined. Acquired data was further used to prepare landslide inventory map within 2 km2 around the Highway.

5.6.2 Remote Sensing

Geomorphological parameters (elevation, slope, aspect, curvature) and drainage were extracted from Shuttle Radar Topography Mission (SRTM) based DEM (30 m × 30 m). Thematic layers were prepared and classified using the natural break method in GIS. Satellite images of Landsat

59

Landslide Susceptibility Mapping by using GIS ______

8 (19-21 November 2017) were acquired from USGS web portal and then pre-processed by using QGIS 2.18’s Semi-Automatic Classification Plugin (SCP), followed by supervised classification of composite images in GISto prepare land cover map.

Construction of thematic maps

Thematic layers of elevation, slope, aspect and curvature were prepared and classified using the natural break method in GIS. Drainage was extracted from SRTM based DEM by arc hydro tool. A buffer polygon of 300 meters was created to measure distance around streams to form thematic layer of distance from drainage. Faults, lithology and seismic zones were digitized from previously published geological and seismic maps. Multiple ring buffer polygons of 500, 1000, 1500 and 2000 meters around digitized faults were produced. Vector layers of distance from fault and drainage, lithology and seismic intensities were then rasterized. Annual rainfall data was interpolated to create precipitation map and then was combined with previously published annual rainfall map of PMD. Thematic layer of land cover was produced from land cover map.

5.6.3 Analytical hierarchy process (AHP)

Analytical hierarchy process (AHP) is a multi-criteria decision making approach to prepare landslide susceptibility maps. It has been used by previous researchers to assign weightage to landslide-controlling (Basharat et al., 2016; Kanwal et al., 2016; Shahabi and Hashim, 2015). It is based on the user’s decision to weigh factors through their pairwise comparisons. Each factor is assigned a score (1-9) depending upon its relative importance, with increasing impact from 1 to 9 (Table 5.2, Saaty, 1990). The values assigned are based on spatial analysis of data, field observations and experience of the user. If the parameter on the x-axis is more important than the one on y-axis, the value ranges between 1 and 9. Conversely, when the factor on y-axis is more important, the values are in reciprocals (1/2-1/9).

Consistency ratio (CR) is a tool to check and avoid inconsistencies and bias in whole process of rating controlling parameters (Basharat et al., 2016; Kanwal et al., 2016; Pourghasemi and Rossi, 2017; Sarkar and Kanungo, 2004; Taherynia et al., 2014).

Equ. (5.1) 퐶푅 = 퐶퐼/푅퐼

Where CR is consistency ratio, CI is consistency index and RI is random index.

60

Landslide Susceptibility Mapping by using GIS ______

Table 5.2: Fundamental scale for pair-wise comparisons (Saaty, 1987).

Intensity of Importance Explanation

1 Equal Importance 2 Weak or slight 3 Moderate importance 4 Moderate plus 5 Strong importance 6 Strong plus 7 Very strong 8 Very, very strong 9 Extreme importance

CI was calculated by using following equation:

Equ. (5.2) 퐶퐼 = (휆푚푎푥 − 푛)⁄푛 − 1

Where λmax is the maximum eigenvalue of matrix and n is the number of controlling parameters involved (Zhou et al., 2016).

Saaty (1987) produced a table (Table 5.3) of random consistency index (RI) after calculation from 500 samples. Values of RI from this table and calculated CI were then compared to find CR.

Table 5.3: Random Consistency Index (RI)

n 2 3 4 5 6 7 8 9 10 RI 0 0.58 0.9 1.12 1.24 1.32 1.41 1.45 1.51

The value of CR indicates the inconsistency in the expert’s decision during weighing the parameters. Values of CR lower than 0.1 prove the decision consistent, while values greater than 0.1 indicate inconsistency and suggest a revision of judgement. Subclasses in each factor were prioritized by using pairwise comparison procedure (Table 5.4). Curvature and distance from drainage comprised of two and one classes respectively. Therefore, influence of these subclasses was easily scaled without AHP procedure. In next step, each parameter was assigned weightage on the completion of procedure (Table 5.5). Value of CR in our study remained below 0.1, which proves comparisons and weighting criteria reliable, unbiased and consistent.

61

Landslide Susceptibility Mapping by using GIS ______

Table 5.4: Pairwise matrix and weights of factor sub-classes. Class (1) (2) (3) (4) (5) (6) (7) (8) % Importance Aspect (1) North 1.00 2.59 2 (2) Northeast 2.00 1.00 5.89 3 (3) East 3.00 2.00 1.00 7.70 4 (4) Southeast 6.00 3.00 2.00 1.00 12.68 5 (5) South 7.00 4.00 2.00 1.00 1.00 16.71 6 (6) Southwest 8.00 4.00 3.00 2.00 2.00 1.00 21.95 7 (7) West 9.00 5.00 4.00 2.00 3.00 2.00 1.00 24.54 8 (8) Northwest 3.00 2.00 1.00 0.50 0.25 0.50 0.33 1.00 7.94 4 Elevation (m) (1) 432-1000 1.00 4.05 1 (2) 1000-2000 7.00 1.00 41.46 8 (3) 2000-3000 6.00 0.50 1.00 22.26 6 (4) 3000-4000 6.00 0.33 1.00 1.00 20.73 6 (5) 4000-4700 4.00 0.20 0.50 0.50 1.00 11.50 4 Slope (1) 0-15 1.00 3.92 1 (2) 15-30 3.00 1.00 11.00 4 (3) 30-45 9.00 4.00 1.00 41.46 9 (4) 45-65 8.00 3.00 0.50 1.00 30.83 6 (5) >65 5.00 1.00 0.25 0.30 1.00 12.80 4 Land Cover (1) Vegetation 1.00 9.68 2 (2) Water 0.33 1.00 4.75 1 (3) Snow 3.00 5.00 1.00 23.08 4 (4) Bare Rock/Soil 6.00 9.00 4.00 1.00 62.49 7 Rainfall Intensity (mm/y) (1) 0-250 1.00 5.68 1 (2) 250-500 3.00 1.00 12.65 3 (3) 500-1000 5.00 3.00 1.00 27.35 5 (4) 1000-1500 8.00 5.00 3.00 1.00 60.00 9 Lithology (1) Group A (SI, AF, HF, KgB, MG, 2.00 4.52 1 TaF) (2) Group B (BeG, SC, TeF) 4.00 1.00 8.14 2 (3) Group C (KoB, KaB, CC, GilF, 6.00 2.00 1.00 14.83 4 GJ) (4) Group D (KA, GirF, KF, RPV, 8.00 3.00 2.00 1.00 26.02 6 SC) (5) Group E (Qu, JC, OM, MS, PS, 4.00 5.00 4.00 2.00 1.00 46.49 9 NPG, BaG) Seismic Intensity (1) I-III 1.00 3.74 2 (2) IV-V 3.00 1.00 7.63 3 (3) V-VI 5.00 3.00 1.00 14.22 6 (4) VII-VIII 7.00 5.00 3.00 1.00 29.77 8 (5) IX-X 8.00 6.00 4.00 2.00 1.00 44.64 9 Distance from a Fault (m) (1) 0 -500 1.00 50.50 9 (2) 501-1000 0.50 1.00 27.04 7 (3) 1001-1500 0.25 0.50 1.00 15.30 5 (4) 1501-2000 0.20 0.25 0.33 1.00 7.15 4

62

Landslide Susceptibility Mapping by using GIS ______

Table 5.5: Pairwise matrix and weights of all controlling parameters.

Distance Controlling Land rainfall Aspect Elevation Fault Lithology from slope curvature seismicity Weight Cover intensity Factors drainage Aspect 1 1.77 Elevation 3 1 3.05

Distance 9 7 1 23.28 from Fault Lithology 6 5 1/3 1 10.28 Land Cover 4 3 1/4 1/3 1 7.04 Distance from 6 5 1/3 1 2 1 8.90 drainage Rainfall 5 4 1/4 1/2 2 1/2 1 7.08 intensity Slope 9 7 1 3 5 3 4 1 23.74 Curvature 1 1/3 1/9 1/6 1/5 1/5 1/5 1/7 1 1.82 Seismicity 7 5 1/3 1 4 2 3 1/3 7 1 13.03

5.6.4 Weighted Overlay Method

Weighted overlay method (WOM) is a simple and direct tool of Arc GIS to produce a susceptibility map (Bachri and Shresta, 2010; Intarawichian and Dasananda, 2010). Many researchers used WOM to produce landslide susceptibility map (Bachri and Shresta, 2010; Basharat et al., 2016; Intarawichian and Dasananda, 2010; Roslee et al., 2017; Shit et al., 2016). We used an overlay of raster layers of all controlling factors to prepare a susceptibility map. Raster layers of each controlling factor were reclassified and weighted according to their importance determined by AHP (Table 5.4 & 5.5). The cumulative weight of all input layers was maintained at 100. All layers were combined by using weighted overlay tool based on following equation (5.3):

∑ 푊푖 푆푖푗 Equ. (5.3) 푆 = ∑ 푊푖

Where Wi is weight of ith factor, Sij represents subclass weight of jth factor and S is spatial unit of the final map. The completion of this process resulted into the ultimate production of a landslide susceptibility map of the Highway.

63

Landslide Susceptibility Mapping by using GIS ______Results

5.7.1 Landslides along the KKH

A total 261 landslides were used to prepare the map (Table 5.6). Broadly, we grouped these into shallow and deep-seated landslides (rock avalanches). Shallow landslides were further divided into slides, falls and flows based on simplified version of Varnes (1978). Four sections (Table 5.6) are characterised by a large number of landslides during heavy rainfall and snowmelt. Highway blockage and traffic interruption in these sections is a regular phenomenon. Presence of a large number of rock/debris falls (37) in Jijal-Dassu section is due to steep topography formed by deep river incision in ultramfics (Jijal Complex), amphibolites (Kamila Amphibolites) and Gabbronorites (Chilas Complex) of Kohistan Island Arc. Whereas, stress release joints with short persistence in Sazin-Chilas section are responsible for huge boulder falls (>6m3). A large number of slides (rock, debris and mud) and flows (debris and mud) in old large landslide deposits characterizes Raikot Bridge section. Hunza Valley section is dominated by slides (rock, debris and mud) in highly sheared rock mass and falls (rock and debris) in over steepened parts of the valley. Steady flow of traffic along the highest section (Sost-Khunjerab Pass) of the Highway is a major problem due to seasonally influenced falls and slides. This section has a large number of large landslides (16), which dammed the Hunza and Khunjrab rivers in past.

Table 5.6: Types of landslides along the KKH.

Deep-seated Rock Flows (Debris, Shallow Section Landslides/Rock Falls Mud) Landslides Avalanches Jijal-Dassu Section (91 Km) 37 17 19 1 Sazin-Chilas Section (90 Km) 10 9 7 4 Raikot Bridge Section (49 Km) 5 8 15 5 Hunza Valley Section (76 Km) 13 6 15 10 Sost-Khunjrab Valley Section 22 8 16 16 (86 Km) Rest of the Highway (321 Km) 2 2 14 0 Total 89 50 86 36 5.7.2 Causative Factors and Spatial Distribution analysis

Geological, morphological, seismo-tectonic, topographic and climatic factors are generally considered as landslide-controlling parameters (Kamp et al., 2008). The following ten causative factors were considered for preparation of the map: Lithology, distance from faults, seismicity, elevation, slope, aspect, curvature, land cover, rainfall intensity and distance from drainage. Size

64

Landslide Susceptibility Mapping by using GIS ______and distribution of the landslides varies locally, depending on the values of the parameters mentioned above. Thus, the creation of an accurate and precise GIS-based landslide susceptibility map is entirely dependent on the availability of data related to controlling factors (Ayalew et al., 2004). Rockslides, debris slides, rock avalanches, rock fall, toppling, wedging, mudflows and debris flows are important landslide processes along the KKH.

Lithology

Time and type of slope failure is determined by the slope-building lithology. Each lithology is unique in terms of response to stresses and therefore features a particular susceptibility to potential slope failure (Vallejo and Ferrer, 2011). The KKH traverses a great variety of lithologies comprising sedimentary, igneous and metamorphic rocks. According to spatial analysis, Quaternary deposits, the Jijal Complex and the Misgar slates exhibit the highest numbers of mass movement events (Fig. 5.8f).

Distance from faults

The Main Mantle Thrust (MMT), the Kamila Jal Shear zone (KJSZ), the Kamila Strike-slip Fault (KSF), the Raikot Fault, the Main Karakoram Thrust (MKT) and the Karakoram Fault are important structural features in close proximity of the Highway (Fig. 5.4). Landslides are concentrated along these active faults where rock mass is highly deformed (Casagli et al., 2017). The fact that 54% of mapped landslides were found within a maximum distance of 1 km from these faults, while 69% were found within a 2km range, impressively substantiates the postulated strong control of structural features (Fig. 5.8e).

Geomorphologic Factors

Slope angle is an important geomorphic factor responsible for initiation of slope movements (Lee et al., 2004), to be considered for preparation of landslide susceptibility maps. Steep slopes are more susceptible to failure as compared to gentle ones. The study area demonstrates variation in topography ranging from steep to gentle slopes, high plains to narrow gorges and high cliffs. Slope steepness in the area has been divided into five classes. Division of slope steepness into classes was based on statistical analysis. Different classes were tried but found this division better in our study area. More than 50% of landslides occurred in class III (30°-45°) areas, whereas least landslide events (2%) occurred in class I and class V (0 - 15° and >65°) areas (Fig. 5.8a). In addition to slope and elevation, aspect and curvature were also considered important factors for preparation of the landslide susceptibility map. However, in our area these parameters seem to have a reduced influence on landslide occurrence (Fig. 5.8b, 5.8c, 5.8d).

65

Landslide Susceptibility Mapping by using GIS ______

Figure 5.8: Frequency distribution histograms of controlling parameters: a) Slope Angle b) Aspect c) Profile Curvature d) Elevation e) Distance from fault f) Geological formation (Abbreviations explained in Fig. 5.5). BaG: Baltit Group, BG: Besham Group, Cc:Chilas complex, GilF: Gilgit Formation, GirF: Gircha Formation, JC: Jijal Complex, KA: Kamila Amphibolite, KaB: Batholith, KoB: Kohistan Batholith, MS: Misgar Slates, NPG: Nanga Parbat Granitic Gneisses, OM: Ophiolitic Melange, PS: Passu Slates, Qu: Quaternary, RpV: Rakaposhi Volcanics, TF: Theilichi Formation, F: Fault, TF: Thrust Fault, MMT-Main Mantle Thrust, KJS-Kamila-Jal Shear Zone, KSF-Kamila Strikeslip Fault, RF-Raikot Fault, KF-Karakoram Fault, MKT-Main Karakoram Thrust.

66

Landslide Susceptibility Mapping by using GIS ______

Hydrology

The proximity of streams has been considered for the preparation of landslide susceptibility maps by many researchers (Akgun et al., 2008; Basharat et al., 2016; Kanwal et al., 2016; Wang et al., 2015). In our study area, many small tributaries feed main rivers, Indus and Hunza. During Monsoon season and after heavy precipitation events, these streams exhibit high-energy flows and large discharge, and are the main source of mud and debris flows. Heavy precipitation during monsoon and the westerlies triggers many landslides by increasing pore water pressure in unconsolidated sediments. Annual precipitation map of the area was prepared based on Pakistan Meteorological Department (PMD) data (Fig. 5.3a). A strong association between precipitation and mass movements along the Highway has been found (Fig. 5.3b ) (Casagli et al., 2017). Peaks in mass movements curve is clearly synchronizing with high precipitation in respective months (Fig. 5.3b). A large number of landslides along the KKH occurred in 1999 leading to traffic blockade. Precipitation map was then overlaid to landslide events (1999). A large number of landslide events were found in south of Sazin with annual precipitation more than 1000 mm/y. Section of the KKH in East of Sazin contained comparatively less landslides due to its location in semi-arid to arid climate zones (>250 mm/y). Similar control of rainfall over landslide events has also been found in rest of the KKH.

Land Cover

Variations in land cover control the spatial distribution of landslides along with other conditioning parameters (lithology, seismology, slope geometry) (Malek et al., 2015). Changes in land cover influence the hydrological condition of the slopes, leading to slope instability. Generally, vegetation tends to resist the erosion process whereas bare rock or soil is more susceptible to slope failure (Reichenbach et al., 2018). Restrepo and Alvarez (2006) found a strong relationship between land use and landslide events. Previous experts used a variety of softwares and techniques to produce a land cover map from satellite imagery. Many of them used Maximum Likelihood (ML) supervised classification tool on GIS(Ahmad and Quegan, 2012; Butt et al., 2015; Escape et al., 2013; Pourghasemi et al., 2012; Reis, 2008; Rwanga and Ndambuki, 2017; Ulbricht et al., 1993). All land cover maps produced by this technique had an accuracy more than 80%. Optical images of Landsat 8 (19-21 November 2017) were downloaded from USGS database. These images were then ortho-rectified and atmospherically corrected by using Semi-Automatic Classification Plugin (SCP) of QGIS 2.18. Composite images were classified by using GIS’s Maximum Likelihood (ML) supervised classification tool. Training data and spectral signature file was created to represent uniform four classes (Vegetation, Water bodies, Snow and Bare

67

Landslide Susceptibility Mapping by using GIS ______rock/soil). Produced map was divided into four classes: Vegetation, Water bodies, Snow and Bare rock/soil. The final land cover map was assessed using the confusion matrix method. Randomly distributed test pixels for each class were taken on the same image that had previously been classified. The accuracy of the land cover map added up to 87%. Due to variations in the mountain ecosystem from Hassan Abdal to Khunjab pass, the KKH is surrounded by vegetation, bare rock/soil, water bodies and ice/snow covered slopes. The section of the Highway between Hassan Abdal and Thakot is heavily vegetated due to considerable mean annual rainfall. From Thakot to Sazin, slopes are sparsely vegetated. From Sazin onward, barren rock slopes characterize the area.

In the end, spatial density analysis was performed to check influence of land cover changes on landslide events. Results revealed that more than 50% of landslide were located in bare rock/soil category whereas vegetation and snow-covered areas contain 23% each. Processed satellite images were captured at the start of winter season (19-21 November 2016). Most of the slopes in North of Gilgit were covered by snow at this specific time. It justifies presence of 23 % landslides in snow/glacial ice class.

5.7.3 Landslide Susceptibility Map

The produced landslide susceptibility map was classified in four classes: low susceptibility, moderate susceptibility, high susceptibility and very high susceptibility (Fig. 5.9 and 5.10). Nine (9) susceptibility levels were converted into four equally with interval of two except high susceptibility, which contains susceptibility levels of 5, 6 and 7. It was done to distinguish the locations that are more hazardous. Areas of 49.9% and 10.4% of the classified map respectively belong to the high susceptibility and very high susceptibility classes, particularly owed to the presence of active faults, seismic zones and steep slopes (Table 5.7). 34.1% and 5.4% of the Highway fall into intermediate and low susceptibility areas, respectively.

68

Landslide Susceptibility Mapping by using GIS ______

Table 5.7: Area of Susceptibility Classes

Classes Area (%) Low Risk 5.4 Intermediate Risk 34.1 High Risk 49.9 Very High Risk 10.4

Owed to lucidity and space reasons, the final version of the map was divided into two parts: Sections (1) Hassan Abdal-Chilas section and (2) Chilas-Khunjerab Pass section (Fig. 5.9 and Fig. 5.10). The Highway section 1 until Thakot, is characterized by broad valleys and gentle slopes covered by vegetation and therefore falls into the low to intermediate susceptibility zones (Fig 9). Contrastingly, the following section north of Thakot, particularly close to Jijal, lies in high and very high susceptibility zones (Fig. 5.9a). Threads arise from the presence of the southern suture (MMT), poor rock mass quality, the active IKSZ and steep slopes. In the section from Pattan to Sazin, more than half of the Highway is at high susceptibility and very high susceptibility (Fig. 5.9b and 9c). This is because multiple shear zones (KJS) cross the highway between Pattan and Dassu and the surroundings of Sazin fall into the reach of Kamila strike-slip fault (KSF) and the active HSZ (Fig. 5.9c). Two locations close to drainage features (Samar and Harbon Nala) near Sazin (Fig. 5.9c) also fall in the very high susceptibility zone. The second section starts in Chilas and ends at the Khunjerab Pass (Chinese border). Some parts of the Highway were found at very high susceptibility (Fig. 5.10). The Raikot bridge section is the most dangerous part of the Highway as it lies directly over the active Raikot Fault (RF) and passes through RSSZ. Steep slopes and continuous erosion of slope toes by Indus River are aggravating the situation (Fig. 5.10a). Due to presence of the Northern suture (MKT), loose glacial deposits and steep slopes in the Hunza section, some locations of the Highway are declared very high susceptibility zones (Fig. 5.10b). Also north of Sost, two locations (Kafir Pahar and Notorious Killing zone) were found in very high susceptibility zones (Fig. 5.10c).

69

Landslide Susceptibility Mapping by using GIS ______

Figure 5.9: Landslide Susceptibility Map (Abbottabad-Chilas): (a) Jijal Section (area in box “X” is shown in Fig. 5.13) (b) Dassu Section (c) Sazin Section.

70

Landslide Susceptibility Mapping by using GIS ______

Figure 5.10: Landslide Susceptibility Map (Chilas-Khunjerab Pass): (a) Raikot Bridge Section (box “Y” is shown in Fig. 5.14) (b) Attabad Section (box “Z” is shown in Fig. 5.15) (c) Sost Section.

71

Landslide Susceptibility Mapping by using GIS ______

Figure 5.11: Example of landslide events (Casagli et al., 2017) (locations of the photos are given in Fig. 5.9 & 5.10).

72

Landslide Susceptibility Mapping by using GIS ______

Accuracy assessment Accuracy assessment of the map is an essential component of the whole process. In the past, different statistical techniques have been employed to check the predictive ability of a landslide susceptibility map: Predication Rate Curve (PRC), Landslide Density Analysis (LDA), Receiver Operating Curve (ROC) and Area Under Curve (AUC) (Deng et al., 2017). ROC is a better choice than other techniques as it is threshold-independent and measures both accuracy and error rate (Fawcett, 2006; Vakhshoori and Zare, 2018). Multiple researchers used ROC for validation of produced maps (Ahmed, 2015; Basharat et al., 2016; Lee, 2005; Zhou et al., 2016). In this study, we also used ROC and LDA to estimate the predictive accuracy of the map. In the first step, map classes were compared with landslide densities in their respective classes. Spatial analysis of the map and landslide events was performed on GIS using the tabulate area tool. According to the obtained results, most of the landslide events were found in high and very high susceptibility areas and very few landslides were present in moderate and low susceptibility zones respectively (Table 5.8). These statistics confirm a strong connection between susceptibility zonation and landslide events. Thus, this assessment indicates an adequate accuracy of the map.

Table 5.8: Areas of susceptibility level of map and observed Landslides

Susceptibility Level Area (km2) Landslides

1 1.3 0

2 342.1 0

3 627.2 0

4 1517.1 1

5 1819.0 8

6 1316.1 29

7 585.3 20

8 67.4 13

9 0.2 1

In the second step, we used ROC for validation and accuracy assessment of the map, following the example of previous (Basharat et al., 2016; Brenning, 2005; Deng et al., 2017; Pradhan et al.,

73

Landslide Susceptibility Mapping by using GIS ______

2010b; Shahabi and Hashim, 2015). ROC is the product of a graphical plot opposing true positive rate (TPR) and false positive rate (FPR) (Fig. 5.12). TPR indicates the correctly predicted events and is plotted on the y-axis while FPR indicates falsely predicated events, and is plotted against the x-axis. Area under curve (AUC) in a graphical plot explains the efficiency of the model. AUC may range from 0.5 to 1 in different cases depending on the accuracy of model. A value close to 0.5 indicates random results while values close to 1 indicate a perfect model (Ahmed, 2015). In this study, we used 72 landslide locations to validate the final version of the map. AUC was found 0.72 indicating a reputable accuracy (72%) of the map (Basharat et al., 2016; Deng et al., 2017).

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3

True Positive TruePositive (TPR)Rate 0.2 0.1 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 False Positive Rate (FPR)

Figure 5.12: ROC based accuracy assessment of the landslide susceptibility map.

Discussion

In this study, AHP based weighted overlay technique has been used to produce landslide susceptibility map along the KKH. Ten landslide controlling factors were considered for production of landslide susceptibility map. Review of previous published articles helped us to finalize causative factors, responsible for occurrence of landslides. Afterwards, spatial analysis was performed to prioritize and rate these parameters. The clustering of landslides in active fault and shear zones indicate their strong control. These tectonic features resulted into highly fractured and jointed rock mass, highly susceptible for failure. Slope gradient and distance from a fault were considered basic condition for slope instability and rated as the most important factor. Major seismic events in active seismic zones in close vicinity of the Highway, triggered landslides in the past. Some lithologies (slates, shales, quaternary) exhibit low shear strength and makes slope

74

Landslide Susceptibility Mapping by using GIS ______susceptible to failure. Therefore, seismicity and lithology are rated high but less important than slope gradient. Torrential Rainfall in Monsoon and Westerlies season triggers landslides and was taken into account. However, its influence decreases in north of Gilgit, where cumulative effect of ice melt with rise in temperature and rainfall cause landslides. Other geomorphic factors (elevation, aspect, curvature) were rated low because of their poor association with landslide events. Quality of the data sets directly influences the quality of results (Fressard et al., 2014). Landslide inventory is the most important and basic one among all them. Inaccuracies regarding location and recent activity of landslides adversely affects accuracy of the final map. We prepared and used landslide inventory by using satellite imagery, landslide activity log encompassing over ten years (Fig. 5.6) and then field surveys to validate locations and extent of landslides. The combination of all these aspect has led into flawless inventory. We used geological maps (1:50000 and 1:250000) (Khan et al., 2000) explaining lithological variations within formation and also having both regional and local faults. Lithological contacts and location of faults and shear zones were verified during field visit. Seismic intensities and PGA (Peak Ground Acceleration) values were derived from shake map of instrumental earthquakes (USGS). Rainfall data of six uniformly distributed weather stations was used to prepare annual rainfall map. Land cover map was prepared from Landsat 8 optical imagery having 30-meter resolution. Images captured in November 2017 (before winter) were used to have snow free slopes along the KKH. Keeping in view the objective to produce landslide susceptibility map along the Highway, four classes (Vegetation, Water bodies, Snow and Bare rock/soil) were derived and temporal variability in landuse was not considered. In this study, quality of the data sets was comparatively better than previously used. Many authors used AHP based model to prepare susceptibility map (Ahmed, 2015; Arizapa et al., 2015; Bachri and Shresta, 2010; Basharat et al., 2016; Intarawichian and Dasananda, 2010; Kamp et al., 2008; Komac, 2006; Park et al., 2013; Pourghasemi et al., 2012, 2016; Pourghasemi and Rossi, 2017; Rahim et al., 2018; Rozos et al., 2011; Shahabi and Hashim, 2015; Yalcin, 2008). Comparison of AHP based model with other models in some studies (Pourghasemi et al., 2016, 2012; Pourghasemi and Rossi, 2017; Shahabi and Hashim, 2015; Yalcin, 2008) proved former more accurate and precise. Accuracy of the produced map in this study is 72% which is satisfactory and but slightly less than previous studies (Basharat et al., 2016; Kanwal et al., 2016; Pourghasemi et al., 2012; Pourghasemi and Rossi, 2017; Shahabi and Hashim, 2015; Yalcin, 2008). AHP is a simple and easy way to rate different parameters consistently. Value of CR remained below 0.10 for each case, indicating appropriate and reliable weighting criteria.

75

Landslide Susceptibility Mapping by using GIS ______

However, AHP based model has been criticized due to its expert opinion based subjective approach. Therefore, we did spatial analysis to rate all controlling parameters to minimize chances of errors related to cognitive limitations of expert. Further, due to low spatial resolution of DEM (30m x 30m), some of cut slopes along the KKH were neglected. Rating system introduced in this study may not fit into any other regions due to variations in geological, seismic, hydrological and other controlling parameters. Lastly, change in existing condition (undercutting of landslides toes for Highway expansion) of landslide controlling parameters may change present susceptibility along the Highway.

5.8.1 Case study

To supplement results and final landslide susceptibility map, three sub-sections were discussed: Jijal sub-section, Raikot Bridge sub-section and Attabad sub-section. Jijal sub-section Part of the Highway N of Jijal town lies in a zone of very high susceptibility (Box of Fig. 5.9a). It comprises highly fragmented ultramafic rocks of the Jijal Complex. Due to its position in the hanging wall of MMT, it is highly jointed and locally sheared. The area is seismically active and located just three kilometers away from the epicentre of the Pattan Earthquake on 28 December, 1974 (M = 6.2, D = 22 km). The seismic intensity (Modified Mercalli Intensity Scale) of this event along this part of the Highway reached VIII (Ambraseys et al 1981). Furthermore, the S of this area is seismically very active (Fig. 5.4c). During the catastrophic October 2005 Kashmir Earthquake (M=7.6), some landslides were reactivated, leading to a closure of the Highway. Topography in this part is steep, with slope angles ranging between 40o and 70o. The area lies in the Monsoon region where average annual rainfall exceeds 1000 mm. The dotted yellow lines (Fig. 5.13a) indicate a big catchment area (1.34 km2), capable of collecting large amounts of water during rainfall, leading to debris flows and debris slides in sheared and highly fragmented rock mass. Rock and debris falls are further promoted by clayey soils that form in joint apertures as a result of serpentinization. Due to heavy rainfall (617 mm) in March and April 2016, a large number of landslides was reactivated leading to blockage of the Highway for two weeks. All these factors (closeness to fault, high seismicity, fragmented rock mass, heavy rainfall, steep topography) are responsible for the very high landslide susceptibility in this area.

76

Landslide Susceptibility Mapping by using GIS ______

Figure 5.13: Very high landslide susceptibility along the KKH near Jijal (Google Earth, 2017, Bing Maps) (for location see Box “X” in Fig. 5.9a). (a) Overview of 7.5 km long small section of the Highway in NE of Jijal: Dotted black line represents MMT; dotted yellow line marks the boundary of catchment area (1.34 km2); Pink arrows showing ongoing rock/debris falls, which travel downslope along with water during heavy rainfall; gully erosion is prominent. (b) Famous “Shaitan Pari Slide” with partially damaged retaining wall. Reactivation of this slide is mostly during heavy rainfall. (c) Highly jointed rock mass is highly susceptible to rock/debris fall.

77

Landslide Susceptibility Mapping by using GIS ______

Raikot Bridge sub-section Stability of the Highway is a challenge for geologists, civil engineers and Highway authorities. The Highway subsidence due river undercutting and presence of active Raikot Fault (RF) is a continuous phenomenon. Hot water springs and a shear zone (c. 125 m wide) indicate the presence of RF. It marks the boundary between Precambrian granitic gneisses of the Indian plate and batholiths and gabronorites of the Kohistan Island Arc. RF is responsible for shallow seismicity along the Highway (Fig. 5.4c). Seismic intensity (Modified Mercalli Intensity Scale) in this part reaches VI. The rock mass is highly jointed and sheared due to presence of the fault. Furthermore, the continuous seepage from hot water springs results in weathering and a lower shear strength of the rock mass. In the past, two large landslides dammed Indus River (Fig. 5.14a). Deposits of these landslides contain retrogressive slope movements and debris flows (Fig. 5.14a). Landslide damming had several effects on terraces and slopes, including the deposition of alluvial and lake deposits. In addition to this, continuous rockfall is adding large quantities of debris to the slopes. Topographically, also this section is very steep. Climatically, it lies in a semi- arid to arid zone with an average annual rainfall of 0 – 250 mm. Rainfall, however, is restricted to a couple of events per year. On 3 and 4 April, 2016, 105 mm rainfall reactivated debris flows and slides. The prevailing circumstances make this part of the Highway highly susceptible to landslides.

78

Landslide Susceptibility Mapping by using GIS ______

Figure 5.14: Very high landslide susceptibility near Raikot Bridge (Google Earth, 2017, Bing Maps) (for location see Box “Y” in Fig. 5.10a) (a) Overview of 13.4 Km long small section in West of the Raikot Bridge: White lines represent main scarps of large landslides/rock avalanches, which dammed the Indus River past; dotted yellow circles represent deposits of these old landslides; white arrows are showing sagging in landslide deposit which is due to retrogressive rotational failure (b) White lines represents scarp of shallow landslide in alluvial deposits (area in red box of a); area in dotted yellow circles represents ongoing rock/debris fall supplying scree/talus for debris flow during rainy season (c) One of the hot water springs (90oC-96oC) along the Highway in this section (d) Another view of small section: white arrows are marking upper limit of the shear zone (c. 125 m wide); dotted black representing active RF marked along shear zone; overhangs above and river erosion below the Highway makes this section highly susceptible to slope failures.

79

Landslide Susceptibility Mapping by using GIS ______

Attabad sub-section Hunza section has a variety of slope failures depending upon prevailing geological and climatic conditions. However, the area shown in Fig. 5.15 is characterized by falls (rock, debris) and some slides (rock, debris). Historically, large landslides (1858, 1980, and 2010) dammed Hunza River in this part (Fig. 5.15a). The area is characterized by highly weathered and jointed granodiorites, orthogneisses and pegmatitic veins of the Kohistan Batholith. The orientation of joints (dipping in the same direction as the slope) has an adverse impact on slope stability. The area is located in the hanging wall of MKT, the main fault in this region, and a local fault exists in close vicinity (Fig. 5.10a). Past earthquakes (Astore, 2002; M=6.3, Muzaffarabad, 2005; M=7.6) produced ground shaking intensity up to V-VI (Modified Mercalli Intensity Scale). Climatically, the valley floor is part of a semi-arid zone (250 mm/yr – 500 mm/yr) while the higher slopes and peaks (> 5000 m) receive precipitation ≥ 1000 mm/yr. Therefore, the area is sparsely vegetated. Rainfall coupled with snowmelt in early spring reactivates old landslides (Ali et al., 2017). Undercutting of landslide toes by Hunza River below the Highway and rock/debris fall upslope are major concerns in this section. All of the above mentioned circumstances yield a high landslide susceptibility for this section (Fig. 5.10)

80

Landslide Susceptibility Mapping by using GIS ______

Figure 5.15: High landslide susceptibility along the Highway in Hunza Valley (Google Earth, 2017, Bing Maps) (for location see Box “Z” in Fig. 5.10b) (a) Overview of 11 km long small section in west of the Attabad Lake: white lines are representing the main scarps of old large landslides/rock avalanches which dammed the Hunza River in past; dotted yellow circles represent deposits of these old landslides; white arrows are showing location of local fault (b) Attabad landslide (in red box of a); red arrows are showing the Highway submerged in lake water.

81

Landslide Susceptibility Mapping by using GIS ______

Conclusions

A set of Landslide Susceptibility Maps of the KKH (CPEC) was prepared using GIS involving multiple techniques: literature review, remote sensing, field surveys. Ten controlling parameters (lithology, seismicity, rainfall intensity, distance from fault, elevation, slope angle, aspect, curvature, land cover and hydrology) were considered of which each one was assigned a numerical weight using AHP. Thematic layers of these parameters were then overplayed using the WOL tool of GIS. Four different classes of landslide susceptibility were then applied to the final map: low susceptibility, moderate susceptibility, high susceptibility and very high Susceptibility. 10 % and 50% of the Highway were found in very high and high susceptibility zones, respectively. Active faults (MMT, KJS, KSF, RF, MKT, KF), seismic zones (IKSZ, HSZ, RSSZ) and steep slopes are responsible for the associated susceptibility in these areas. The Highway is characterized by a variety of mass movements: rockfall, debris fall, rockslide, debris slide, debris flows and mudflows. The threatened sections are especially unstable in Monsoon and Westerlies seasons every year. Altogether, a detailed investigation is inevitable to enable hazard free and safe traveling. About 40% of the Highway lies in low and moderate susceptibility zones, which remain almost stable throughout the year. The Highway sections from Hassan Abdal to Thakot, near Chilas, Gilgit and Sost fall into these moderate and low susceptibility zones and are quite stable due to their course in broad U-shaped valleys with gentle slopes. Although the part of the Highway between Hassan Abdal and Thakot receives heavy precipitation in Monsoon season, the area is stable owed to mature geomorphology. Due to closeness with MMT, higher seismic intensity and steep topography, the KKH near Raikot Bridge and Jijal was found at very risk. Furthermore, extreme weather conditions, highly shattered and weathered rock mass, active faults and long steep slopes are responsible for very high and high susceptibility around Attabad, notorious Killing Zone and Kafir Pahar sites. According to results, active faults, slope gradient, seismicity and lithology have a strong influence on landslide events along the Highway. In the final step, the predictive accuracy of the map was determined by using LDA and ROC. The accuracy of the map was rated to a satisfactory 72%, which is suitable for mitigation planning.

82

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

6 Empirical Assessment of Rockfall and Debris Flow Risk along the Besham- Chilas Section of the KKH

Abstract

The Karakoram Highway links north Pakistan with southwest China. It passes through unique geomorphological, geological and tectonic setting. This study focused 200 km long section of the Highway starting from Besham until Chilas. Landslides are frequent are mostly triggered by torrential rain during Monsoon and Westerlies, leading to Highway blockade. Rockfall and debris flow are prime mode of slope failures. Regional to site specific approach was implemented to assess risk associated to these two modes. Remote sensing based techniques were used to identify potential hazardous sites, which were further investigated for risk assessment. Modified Pierson’s Rockfall Hazard Rating System (RHRS) rated potential rockfalls whereas semi- quantitative technique was employed to assess debris flows. Normalized scores of each site shaped the final map, further classified into four zones: very high, high, intermediate and low risk. Introduction

The Karakoram Highway (KKH) is a lifeline of North Pakistan, more specifically of the Gilgit Baltistan. It is also an important trade route between Pakistan and China (Fig.6.1a). It passes through one of the most challenging and rugged terrain of the world. Stability of the 840 km long KKH is in question since its completion in 1979 (Ali et al., 2018b; Casagli et al., 2017). Its construction by aggressive uncontrolled blasting resulted into large number of landslides. Variation in geological, seismological and atmospheric environments make it a unique area to study and investigate landslide hazard. Interruption of traffic along the KKH, due to landslides, during rainy seasons (Monsoon and Westerlies) is a regular phenomenon. Additionally, it results into life losses and economic losses (Fig. 6.2). Depending upon magnitude of the landslide, traffic blockade may extend to a couple of weeks. On April 2, 2016, due to torrential rainfall, landslides at more than hundred locations blocked the KKH for more than two weeks and killed 16 people (Petley, 2016). Furthermore, landslides at Chuchang and Kiyal Bridge completely eroded the

83

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Highway. Previously most of the research work along the KKH was regional landslide susceptibility mapping based on remote sensing techniques (Kamp et al., 2008; Ahmed et al., 2014; Kanwal et al., 2016; Basharat et al., 2016; Bacha et al., 2018; Khan et al., 2018; Rahim et al., 2018; Ali et. al 2019). In this paper, an attempt is made to assess risk associated to landslide along 200 km long Besham-Chilas section of the KKH (Fig. 6.1). Rockfall, debris slide and debris flow are dominant mode of slope failures in the section.

Risk associated with these processes is defined as the probability of their occurrence and capacity to damage exposed elements (Mineo et al., 2018). Previous researchers used empirical and semi- empirical techniques to evaluate hazard and risk associated with above landslide processes (Capra et al., 2002; Hürlimann et al., 2006; Gentile et al., 2008; Liu et al., 2009; Baumann et al., 2010; Lari et al., 2011; Santi et al., 2011; Wrachien and Mambretti, 2011; Chen et al., 2012; Liang et al., 2012; Cui et al., 2013; Ma et al., 2013; Rickenmann, 2016; Cao et al., 2016). One element is common that all techniques are based on factors responsible for occurrence of landslide and capacity to damage exposed elements. Factors controlling occurrence and its probability are different for each landslide process (rockfall and debris flow). Unique and relevant risk assessment methods are necessary for each landslide process. Therefore, primary knowledge about each failure mode is essential to define, design and implement risk assessment methods.

Extremely rapid downslope movement of a detached block by free fall, rolling, leaping and bounding is known as rockfall (Varnes, 1978a). Frequency of rockfall is high along roads and highways, which have been excavated along steep cliffs. Presence of faulted and sheared massive ultramafics, mafics and batholiths made this section highly susceptible to rockfalls.

84

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Figure 6.1: Overview of study area (a) Location of the study area (b) Overview of rainfall conditions and tectonics (c) Geology of the study area: DF-Debris Flow, DS-Debris Slide, RF-Rockfall, CC-Chilas Complex, KA-Kamila Amphibolites, JC-Jijal Complex, GGD- Gabbronorites, Gabbros and diorites, D-Diorites, SG-Sheared Gabbronorites, AB- Amphibolites Banded, AMS-Sheared Amphibolites, G-Garnet Gabbros, P-Pyroxenites, Du- Dunites, PS-Pyroxenites and Serpentinites, BG-Besham Group, PG-Porphyritic Granites, Q-Quaternary, MMT-Main Mantle Thrust, KaF-Karkoram Fault, KF-Kamila Fault, KoF- Kohistan Fault, KJS-Kamila Jal Shear Zone, MKT-Main Karkoram Thrust, MBT-Main Boundary Thrust. Red circles and dotted yellow rectangle showing locations of Fig. 6.3a, 6.3b, 6.3c and 6.3d.

85

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Previously, Event Trees (Peila and Guardini, 2008; Budetta et al., 2016; Mineo et al., 2017) and rating approach (Budetta, 2004; Ferlisi et al., 2012; Regmi et al., 2016) has been widely used for rockfall risk assessment. Availability of relevant data regarding rockfall events and objective of the study influence the choice of method.

In this case study, the documentation of historical events is very poor and is not enough to determine recurrence of rockfall. Therefore, we did not employed statistical method and used modified Rockfall Hazard Rating System initially developed by Pierson et al. (1993). We used previously developed susceptibility map (Ali et al., 2018a), limited historical landslide inventory, slope angle distribution (SAD) and runout assessment to zoom into potential rockfall sites.

In addition to rockfalls, a large number of debris slides and flows characterizes this section of the KKH. Debris slide and flow are two different slope failures but their progressions have slight commonality (Sletten et al., 2015). Presence of collapsed material in terms of talus, scree or moraine deposits on steep slopes is highly susceptible to erosion. Melting of snow coupled with precipitation triggers debris slides and flows. Initial rise in pore water pressure leads into debris slide and further upsurge in water content leads into flow of collapsed material into a channel. Sometimes, light rain removes fine particles from collapsed material leaving behind coarser one with low shear strength. Risk assessment of debris slide and flow involves identification of parameters essential for their formation and initiation. Previous researchers used regional approach involving remote sensing, detailed field survey for simulations and statistical analysis of causative factors for risk assessment of debris slide and flows ( Capra et al., 2002; Hürlimann et al., 2006; Gentile et al., 2008; Liu et al., 2009; Baumann et al., 2010; Santi et al., 2011; Wrachien and Mambretti, 2011; Lari et al., 2011; Liang et al., 2012; Chen et al., 2012; Cui et al., 2013; Ma et al., 2013; Rickenmann, 2016;). In this study, maximum probability of debris flow was calculated by using predefined sources on FLOW R. Afterwards, detailed field survey was conducted to rate indexes responsible for its formation and initiation.

86

Figure 6.2: Impact of rockfalls and debris flows on traffic and travelers: (a) Rockfall near Dassu crashes cars. Yellow arrows are indicating crashed cars (KKH Updates 2016). (b) Debris flow trapped a car along the KKH (c) Small debris flow blocked the Highway for couple of hours and two yellow lines are showing long line of vehicles on one side of deposit.

87

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______General Situation of the Study Area

6.3.1 Location and weather conditions

The study area is a section of the KKH, located in North Pakistan (Fig. 6.1b). It starts from town of Besham and ends at Chilas. The Jijal, Pattan,Dassu and Sazin are important towns located along the section of the Highway. It runs through deep gorges and along the Indus River. From Besham onwards, it follows right bank of the river until Dassu where it crosses the river and then runs along left bank. The section is characterized by steep (45o-80o) and long slopes (2~3km). The study area receives heavy rainfall in two seasons: monsoon and westerlies. However, weather conditions along the section are not uniform. Area in North of Besham until Sazin, receives heavy rainfall (1000-1500mm/y) whereas, east of Sazin until Chilas lies in semi-arid to arid climate (0-250mm/y). In April 2016, due to torrential rainfall in March and April (281.8 and 335 respectively) triggered multiple landslides leading to blockage of the KKH for more than two weeks.

6.3.2 Geology, geomorphology, tectonics and seismology

Southern part is located at northern part of Indian plate whereas northern is situated at central part of Kohistan Island Arc (KIA). It demonstrates a wide range of lithology, different tectonic environments and seismic zones. It runs through cliff forming granitic gneisses of Indian plate, mafic and ultramafics of KIA. Slopes in study area are composed of thirteen (13) lithological units (Fig. 6.1c). Four important tectonic features crosses or lie in close vicinity of the KKH in this section: Main Mantle Thrust (MMT), Pattan Fault (PF), Kamila-Jal Shear Zone (KJS) and Kamila Fault (KF). Rock mass is closely jointed and highly fractured close to these tectonics features (Casagli et al., 2017). Due to position at hanging wall of MMT, dunnites, pyroxenites and serpentinites of Jijal Complex are closely jointed and sheared at places. Therefore, it exhibits a large of number of slope failure. Differential erosion in massive and sheared Kamila amphibolite is responsible for dangerous rockfalls. Gabbronorites and diorites of Chilas complex contains stress release joints with short persistence leading to block formation of size more than 6m3 (Ali et al., 2018b). Furthermore, ongoing rockfall process, has led to formation huge talus deposits on slopes. These talus deposits are highly susceptible to erosion during heavy rainfall and snowmelt at the start of spring season.

Seismically, the area is part of two active seismic zones: the Indus Kohistan Seismic Zone (IKSZ) and the Harman Seismic Zone (HSZ) (Ali et al., 2018a). Former comprised of two events (Pattan

88

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

1974: M=6.2 and Kashmir Earthquake: M=7.6) which severely damaged the Highway (Ambraseys et al., 1981). Close to Jijal area, maximum seismic intensity of Pattan Earthquake was VIII with PGA value of 64.3% g (USGS) (Fig. 6.3). HSZ also contains instrumental seismicity with one major event (1981: M=6.2) (USGS).

The studied part of the Highway passes through steep sided v-shaped valley. Valley widens at some places (Pattan and Sazin). After Sazin, it narrows down into steep sided v-shaped valley again. Further, it enters into “Chilas Plains” with valley width between 0.5 to 2 km (Kibria and Masud, 2006). Slope angle varies between 40o and 85o, depending upon lithology and structural condition of the slope. Sub-vertical to vertical cut slopes and overhangs are important character of the section. Identification of potential hazardous sites (Regional)

The methodology for hazard and risk assessment of rockfalls and debris flows along the section comprised of two steps: (1) Identification of potential slope failures by using regional approach (2) Field surveys to measure all the parameters for hazard and risk assessment. Previous publications, remote sensing and field data was used to prepare risk assessment map of the studied section. Regional to site specific approach was used.

Remote sensing, previous events based inventory and empirical model (Flow-R) was employed to identify potential rockfalls and debris flows. Landslide inventory gave an overview of rockfalls and debris flows. Slope angle distribution (SAD) explored rockfall source areas. Satellite imagery was used to map collapsed material for debris flows and then Flow-R assessed runout, propagation and probability of both rockfalls and debris flows.

6.4.1 Landslide Inventory

Firstly, multi-temporal, Google Earth 3D images were analysed to mark previous slope failures along the section. Then, published reports (Fayaz et al., 1985; Khan et al., 2003, 1986) and articles (Derbyshire et al., 2001; Hewitt, 2001, 1998; Hewitt et al., 2011; Kibria and Masud, 2006; Santi et al., 2011) were used to prepare landslide inventory map of the section. Secondly, road clearance data for three different periods (1982-82, 1996-2000, 2014-16) was also supplemented to know about the frequency of landslide events (Fig. 6.4). Thirdly, field surveys were conducted to validate and complete the landslide inventory map and to further acquire parameters required for risk assessment (Fig. 6.1c).

89

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Figure 6.3: Identification of potential rockfalls and debris flows: Some parts of the studied section are shown as examples (a) Published susceptibility map (Ali et al. 2018 b), (b) FLOW-R runout assessment for debris flows indicating maximum probability along highly active zone, (c) Slope angle distribution to extract rockfall source areas, (d) Shake map of Pattan Earthquake 1974, explaining seismic intensity variations along the Highway (USGS database 2019).

6.4.2 Slope Angle Distribution (SAD)

A morphological unit tends to have several mean slope angles with least standard deviation (Strahler, 1950). Generally, four types of morphologic units have commonly been discussed (Oppikofer et al., 2007; Loye et al., 2009): Plains, foot slopes, steep slopes and cliffs. Fluvial- glacial deposits constitutes plains, marked by low angle. Alluvial, debris flow and landslide deposits compose foot slopes, with slope angle ranging from 10o to 30o. Whereas scree/talus/till covered, slopes are steeper with angle 30o-50o. Cliffs are the steepest of all morphological units. Slope angles here correspond to steeper than 50o. Slope steepness is considered an important element for rockfall initiation. SAD has been widely used for detection of rockfall source zones.

90

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Guzzetti et al., (2010), Toppe, (1987) and Jaboyedoff and Labiouse, (2003) took into account cliffs a rockfall source with steepness more than 45o, 60o and 37o respectively.

Previously, researchers (Loye et al., 2009; Michoud et al., 2012; Losasso et al., 2016) used DEM for construction of normal distribution of slope angle. Afterwards, it has been decomposed into normal distribution of different topographic units: plains, foot slopes, steep slopes, cliffs etc. Spatial resolution of DEM has a direct impact on quality of curve. We used 12m (Tandem-X) and 30m (ASTER) DEM for construction of slope angle distribution to zoom into steep slope and cliffs. The area studied in this paper has a uniform morphotectonic setting. Therefore, we did not divided it into several parts. Slopes and cliffs with angle more than 55o have been considered a potential rockfall source (Fig. 6.3c).

6.4.3 Material Source of Debris Flow

The origination of debris flow is sum of three basic requirements: slope, water supply and erodible unconsolidated sediments. Presence of collapsed material in terms of unconsolidated soil, moraines, talus, scree etc. in the catchment area is a basic essential of debris flow initiation (Cui et al., 2013). Because of large area of the catchment areas, estimation and measurement of the source material has always been challenge for investigators. However, remote sensing has made easy to calculate area of catchment and collapsed material. Still to find exact volume is a difficult and time taking task. Software based runout assessment utilized these mapped collapsed material for propagation, probability and extent of debris flows. Ratio between these two mapped parameters was further considered for grading debris flow site in terms of hazard and risk assessment.

6.4.4 Runout Assessment

Rockfalls and debris flows are only hazardous when they can reach any sort of infrastructure, in our case it is the Highway. Therefore, runout out assessment has been performed to shortlist potential rockfalls and debris flow with high probability. Flow-R is numerical model based an open source software, initially developed for runout assessment of debris flows at regional level (Horton et al., 2011; Jaboyedoff et al., 2011; Oppikofer et al., 2014). Afterwards, it has been successfully used for rockfalls, avalanches and other modes of mass movements. Its operation capacity with limited data sets and customization according to local morpho-tectonic conditions motivated for its usage.

Runout calculation processes involve two steps: source identification and spreading assessment. Identification of source areas can be calculated by software itself either by using different data

91

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______sets like geological maps, DEM, and other derived parameters or by using user-defined sources. In our case, we use user-defined sources for runout assessment. Source areas for rockfalls and debris flows were identified by SAD and mapping of collapsed material respectively. During identification process, mass of source has not been considered, as it is difficult to calculate its volume for such a big area. Spreading assessment is a combination of two customizable algorithms: flow direction and energy balance responsible for maximum travel distance. Flow direction algorithm determines probability of debris flow direction and distribution from a cell. There is wide range of algorithms (D8, Dx, p8, Holmgren) but Holmgren’s algorithm has been extensively used and have also been preferred in our case. It has ability to be modified from multiple flows to single convergent flow according to local conditions. In the end, areas with high probability are combined. Energy based algorithms determines travel distance of debris flow or rockfall by considering basic energy of a cell. Due to unidentified mass of whole deposit, whole process relies on unit mass.

SRTM DEM and Tandem-x with spatial resolution of 30 m and 12 m have been used for runout assessment. User predefined sources were imported. Flow accumulation threshold was set at 1 ha. and distance from stream with buffer of 10 m. Holmgren’s exponent was customized at 4. Friction loss function was customized with setting of travel angle at 12o and energy limitation of 10 mps. Slope angle and maximum velocity relationship (Horton et al., 2011) determined debris flow’s velocity. Due to variation in slope angle along flow path, mean slope angle was considered. Some of the recent debris flow events in the study area were compared to simulation results with different parameters but earlier discussed were found closer. Detailed Mapping and Risk Assessment (Site Specific)

Potential rockfalls and debris flows, identified during first stage, were surveyed for risk assessment. Due to incomplete information regarding event recurrence, statistical techniques have not been used. Two different empirical methods (for rockfall, debris flow), based on morphological, geological, hydrological criterion were used. Measuring and scoring of some parameters is quite easy while other’s evaluation is difficult and subjective. Exponential scoring based scoring criteria has been used to sharply differentiate sites with very high risk, which is ultimate objective of this study.

6.5.1 Rockfall Hazard Rating System (RHRS)

Several researchers assessed rockfall hazard and risk along highways by using RHRS (Budetta, 2004; Pritchard et al., 2005; F. Guzzetti et al., 2010; Regmi et al., 2016). Eleven parameters,

92

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______grouped into four categories, were utilized for risk assessment: slope morphology, geology, failed rock characteristics and highway related factors (Table. 6.1). Slope height indicates potential energy of falling block and determines its runout extent. Effective slope height from source area was measured. We considered number of joint sets, their attitude, spacing, planes and their ultimate effect on block formation and its failure. Seismicity in close vicinity of the Highway is a regular phenomenon Slopes were rated according to intensity level (MMI), taken from shake maps of instrumental seismic events (USGS database 2019). MMI measures the effects of an earth at a specific site. Block size (volume) and mass of failed rock mass indicate amount of energy and its capacity to damage property. For rockfall history, road clearance logs were acquired from Frontier Works Organization (FWO 2016). Vulnerability, sums vehicle risk and damage to road property, comprised of three parameters: distance between road and slope (ditch effectiveness), effective countermeasures and decision sight distance (DSD). Distance between road and slope can avoid damage and minimize risk by trapping falling rock mass. More the distance is, safer it would be. Highway authorities and planners construct barriers, embankments, meshes etc. to protect road and vehicles. In this rating criteria, presence and effectiveness of these protective measures to stop falling mass to reach the Highway has also been considered. DSD is an actual distance required to make instantaneous decision, in case of obstacle created by rockfall. It is calculated by using following relationship:

퐴푆퐷 Equ. (6.1) 푃퐷푆퐷 = ( ) ∗ 100 퐷푆퐷

Where PSDS is percentage of DSD, ASD is actual sight distance. DSD varies according to maximum speed fixed for that particular highway. In our case, it is 60 km/h and actual sight distance is 160 m.

Kinematic Analysis Kinematic analysis is an important technique to know possible rockfall process based on joints and slope orientation. Field survey were conducted to acquire joint data (orientation, spacing, aperture, filling, persistence) from rockfall sites. Then, kinematic analysis was carried out by using DIPS 7.0 software, to interpret joint data to know possible occurrence (Table 6.2). Most of the landslides including rockfalls, took place on slope with angle more than 30o and was considered as friction angle in this area. Daylight envelop and friction cone were drawn to evaluate number of critical poles. Intersections between circle representing slope face and cone of friction are critical (Fig. 6.4a) (Hoek and Bray, 1981). Possibility of three possible failure modes (planer,

93

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______toppling, wedging) has been assessed. In the end, whole information was supplemented with RHRS.

Table 6.1: Rockfall Hazard Rating System (Modified after Pierson 1993) for risk assessment.

Score 3 Points 9 Points 27 Points 81 Points

Slope Morphology

Slope Height (m) <7.62 15.24 22.84 30.48

Geology

2 sets and 3 sets and 2 sets and Number of joint sets 3 sets and more less more less Adverse and their effect on Adverse No adverse No adverse orientation slope orientation exists orientation orientation exists

Joint Spacing (m) <0.2 0.2-.5 0.5-1 >1

Joint Planes Rough Undulating Planer Clay Filling

Hazard Seismic Intensity <6 6-7 7-8 >8

Failed Rock Characteristics

Size of Single Block 0-0.25 0.25-0.5 0.5-1 >1 (m3)

Occasional Rockfall History Few Falls Many Falls Constant Falls Falls

Failed Mass at slope Well-developed No material A few Scattered pile toe pile

Percentage of decision sight 80-100 60-80 40-60 <40 distance (%)

Good Moderate Limited Ditch Effectiveness No Catchment Catchment Catchment Catchment

Vulnerability Fully Partially No Protection measures Ineffective Effective Effective Countermeasures

94

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

6.5.2 Debris-Flow Risk Assessment

Risk is a combination of hazard and vulnerability. In our case, vulnerability is highway’s blockade, burial, erosion and traffic disruption directly affected by debris flows. All rating parameters were based on three main debris flow processes: initiation, transportation and deposition (Table. 6.2). Rainfall water mobilizes debris downslope and therefore, maximum rainfall for twenty-four hours was considered for risk assessment. Stream steepness reflects its capacity to mobilize material downslope. It has been classified into three categories. Availability of debris and enough water supply in catchment area is a basic requirement for debris flow initiation. Quantity of collapsed material in terms of debris has a direct relation with magnitude and spreading of flow. Proportion of area covered by collapsed material (debris, scree, talus, moraines) to total catchment area has been calculated by satellite imagery. Furthermore, potential and recent slope failures were identified in catchment area to determine which can be source of debris flow. Vegetation effect debris flow phenomenon by hindering or decelerating it. Therefore, its subjective evaluation was executed. Debris flow history, disaster mode and countermeasures to avoid disaster were taken into account for vulnerability assessment. Road clearance logs, field surveys and local community’s opinion was taken into account to know history of previous events.

Results and Discussion

6.6.1 Analysis of Landslide Inventory

According to road clearance logs, mass movements interrupted traffic flow along this particular section (200 km) for two hundred and twenty six (226) times at one hundred and four (104) locations during a period of four years (August 1996 – July 2000) (Khan et al., 2003). The mass movements are mainly rockfalls (47), debris flows (26), debris slides (26) and historical deep- seated rockslides (5). Size of these landslides varies from minimum ninety square meters (shallow) to maximum ten square kilometers (deep seated). After interpretation of satellite imagery, previous publications and field surveys, rockfalls and debris flow along this section have been found problematic and were investigated in this study. Due to uncontrolled blasting, the highly fragmented Jijal Complex and massive amphibolites has a large number of rockfalls along cut slopes. The Pattan earthquake triggered many rockfalls leading to severe damage on these two locations. Near Kandia valley, short persistence stress release joints in massive diorite of Chilas complex led into formation of large blocks (>6m3).

95

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Table 6.2: Structural data and kinematic analysis of rockfalls, for location see Fig. 6.6 and 6.7.

(A) (B) (C) (D)

(E) (F) (G) (H)

(I) (J) (K) (L)

(M) (N) (O) (P)

It is sometimes impossible to remove these large blocks from the Highway without blasting. Debris flows along Jijal-Pattan and Sazin-Chilas subsections are common. Presence of active faults (MMT and Pattan Fault) in former sub-section made debris available and torrential rains in Monsoon transport it. Whereas, later has sparsely vegetated relatively large catchment areas.

96

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Table 6.3: Rating criteria for debris flows risk assessment

Score 3 Points 9 Points 27 Points

24h maximum rainfall (mm) <30 30-80 >80

o o o o Maximum Stream Steepness <10 10 -30 >30

Area with slope gradient o 2 <0.08 0.08-0.20 >0.20 greater than 30 (km )

Seepage Dry Dripping Flowing

Slope Failures in Catchment Old scarps & Potential Active & Potential No Area failures failures

Hazard

Thickly Sparsely vegetated/No Area covered by Vegetation Thinly vegetated vegetated vegetation

Massive rock Rock mass Characteristics Banded rock mass Sheared rock mass mass

Proportion of collapsed <0.1 0.1-0.3 >0.3 material to catchment area

Past events without traffic Past events with traffic Debris Flow History No Past events disruption disruption

Effective Ineffective Countermeasures No countermeasures countermeasure Countermeasures

Vulnerability Damage of Debris flooding on Potential disaster Mode Outflow of embankment bridge road/Damage of road

97

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______Risk Analysis of Rockfalls

Potential rockfall sites identified after landslide inventory analysis and runout assessment by empirical model Flow-R were surveyed in detail. Keeping in mind controlling parameters in our study area, modified Pierson’s RHRS was applied (Table 6.1). Four rockfall sites (Jijal, Chochang, Lotar and Summar nala) are discussed as examples.

6.7.1 Jijal Rockfall

This rockfall is along a cut slope and active since construction of the Highway. It located in close vicinity of Jijal (Fig. 6.6a). Being part of hanging wall of MMT, it is highly jointed. Furthermore, excavation by uncontrolled blasting aggravated the situation. The Pattan and Muzaffarabad earthquake triggered rockfall leading to complete destruction during former event. This site had high peak ground acceleration (PGA) value (64.73%; USGS database 2019) during this event with intensity (MMI) of seven. Effective slope is 65 meters high and comprised of pyroxenite and serpentinite. It is characterized by three joint sets with open apertures and an adverse impact on slope stability. Foliation has also been considered as a joint set. Joints are closely spaced leading to formation of block with maximum size >0.25 m3. Rockfall are frequent throughout the year, especially in Monsoon rainy season and at the end of continuous freeze-thaw cycles. While travelling towards Besham, PDSD is 13%, which is very low whereas towards Chilas it is more than 80%. Distance between highway and slope is less than 2 meters, which is not enough to trap blocks falling down. Wedging was found major failure mode along with toppling. All these parameters and absence of protective measure has aggravate situation. Sum of score of all parameters (Table 6.1) was 716.5, which was then normalized to clearly understand risk assessment by using following equation:

Equ. (6.2) 퐻푛 = (퐻 − 퐻푚푖푛)/(퐻푚푎푥 − 퐻푚푖푛)

Where Hn is normalized value, H is total score of this site, Hmax and Hmin are maximum and minimum score along this section of the Highway. Calculated normalized value of this site was 0.98, and therefore was placed under very high-risk category.

6.7.2 ChoChang Rockfall

Overhangs produced by uncontrolled blasting with heavy explosives for road excavation characterize this section. Rockfall in these overhangs along this part of the Highway is regular phenomenon during rains. In April 2016, rockfall damaged, buried highway and blocked traffic for a week at this particular location (Fig. 6.6c). Tectonized banded amphibolies constitutes 148

98

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______meters high slope in this area. Three joint sets exist with open apertures, undulating joint planes, with clay filling and with an overall adverse impact on slope stability (Table 6.2F). Based on previous seismic events, seismic intensity was marked as VI at this site. Deposit of recent rockfall indicates variation in block size from few cm3 to two m3. Absence of countermeasure and effective ditch along the Highway makes it vulnerable. However, PDSD is 100% from both direction of the Highway decreasing vehicle’s vulnerability. During kinematic analysis, toppling was found major mode of failure. Computed normalized value for this site was 0.74, placing it under high risk.

6.7.3 Lotar Rockfall

This rockfall site is located near Dynter Valley Bridge (Fig. 6.6c). This area marks the entry of THE Kamila Fault (KF) in valley. Slope is combination of both cut and natural slope, with effective height of 215 meters. Gabbronorites are primary slope component, which is jointed and sheared at different places. During discontinuity survey, three joints were found with one joint dipping along the slope, facilitating plane failure (Table 6.2H). Joints are closely spaced with open apertures leading to formation of blocks <0.06m3. The site has seismic intensity of VI in previous earthquakes. Well-developed pile of failed mass at toe indicates frequency and history of rockfalls. Distance between highway and slope is less than 2 meters, indicating low ditch effectiveness. Furthermore, while travelling towards Besham, PDSD is quite low (26%). There is no protection for the road and travelers adding more to vulnerability. Overall, this zone was ranked at high risk with score of 0.71.

6.7.4 Yadgar Rockfall

This site located between Kandia valley bridge and Summer Nala (35.496822° N, 73.298431°E) (Fig. 6.7a). Torrential rains in Monsoon and Westerlies triggers rockfalls every year. Strongly foliated, massive gabbro-norites and diorites constitute 411 meter high slopes in this zone. During discontinuity survey, data related to 58 joints was collected. We found three joint sets with their mean dip direction/dip 105/60, 218/54 and 070/50 respectively. In addition to these joints, stress release joints with short persistence were found. These short persistence stress release joints and large normal joint spacing led into formation of huge blocks of more size more than 6m3 (Fig. 6.4). Removal and mobilization of these blocks without blasting is almost impossible. During kinematic analysis, both plane and toppling were found prominent failure modes (Fig. 6.4). While travelling towards the Chilas, an impulsive turn before this sites, limits decision distance and

99

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______therefore PDSD was found low (20%). All discussed parameters, ineffective ditch and lack of countermeasures turned this zone into very high-risk zone with normalized score of 1.

Figure 6.4: (a) An overview of Yadgar Rockfall (see location in 6.7a): It is a cutslope with an effective long natural slope (not visible). Large joint spacing and aperture has led into formation of huge blocks. Green circle is representing fresh scar. Red rectangle contains a losely held block, which can be a possible failure. Deposit of failed rockmass is also visible. Green arrow is pointing towards Highway’s turn just before site, decreasing driver’s decision sight distance. Pink and black rectangles showing locations of Fig. 6.4b and 6.4c. Kinematic analysis is explaining orientation of joints, slope and major possible failure mode as direct toppling. (b) Size of block >6m3 (Scale: man in red circle). (c) Plane failure in losely held blocks.

100

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______Risk Analysis of Debris Flows

Potential debris flow identified during the first step, were then rated using Table. 6.2. Each parameter was cautiously evaluated. Afterwards, total score of each site was normalised and classified. Following two examples are explained to give an overview of the rating criteria.

6.8.1 Serai Debris Flow

The site is located at 30 km from Besham toward Chilas (35° 4'13.00"N, 72°57'18.91"E; Fig. 6.6a), in Monsoon region, receiving maximum daily rainfall more than 80 mm. Blockade of highway at this site due to torrential rainfall is common. In April 2017 and 2018, 30 meters of the Highway was completely buried under its deposits. The catchment area (1.20 km2) is thinly vegetated, has 18% area with collapsed material produced by continuous slope failures in highly fragmented garnet pyroxenites. The channel is 1.59 km long with gradient of 1.23 km. Moreover, it is young and steep (>40o) at places having a lot of debris. Both active and potential slope failures have been found in catchment area, which are possible source of upcoming debris flows. There was continuous water seepage in the channel. Previously, due to absence of efficient drain or effective countermeasures, debris flow was directly flooding over the Highway. There is an embankment on opposite side, to avoid the Highway’s erosion, which is not enough to minimize damage. The calculated normalized score was 0.91 and ranked as very high-risk site.

6.8.2 Harbon Debris Flow

The Harbon site is located at 35°31'23.21"N and 73°39'36.03"E, 5 km east of Harbon Nala and in non-monsoon region with daily rainfall less than 45 mm (Fig. 6.7b). The site is a part of an historical large rockslide deposit (~4 km3), which dammed Indus River (age is unknown). Gully erosion is prominent in this deposit and furthermore, all gullies converge to a single gully, with steepness more than 30o at places. The catchment area (6 km2) is sparsely vegetated with high percentage of collapsed material (33%). Furthermore, slope failures in scarp and in deposit of old landslide are providing debris material. In April 2016, heavy rain initiated debris flow, which flooded, buried and blocked the Highway for traffic. Absence of culvert, bridge and other effective measure placed this site at very high risk with normalized risk score of 0.82.

101

Figure 6.5: (a) Overview of Serai debris flow (Google Earth 2017). White arrows expressing different elevations in catchment area. White rectangle is showing a slope failure within catchment area. Red arrows marks the Highway. (b) Red rectangle representing small retaining wall over drain, now completely blocked by deposit. Red circle showing highly fragmented rockmass, which is almost same throughout the catchment. White arrows representing the highest part of the catchment. (c) Debris flooding the Highway directly. (d) Variety of grain sizes, from silt to large boulder (>2m3).

102

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______Risk Map

It was an impossible task to assess debris flow and rockfall risk along entire 200 km stretch of the Highway, by surveying each site in detail. Therefore, potential hazardous sites were identified by using regional approach, which were further investigated in detail during field surveys. Rockfalls (47) and debris flows (26) were rated by using Table 6.1 and 6.2. Score of each site was normalized by using Eq. 2, further classified into four classes by equal interval technique: low, intermediate, high and very risk (Fig. 6.6, Fig. 6.7). Very high and high risk were found along 12.56 % and 18.67 % of the section respectively. Whereas 61 % of this section accounts for low and intermediate risk (Table 6.3).

The Highway near Jijal, Pattan, Kiru, Chochang, Lotar, Samar Nala and Kandian was at very high risk. Active faults (MMT, PF, and KF) and shear zones in these parts degraded rock mass where differential erosion triggered rockfall during heavy rain and seismicity.

The subsection between Jijal and Sazin is characterized by steep topography, having rockfall a major concern. Whereas section between Sazin and Chilas passes through broad valley except few places after Harbon Nala. Mostly it lies at low to intermediate risk except some locations. In past, sparsely vegetated, large catchment areas with enough debris supply generated hazardous debris flows in this area. Overall absence of countermeasures is a serious threat to Highway’s stability. Construction of effective countermeasures can change or minimize risk level of all sites.

Table 6.4: Percentage of different classes in final risk map

Risk Level Length (km) Percentage of total length (%)

Very High 25.13 12.56

High 37.34 18.67

Intermediate 57 28.48

Low 80.56 40.28

103

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______

Figure 6.6: Risk map of the studied section, (a) Besham-Pattan, (b) Pattan-Dassu, (c) Dassu- Kandia Valley. Numbers in white colours are showing location of explained examples: 1-Jijal Rockfall, 2-Serai Debris Flow, 3-Chochang Rockfall, 4-Lotar Rockfall. Red star giving location of Fig. 6.5. Whereas, capital letters in white colour is showing the locations of Table 6.2.

104

Figure 6.7: Risk map of the studied section, (a) Kandia Valley-Harbon Nala, (b) Harbon Nala-Chilas. Numbers in white colour are showing location of explained examples: 5-Yadgar Rockfall, 6-Harbon Debris Flow. Red star giving location of Fig. 6.4. Whereas, capital letters in white colour is showing the locations of Table 6.2.

105

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______Discussion and Conclusion

In this study, regional to site-specific approach (quantitative to semi-quantitative) was used to prepare risk map along a section of the KKH. In first phase, remote sensing was used to find potential hotspots, required to be investigated in detail. Source areas were defined by using two different techniques. DEM based slope angle distribution (SAD) was used to identify rockfall source areas. High resolution DEM has ability to detect even smaller source areas (Michoud et al., 2012) and has an ultimate impact on quality of results. Therefore, DEM with 12 m resolution was used for most of the area. Presence of no-data values in 12 m DEM has forced us to use low resolution (30 m) for some areas.

For debris flows, predefined mapped sources were used for runout assessment. FLOW-R was employed to assess approximate extent and impact of debris flows and rockfalls on the Highway.

Basic advantage of this software is that it can work with limited data and customised options, which allowed us to modify it according to local conditions. However, it had also some limitations, as it did not accounted some local parameters (weathering, structural conditions etc.) and volume but only surface area. Propagation extent of flows and falls was marginally more than real events, which decreased possibility of exclusion of even single observed event. Potential sites identified during this process were then surveyed in detail. Keeping in view nature of slope failures, two different rating criteria were used for risk assessment.

Modified Pierson’s RHRS was used to assess risk related with rockfalls. Eleven parameters grouped in four categories were considered for rating (Table 6.1). Evaluation of some parameters was subjective, depending upon expert judgment. RHRS is flexible and can be modified according to local geomorphological, geological and climatic conditions. PDSD was carefully calculated and Highway authorities cleared confusion created by absence of speed limit posters along some parts of the KKH. Due to insufficient daily data regarding vehicles passing through section, Average Vehicle Risk (AVR) was not taken into account. Width of the Highway in this section is constant and therefore, has not been considered. Kinematic analysis showed toppling and wedging major rockfall process along the studied section. Risk assessment of debris flow was executed by using semi-quantitative technique. Keeping in mind, basic components of debris flow (formation, initiation, transportation and deposition), eleven parameters were opted (Table. 6.2). Rainfall intensity is not uniform along studied section; therefore, daily maximum rainfall was included. Three parameters in both rating criteria were about current state of countermeasures

106

Empirical Assessment of Rockfall and Debris Flow Risk along Besham-Chilas Section ______and their capacity to protect Highway and traveller from hazard. Change in their current status can lead to decrease or increase vulnerability of the Highway.

Sum of scores of all parameters was then normalised using Eq. 6.2. It pertinent to mention that scores of rockfalls and debris flows were normalised separately. Equal interval classification method was applied to categorise the final risk map into four: very high, high, intermediate and low risk. Sites near Jijal, Kiru, Chochang, Yadgar, Kandian valley and Harbon Nala were found at very high risk. Section between Besham and Dubair, Dudishal and Chilas were safer with low to intermediate risk except a few sites with high score. Highway authorities can plan curative work according to risk level calculated. Sites with very high risk requires immediate inspection and countermeasures. Otherwise, Highway can potentially be blocked for more than two days at these sites. Highway section with high-risk has capacity to block traffic for less than a day. Warning signs and regular inspection are obligatory for these sites. Intermediate risk means no blockade but can damage vehicle or traveller. Low risk areas are quite safe and requires no countermeasures. Methods and techniques applied in this study can further be used for prioritization of road and highways affected by rockfalls and debris flows.

107

108

Discussion and conclusions ______

7 Conclusions and Outlook

Conclusions

This chapter gives an overview of important outcomes of research and further suggests future possible work. Broadly, this work can be grouped into three stages: (a) acquisition, compilation and digitization of published geological, inventory maps, structural, seismic and climatic data, to know possible control of mass movements (b) preparation of landslides susceptibility maps at regional level (c) site specific study to prepare risk map of highly susceptible sections of the Highway. In following paragraphs, these topics would be briefly explained to discuss what we achieved and what can be done in future.

Geological road log (Khan et al., 2000), published book (Khan et al., 2003), articles ( Jones et al., 1983; Shroder and Bishop, 1998; Hewitt, 2001, 1998; Derbyshire et al., 2001; Kibria and Masud, 2006; Hewitt et al., 2011) and reports (Fayaz et al., 1985) were reviewed and enclosed maps were digitised (Fig. 2.4 and Fig. 2.10). Climatic data was acquired from Pakistan Meteorological Department (PMD) to prepare precipitation map (Fig. 5.3). Road clearance logs prepared by Frontier Works Organization (FWO) and reports and books published by Geological Survey of Pakistan (GSP) (Fayaz et al., 1985; Khan et al., 2003) were interpreted to produce multi-temporal landslide inventory map (Fig. 5.6) for period of ten years (1982, 83, 1996, 97, 98, 99, 2000, 2014, 15, 16). Primary inventory elaborated frequency and location of landslides. Afterwards, these locations and satellite imagery were explored to prepare complete inventory map describing extents and types of landslides. Afterwards, during field campaigns, both inventory and geological maps were validated.

Spatial coincidence between landslides and lithological variation was analysed. Lithologies (amphibolites, dunnites, pyroxinites, schists, slates and quaternary) witnessed high landslide density. In addition to lithology, structure had significant role in determining landslide initiation and type. Relative position of lithologies with active faults, fragmentation of lithology and attitude and frequency of discontinuities strongly influenced landslide events. Translational failures were found in dip slope whereas rotational were found in reverse slope setting (Guzzetii et al., 1996). Moreover, strong negative association was found between distance from active fault and number of mass movements, blocked the Highway (Fig. 3.4b).

109

Discussion and conclusions ______

Mean monthly precipitation data was correlated with number of mass movements for the same period. Keeping in view variations in weather conditions, the Highway was divided into sections: (a) Hassan Abdal-Gilgit section (b) Gilgit-Khunjerab Pass section. Former lies in comparatively warmer while latter in colder area with pronounced effect of rainfall with snow melt. Former exhibited strong positive association between mean monthly precipitation and number of mass movements, concentrated in Monsoon and Westerlies seasons (Fig. 3.4a). Whereas in latter section, mass movements were concentrated in thawing period of early spring, when ice melting pronounced effect of precipitation (Fig. 3.4c). It is pertinent to mention that number of mass movements are meant by number of events took place, blocking the Highway. Subsequently, statistical analysis of meteorological variation (mean daily precipitation and minimum and maximum daily temperatures) were performed to know possible cause of rockfall initiation. Three sections of the Highway and then specific sites were considered separately to avoid any bias related to variations in local meteorological conditions and lithology respectively. Shattering effect due to freeze-thaw conditions was profound in north with extreme cold weather conditions. Also loss of cohesion and differential erosion due to ice melt induced rockfalls (D’Amato et al., 2016).

Geological, precipitation and inventory maps, produced earlier, were used to prepare landslide susceptibility map. Spatial analysis and previous publications were used to finalise ten parameters for this purpose. Some factors (lithology, active faults, seismicity, and rainfall intensity) were taken from above maps whereas other factors (slope angle, curvature, and aspect, elevation and hydrology) were derived from Digital Elevation Model (DEM). Land cover map was produced by using supervised classification. Afterwards, all causative parameters were classified into multiple classes based on spatial coincidence of landslides to avoid any personal biasness. AHP based scoring was applied to keep rating consistent. Because of strong influence over landslide distribution, slope steepness and distance from fault were highly ranked. Weighted overlay method was employed to produce the final map (Fig. 5.9 and Fig. 5.10), which was further classified into four classes: low, intermediate, high and very high susceptibility. Some sections were easily segregated on the basis of high landslide susceptibility: Jijal, Dassu, Sazin, Raikot Bridge, Chalt and north of Sost-Deh. All of these sections have highly fragmented rock mass due to presence of active faults and shear zones (MMT, KJS, KSF, RF, MKT and KF). Due to low spatial resolution, small cut slopes were ignored. Variations in land cover has not been considered.

Keeping in view susceptibility level and current state of countermeasures, Besham-Chilas section was chosen for detailed investigation for risk assessment. It is a case study, validating “regional

110

Discussion and conclusions ______to site specific approach” in comprehensive way. Landslide inventory, susceptibility map and runout assessment at regional level were employed to find potential hazardous sites. These sites were further surveyed in detail by using empirical assessment. Risk associated with rockfall sites was assessed by Modified Pierson’s RHRS. Eleven parameters were rated exponentially to clearly differentiate hazardous sites. Keeping in view basic debris flow processes (formation, initiation and transportation), eleven parameters were rated to calculate associated risk. Normalised score of each site was further applied to produce final risk map of the section (Fig. 6.6 and Fig. 6.7), which was further classified into four categories: low, intermediate, high and very high risk. Lastly, based on each risk category, remedial work was proposed.

Outlook

Detailed field surveys and site investigation gave an overview of the recent landslide activity, potential hazard and current state of countermeasures. Northern part of the Highway, from Raikot Bridge onwards and central part between Dassu and Thakot has been recently constructed. At places, countermeasures were built to avoid and minimise associated risk. During construction of countermeasure, protection or restraint work was focussed. While, prevention and control work was ignored. Rock sheds, catch walls (concrete and gabion), retaining walls, causeways and culvert were constructed (Fig. 7.1). Some of them are effective while some are ineffective because of poor understanding of landslide phenomenon and unawareness regarding its type, size and volume. Drainage of surface water is a primary reason for mass movements and needs special attention. Most of the landslides occurs during rainfall, where it infiltrates at some sites triggering debris slides and rockfall and becomes runoff other places triggering debris flows. So, draining surface water in a proper way, can avoid this process. Additionally, rock mass reinforcement by anchoring and bolting should be performed. Furthermore, construction of check and sabo dams upstream of the Highway along with existing causeways and culverts will improve the situation. The Highway at a few places, is at high risk and should be relocated to other bank of the rivers (Fig. 7.2). Most importantly, there is dire need to investigate slopes in detail and to evaluate factor of safety prior designing and construction of countermeasures.

Detailed structural mapping is required to know variation in fragmentation with distance from active faults and its ultimate impact on landslide occurrences. Hourly data can give more insight of precipitation effect and help to establish rainfall threshold for landslide initiation.

111

Figure 7.1: Overview of countermeasures against landslides: (a) rock shed to discharge debris flow without damaging the Highway (b) Tunnel to avoid notorious active rockfall site “Kafir Pahar” (c) Retaining wall to minimize damage by debris slide (d) Barrier wall or catch wall to protect the Highway from rock or scree fall (e) Gabion damaged by fallen block (f) Causeway to minimize hazard associated with debris flow.

112

Discussion and conclusions ______

Figure 7.2: Condition of the Highway, damaged by continuous erosion by Indus River (visible in zoomed view). Yellow line indicating possible relocated route while red line showing existing route. Raikot Bridge is located at a distance of almost 1 km in northeast, from where Highway follows this dangerous bank. So, if Highway follows the same western bank and then shifts to eastern bank after this problematic part, hazard related to erosion and landslides will be avoided.

113

114

References ______

8 References

Ahmad, A., Quegan, S., 2012. Analysis of Maximum Likelihood Classification on Multispectral Data. Appl. Math. Sci. 6, 6425–6436. doi:10.12988/ams.2013.34214 Ahmed, B., 2015. Landslide susceptibility mapping using multi-criteria evaluation techniques in Chittagong Metropolitan Area, Bangladesh. Landslides 12, 1077–1095. doi:10.1007/s10346- 014-0521-x Ahmed, M.F., Rogers, J.D., Ismail, E.H., 2014. A regional level preliminary landslide susceptibility study of the upper indus river basin. Eur. J. Remote Sens. 47, 343–373. Akgun, A., Dag, S., Bulut, F., 2008. Landslide susceptibility mapping for a landslide-prone area (Findikli, NE of Turkey) by likelihood-frequency ratio and weighted linear combination models. Environ. Geol. 54, 1127–1143. doi:10.1007/s00254-007-0882-8 Ali, S., Biermanns, P., Haider, R., Reicherter, K., 2018a. Landslide Susceptibility Mapping By Using Gis Along the China Pakistan Economic Corridor (Karakoram Highway), Pakistan . Nat. Hazards Earth Syst. Sci. 1–40. doi:10.5194/nhess-2018-39 Ali, S., Haider, R., Rashid, A., Reicherter, K., 2019. Meteorological variation and rockfall occurence along the Karakoram highway, in: 2nd International Conference on Natural Hazards & Infrastructure 23-26 June, 2019, Chania, Greece. Ali, S., Schneiderwind, S., Reicherter, K., 2017. Structural and Climatic Control of Mass Movements Along the Karakoram Highway, in: Advancing Culture of Living with Landslides. Springer International Publishing, pp. 509–516. doi:10.1007/978-3-319-53485-5_60 Ali, S., Tayyib Riaz, M., Yasir Sarfraz, Shafique, M., Basharat, M., Klaus Reicherter, 2018b. Rock Fall and Debris Flow Hazard and Risk Assessment along Karakoram Highway (Besham- Chilas Section), in: EGU General Assembly 2018. Ambraseys, N., Lensen, G., Moinfart, A., Penningtons, W., 1981. The Pattan (Pakistan) earthquake of 28 December 1974 : field observations The earliesknownt earthquakes of the Northwest Frontier Provinces ( NWFP ) Geological and tectonic setting Evidence of regional Quarternary and Recent movements. Q. J. Eng. Geol. Hydrogeol. 14, 1–16. Arizapa, J.L., Combalicer, E.A., Tiburan, C.J.L., 2015. Landslide Susceptibility Mapping of Pagsanjan – Lumban Watershed using GIS and Analytical Hierarchy Process. Ecosyst. Dev. J. 5, 23–32. Ayalew, L., Yamagishi, H., Marui, H., Kanno, T., 2005. Landslides in Sado Island of Japan: Part II. GIS-based susceptibility mapping with comparisons of results from two methods and verifications. Eng. Geol. 81, 432–445. doi:10.1016/j.enggeo.2005.08.004 Ayalew, L., Yamagishi, H., Ugawa, N., 2004. Landslide susceptibility mapping using GIS-based weighted linear combination, the case in Tsugawa area of Agano River, Niigata Prefecture, Japan. Landslides 1, 73–81. doi:10.1007/s10346-003-0006-9

115

References ______

Bacha, A.S., Shafique, M., van der Werff, H., 2018. Landslide inventory and susceptibility modelling using geospatial tools, in Hunza-Nagar valley, northern Pakistan. J. Mt. Sci. 15, 1354–1370. doi:10.1007/s11629-017-4697-0 Bachri, S., Shresta, R.P., 2010. Landslide hazard assessment using analytic hierarchy processing (AHP) and geographic information system in Kaligesing mountain area of Central Java Province Indonesia, in: 5th Annual International Workshop & Expo on Sumatra Tsunami Disaster & Recovery, 23-24 Nov 2010, Indonesia (Banda Aceh). Bard, J.P., Maluski, H., Matte, P., Proust, F., 1980. The Kohistan Sequence: Crust and Mantle of an Obducted Island Arc, in: Proc. Intern. Commit. Geodynamics, Grp. 6, Mtg. Peshawar, Nov. 23-29, 1979: Spec. Issue, Geol. Bull. Univ. Peshawar, Vol. 13, 1980. Basharat, M., Shah, H.R., Hameed, N., 2016. Landslide susceptibility mapping using GIS and weighted overlay method: a case study from NW Himalayas, Pakistan. Arab. J. Geosci. 9, 1–19. doi:10.1007/s12517-016-2308-y Baumann, V., Wick, E., Horton, P., Jaboyedoff, M., 2010. Debris flow susceptibility mapping at a regional scale along the National Road N7, Argentina. Proc. 14th Pan-American Conf. Soil Mech. Geotech. Eng. Bishop, M.P., Shroder, J.F., Bonk, R., Olsenholler, J., 2002. Geomorphic change in high mountains: A western Himalayan perspective. Glob. Planet. Change 32, 311–329. doi:10.1016/S0921-8181(02)00073-5 Borneman, N.L., Hodges, K. V., Van Soest, M.C., Bohon, W., Wartho, J.A., Cronk, S.S., Ahmad, T., 2015. Age and structure of the Shyok suture in the Ladakh region of northwestern India: Implications for slip on the Karakoram fault system. Tectonics 34, 2011–2033. doi:10.1002/2015TC003933 Brenning, A., 2005. Spatial prediction models for landslide hazards: review, comparison and evaluation. Nat. Hazards Earth Syst. Sci. 5, 853–862. doi:10.5194/nhess-5-853-2005 Budetta, P., 2004. Assessment of rockfall risk along roads. Nat. Hazards Earth Syst. Sci. 4, 71– 81. doi:10.5194/nhess-4-71-2004 Budetta, P., De Luca, C., Nappi, M., 2016. Quantitative rockfall risk assessment for an important road by means of the rockfall risk management (RO.MA.) method. Bull. Eng. Geol. Environ. 75, 1377–1397. doi:10.1007/s10064-015-0798-6 Burg, J.P., Jagoutz, O., Dawood, H., Shahid Hussain, S., 2006. Precollision tilt of crustal blocks in rifted island arcs: Structural evidence from the Kohistan Arc. Tectonics 25, 1–13. doi:10.1029/2005TC001835 Butt, A., Shabbir, R., Ahmad, S.S., Aziz, N., 2015. Land use change mapping and analysis using Remote Sensing and GIS: A case study of Simly watershed, Islamabad, Pakistan. Egypt. J. Remote Sens. Sp. Sci. 18, 251–259. doi:10.1016/j.ejrs.2015.07.003 Canuti, P., Garduño, V.H., Garzonio, C.A., Iotti, A., 1993. Slope evolution and mass movements in the Mt. Amiata region (Tuscany, Italy), in: International Workshop on Environmental Volcanology, January 1993. Riassunto, pp. 35–36. doi:10.1123/tsp.5.1.15

116

References ______

Cao, C., Xu, P., Chen, J., Zheng, L., Niu, C., 2016. Hazard Assessment of Debris-Flow along the Baicha River in Heshigten Banner, Inner Mongolia, China. Int. J. Environ. Res. Public Health 14, 30. doi:10.3390/ijerph14010030 Capra, L., Macías, J.L., Scott, K.M., Abrams, M., Garduño-Monroy, V.H., 2002. Debris avalanches and debris flows transformed from collapses in the Trans-Mexican Volcanic Belt, Mexico - Behavior, and implications for hazard assessment. J. Volcanol. Geotherm. Res. 113, 81– 110. doi:10.1016/S0377-0273(01)00252-9 Cardinali, M., Ardizzone, F., Galli, M., Guzzetti, F., Reichenbach, P., 2000. Landslides triggered by rapid snow melting: the December 1996-January 1997 event in Central Italy, in: Proceedings Plinius Conference ’99, Maratea, 14-16 October 1999. pp. 439–448. Casagli, N., Yin, Y., Sassa, K., Forms, L., 2017. Advancing Culture of Living with Landslides. doi:10.1007/978-3-319-53485-5 Chen, H., Dadson, S., Chi, Y.-G., 2006. Recent rainfall-induced landslides and debris flow in northern Taiwan. Geomorphology 77, 112–125. doi:10.1016/J.GEOMORPH.2006.01.002 Chen, H.X., Zhang, L.M., Chang, D.S., Zhang, S., 2012. Mechanisms and runout characteristics of the rainfall-triggered debris flow in Xiaojiagou in Sichuan Province, China. Nat. Hazards 62, 1037–1057. doi:10.1007/s11069-012-0133-5 Coco, L., Buccolini, M., 2015. The Effect of Morphometry, Land-use and Lithology on Landslides Susceptibility: An Exploratory Analysis. [WWW Document]. IOS Press. doi:10.3233/978-1- 61499-580-7-779 Corominas, J., Moya, J., 2008. A review of assessing landslide frequency for hazard zoning purposes. Eng. Geol. doi:10.1016/j.enggeo.2008.03.018 Coward, M.P., Rex, D.C., Asif Khan, M., Windley, B.F., Broughton, R.D., Luff, I.W., Petterson, M.G., Pudsey, C.J., 1986. Collision tectonics in the NW Himalayas. Geol. Soc. Spec. Publ. 19, 203–219. doi:10.1144/GSL.SP.1986.019.01.11 Cui, P., Xiang, L.Z., Zou, Q., 2013. Risk assessment of highways affected by debris flows in Wenchuan earthquake area. J. Mt. Sci. 10, 173–189. doi:10.1007/s11629-013-2575-y D’Amato, J., Hantz, D., Guerin, A., Jaboyedoff, M., Baillet, L., Mariscal, A., 2016. Influence of meteorological factors on rockfall occurrence in a middle mountain limestone cliff. Nat. Hazards Earth Syst. Sci. 16, 719–735. doi:10.5194/nhess-16-719-2016 Dadson, S.J., Hovius, N., Chen, H., Dade, W.B., Lin, J.C., Hsu, M.L., Lin, C.W., Horng, M.J., Chen, T.C., Milliman, J., Stark, C.P., 2004. Earthquake-triggered increase in sediment delivery from an active mountain belt. Geology 32, 733–736. doi:10.1130/G20639.1 Debon, F., Le Fort, P., Dautel, D., Sonet, J., Zimmermann, J.L., 1987. Granites of western Karakorum and northern Kohistan (Pakistan): A composite mid-cretaceous to upper cenozoic magmatism. LITHOS. doi:10.1016/0024-4937(87)90022-3 DeGraff, J. V., 1985. Using Isopleth Maps of Landslide Deposits as a Tool in Timber Sale Planning. Environ. Eng. Geosci. xxii, 445–453. doi:10.2113/gseegeosci.xxii.4.445

117

References ______

Delonca, A., Gunzburger, Y., Verdel, T., 2014. Statistical correlation between meteorological and rockfall databases. Nat. Hazards Earth Syst. Sci. 14, 1953–1964. doi:10.5194/nhess-14- 1953-2014 Deng, X., Li, L., Tan, Y., 2017. Validation of Spatial Prediction Models for Landslide Susceptibility Mapping by Considering Structural Similarity. ISPRS Int. J. Geo-Information 6, 103. doi:10.3390/ijgi6040103 Derbyshire, E., Fort, M., Owen, L.A., 2001. Geomorphological hazards along the Karakoram Highway: Khunjerab Pass to the Gilgit River, northernmost Pakistan. Erdkunde 55, 49–71. doi:10.3112/erdkunde.2001.01.04 Ding, M., Wei, F., Hu, K., 2012. Property insurance against debris-flow disasters based on risk assessment and the principal-agent theory. Nat. Hazards 60, 801–817. doi:10.1007/s11069- 011-9897-2 DiPietro, J. a., Hussain, a., Ahmad, I., Khan, M. a., 2000. The Main Mantle Thrust in Pakistan: its character and extent. Geol. Soc. London, Spec. Publ. 170, 375–393. doi:10.1144/GSL.SP.2000.170.01.20 DiPietro, J.A., Pogue, K.R., 2004. Tectonostratigraphic subdivisions of the Himalaya: A view from the west. Tectonics 23. doi:10.1029/2003TC001554 Douglas, G.R., 1980. Magnitude frequency study of rockfall in County Antrim, Northern Ireland. Earth Surf. Process. 5, 123–129. doi:10.1002/esp.3760050203 Ercanoglu, M., 2005. Landslide susceptibility assessment of SE Bartin (West Black Sea region, Turkey) by artificial neural networks. Nat. Hazards Earth Syst. Sci. 5, 979–992. doi:10.5194/nhess-5-979-2005 Escape, C.M., Alemania, M.K., Luzon, P.K., Felix, R., 2013. Comparison of various remote sensing classification methods for landslide detection using ArcGIS [WWW Document]. UP NOAH Cent. URL https://center.noah.up.edu.ph/comparison-of-various-remote-sensing- classification-methods-for-landslide-detection-using-arcgis/ (accessed 3.5.17). Fawcett, T., 2006. An introduction to ROC analysis. Pattern Recognit. Lett. 27, 861–874. doi:10.1016/j.patrec.2005.10.010 Fayaz, A., Latif, M., Khan, K.S.A., 1985. Landslide Evaluation and Stabilization Between Gilgit and Thakot Along the Karakoram Highway. Geological Survey of Pakistan, Islamabad. Ferlisi, S., Cascini, L., Corominas, J., Matano, F., 2012. Rockfall risk assessment to persons travelling in vehicles along a road: The case study of the Amalfi coastal road (southern Italy). Nat. Hazards 62, 691–721. doi:10.1007/s11069-012-0102-z Francipane, A., Arnone, E., Lo Conti, F., Puglisi, C., Noto, L.V., 2014. A Comparison Between Heuristic, Statistical, And Data-Driven Methods In Landslide Susceptibility Assessment: An Application To The Briga And Giampilieri Catchments. 11o Int. Conf. Hydroinformations 9. Fressard, M., Thiery, Y., Maquaire, O., 2014. Which data for quantitative landslide susceptibility mapping at operational scale case study of the pays d’auge plateau hillslopes (Normandy, France). Nat. Hazards Earth Syst. Sci. 14, 569–588. doi:10.5194/nhess-14-569-2014

118

References ______

Gaetani, M., Angiolini, L., Garzanti, E., Jadoul, F., Leven, E.Y.A., Nicora, A., Sciunnach, D., 1995. Permian Stratigraphy in the Northern Karakorum, Pakistan. Riv. Ital. di Paleontol. e Stratigr. 101, 107–152. Gentile, F., Bisantino, T., Trisorio Liuzzi, G., 2008. Debris-flow risk analysis in south Gargano watersheds (Southern-Italy). Nat. Hazards 44, 1–17. doi:10.1007/s11069-007-9139-9 Gökceoglu, C., Aksoy, H., 1996. Landslide susceptibility mapping of the slopes in the residual soils of the Mengen region (Turkey) by deterministic stability analyses and image processing techniques. Eng. Geol. 44, 147–161. doi:10.1016/S0013-7952(97)81260-4 Goudie, A.S., Brunsden, D., Collins, D.N., Derbyshire, E., Ferguson, R.I., Hashnet, Z., Jones, D.K.C., Perrot, F.A., Said, M., Waters, R.S., Whalley, W.B., 1984. The geomorphology of the Hunza Valley, Karakoram mountains, Pakistan. Int. Karakoram-Project 2, 359–411. Guzzetii, F., Cardinali, M., Reichenbach, P., 1996. The Influence of Structural Setting and Lithology on Landslide Type and Pattern. Environ. Eng. Geosci. 2, 531–555. Guzzetti, F., Cardinali, M., Reichenbach, P., Carrara, A., 2000. Comparing landslide maps: A case study in the upper Tiber River basin, central Italy. Environ. Manage. 25, 247–263. doi:10.1007/s002679910020 Guzzetti, F., Carrara, A., Cardinali, M., Reichenbach, P., 1999a. Landslide hazard evaluation: A review of current techniques and their application in a multi-scale study, Central Italy, in: Geomorphology. doi:10.1016/S0169-555X(99)00078-1 Guzzetti, F., Carrara, A., Cardinali, M., Reichenbach, P., 1999b. Landslide hazard evaluation: A review of current techniques and their application in a multi-scale study, Central Italy. Geomorphology 31, 181–216. doi:10.1016/S0169-555X(99)00078-1 Guzzetti, Fausto, Mondini, A.C., Cardinali, M., Fiorucci, F., Santangelo, M., Chang, K.T., 2010. Landslide inventory maps: New tools for an old problem. Earth-Science Rev. 112, 42–66. doi:10.1016/j.earscirev.2012.02.001 Guzzetti, F., Reichenbach, P., Wieczorek, G.F., 2010. Rockfall hazard and risk assessment in the Yosemite Valley, California, USA. Nat. Hazards Earth Syst. Sci. 3, 491–503. doi:10.5194/nhess-3-491-2003 Harp, E.L., Noble, M.A., 1993. An Engineering Rock Classification to Evaluate Seismic Rock-Fall Susceptibility and its Application to the Wasatch Front. Environ. Eng. Geosci. xxx, 293–319. doi:10.2113/gseegeosci.xxx.3.293 Hewitt, K., 2009. Glacially conditioned rock-slope failures and disturbance-regime landscapes, Upper Indus Basin, northern Pakistan. Geol. Soc. London, Spec. Publ. 320, 235–255. doi:10.1144/SP320.15 Hewitt, K., 2002. Styles of rock-avalanche depositional complexes conditioned by very rugged terrain, Karakoram Himalaya, Pakistan. Rev. Eng. Geol. XV, 345–377. doi:10.1130/REG15- p345 Hewitt, K., 2001. Catastrophic rockslides and the geomorphology of the Hunza and Gilgit River Valleys, Karakoram Himalaya. Erdkunde 55, 72–93. doi:10.3112/erdkunde.2001.01.05

119

References ______

Hewitt, K., 1998. Catastrophic landslides and their effects on the Upper Indus streams, Karakoram Himalaya, northern Pakistan. Geomorphology 26, 47–80. doi:10.1016/S0169- 555X(98)00051-8 Hewitt, K., Gosse, J., Clague, J.J., 2011. Rock avalanches and the pace of late Quaternary development of river valleys in the Karakoram Himalaya. Bull. Geol. Soc. Am. 123, 1836– 1850. doi:10.1130/B30341.1 Hoek, E., Bray, J., 1981. Rock Slope Engineering, 3rd ed, Institution of Mining and Metallurgy. Institution of Mining and Metallurgy, London. doi:10.1007/0-387-30842-3_59 Horton, P., Jaboyedoff, M., Zimmermann, M., Mazotti, B., Longchamp, C., 2011. Flow-R, a model for debris flow susceptibility mapping at a regional scale - some case studies. Proc. 5th Int. Conf. Debris-Flow Hazards Mitig. Mech. Predict. Assess. - Ital. J. Eng. Geol. Environ. 875– 884. doi:10.4408/IJEGE.2011-03.B-095 Hovius, N., Stark, C.P., Allen, P.A., 1997. Sediment flux from a mountain belt derived by landslide mapping. Geology 25, 231–234. Hürlimann, M., Copons, R., Altimir, J., 2006. Detailed debris flow hazard assessment in Andorra: A multidisciplinary approach. Geomorphology 78, 359–372. Intarawichian, N., Dasananda, S., 2010. Analytical hierarchy process for landslide susceptibility mapping in lower Mae Chaem watershed, Northern Thailand. Suranaree J. Sci. Technol. 17, 277–292. Jaboyedoff, M., Labiouse, V., 2003. Preliminary assessment of rockfall hazard based on GIS data. Rock Mech. 575–578. Jaboyedoff, M., Rudaz, B., Horton, P., 2011. Concepts and parameterization of Perla and FLM model using Flow-R for debris flow. Proc. 5th Can. Conf. Geotech. Nat. Hazards 14, 11414. Jade, S., 2004. Estimates of plate velocity and crustal deformation in the Indian subcontinent using GPS geodesy. Curr. Sci. 86, 1443–1448. Jones, D.K.C., Brunsden, D., Goudie, A.S., 1983. A preliminary geomorphological assessment of part of the Karakoram Highway. Q. J. Eng. Geol. Hydrogeol. 16, 331–355. doi:10.1144/GSL.QJEG.1983.016.04.10 Kamp, U., Growley, B.J., Khattak, G.A., Owen, L.A., 2008. GIS-based landslide susceptibility mapping for the 2005 Kashmir earthquake region. Geomorphology 101, 631–642. doi:10.1016/j.geomorph.2008.03.003 Kanwal, S., Atif, S., Shafiq, M., 2016. GIS based landslide susceptibility mapping of northern areas of Pakistan, a case study of Shigar and Shyok Basins. Geomatics, Nat. Hazards Risk 5705, 1–19. doi:10.1080/19475705.2016.1220023 Kazmi, A.H., Jan, M.Q., 1997. Geology and Tectonics of Pakistan. Graphic Publishers, Karachi. Keefer, D.K., 1984. Landslides caused by earthquakes: Discussion. Bull. Geol. Soc. Am. 95, 406– 421. Khan, H., Shafique, M., Khan, M.A., Bacha, M.A., Shah, S.U., Calligaris, C., 2018. Landslide

120

References ______

susceptibility assessment using Frequency Ratio, a case study of northern Pakistan. Egypt. J. Remote Sens. Sp. Sci. 22, 11–24. doi:10.1016/j.ejrs.2018.03.004 Khan, K.S.A., Fayaz, A., Hussain, M., Latif, M., 2003. Landslides Problems and Their Mitigation along the Karakoram Highway, 1st ed. Geological Survey of Pakistan, Islamabad. Khan, K.S.A., Fayaz, A., Latif, M., Wazir, A.K., 1986. Rock and Debris Slides Between Khunjrab Pass and Gilgit along the Karakoram Highway. Islamabad. Khan, K.S.A., Latif, M., Fayaz, A., Khan, N.A., Khan, S.Z., 2000. Geological Roadlog along the Karakorum Highway from Islamabad to Khunjrab Pass. Khan, M.A., Jan, M.Q., 1991. An interpretation of the Himalaya-Karakoram region of Pakistan in the light of plate tectonics and rock ages. Pakistan J. Geogr. 1, 114–136. Khan, M.A., Jan, M.Q., Windley, B.F., Tarney, J., Thirlwall, M.F., 1989. The Chilas mafic- ultramafic igneous complex; the root of the Kohistan island arc in the Himalaya of northern Pakistan. Geol. Soc. Am. 232, 75–94. Khan, T., Jan, M.Q., Khan, M.A., Kausar, A.B., 1997. High-grade metasedimentary rocks (Gilgit Formation) in the vicinity of Gilgit, Kohistan, northern Pakistan. J. Min. Petr. Econ. Geol. 92, 465–479. Kibria, S., Masud, B., 2006. Tackling Landslides during Extension of KKH, in: Pakistan Engineering Congress, 70th Annual Session Proceedings. pp. 264–292. Komac, M., 2006. A landslide susceptibility model using the Analytical Hierarchy Process method and multivariate statistics in perialpine Slovenia. Geomorphology 74, 17–28. doi:10.1016/j.geomorph.2005.07.005 Lari, S., Frattini, P., Crosta, G., Jaboyedoff, M., Horton, P., 2011. Rockfall and debris flow societal and economic risk assessment at the regional scale. Acts 10th World Water Day, Accad. Naz. dei Lincei - Atti dei Convegni Lincei 179–187. Lee, S., 2005. Application of logistic regression model and its validation for landslide susceptibility mapping using GIS and remote sensing data. Int. J. Remote Sens. 26, 1477–1491. doi:10.1080/01431160412331331012 Lee, S., Chwae, U., Min, K., 2002. Landslide susceptibility mapping by correlation between topography and geological structure: The Janghung area, Korea. Geomorphology 46, 149– 162. doi:10.1016/S0169-555X(02)00057-0 Lee, S., Ryu, J.-H., Won, J.-S., Park, H.-J., 2004. Determination and application of the weights for landslide susceptibility mapping using an artificial neural network. Eng. Geol. 71, 289– 302. doi:10.1016/S0013-7952(03)00142-X Liang, W. jie, Zhuang, D. fang, Jiang, D., Pan, J. jun, Ren, H. yan, 2012. Assessment of debris flow hazards using a Bayesian Network. Geomorphology 171–172, 94–100. doi:10.1016/j.geomorph.2012.05.008 Liao, Z., Hong, Y., Kirschbaum, D., Adler, R.F., Gourley, J.J., Wooten, R., 2011. Evaluation of TRIGRS (transient rainfall infiltration and grid-based regional slope-stability analysis)’s

121

References ______

predictive skill for hurricane-triggered landslides: A case study in Macon County, North Carolina. Nat. Hazards 58, 325–339. doi:10.1007/s11069-010-9670-y Liu, K.-F., Li, H.-C., Hsu, Y.-C., 2009. Debris flow hazard assessment with numerical simulation. Nat. Hazards 49, 137–161. doi:10.1007/s11069-008-9285-8 Losasso, L., Derron, M.H., Horton, P., Jaboyedoff, M., Sdao, F., 2016. Definition and mapping of potential rockfall source and propagation areas at a regional scale in Basilicata region (Southern Italy). Rend. Online Soc. Geol. Ital. 41, 175–178. doi:10.3301/ROL.2016.122 Loye, A., Jaboyedoff, M., Pedrazzini, A., 2009. Identification of potential rockfall source areas at a regional scale using a DEM-based geomorphometric analysis 1643–1653. Luckman, B.H., 1976. Rockfalls and rockfall inventory data: Some observations from surprise valley, Jasper National Park, Canada. Earth Surf. Process. 1, 287–298. doi:10.1002/esp.3290010309 Ma, C., Hu, K., Tian, M., 2013. Comparison of debris-flow volume and activity under different formation conditions. Nat. Hazards 67, 261–273. doi:10.1007/s11069-013-0557-6 Malek, Ž., Zumpano, V., Schröter, D., Glade, T., Balteanu, D., Micu, M., 2015. Scenarios of land cover change and landslide susceptibility: An example from the buzau subcarpathians, romania. Eng. Geol. Soc. Territ. - Vol. 5 Urban Geol. Sustain. Plan. Landsc. Exploit. 743– 746. doi:10.1007/978-3-319-09048-1_144 Mateos, R.M., García-Moreno, I., Azañón, J.M., 2012. Freeze-thaw cycles and rainfall as triggering factors of mass movements in a warm Mediterranean region: The case of the Tramuntana Range (Majorca, Spain). Landslides 9, 417–432. doi:10.1007/s10346-011- 0290-8 Matsuoka, N., 1994. Diurnal freeze–thaw depth in rockwalls Field measurements and theoretical considerations. Earth Surf. Process. Landforms 19, 423–435. Michoud, C., Derron, M.H., Horton, P., Jaboyedoff, M., Baillifard, F.J., Loye, A., Nicolet, P., Pedrazzini, A., Queyrel, A., 2012. Rockfall hazard and risk assessments along roads at a regional scale: Example in Swiss Alps. Nat. Hazards Earth Syst. Sci. 12, 615–629. doi:10.5194/nhess-12-615-2012 Mineo, S., Pappalardo, G., D’Urso, A., Calcaterra, D., 2017. Event tree analysis for rockfall risk assessment along a strategic mountainous transportation route. Environ. Earth Sci. 76. doi:10.1007/s12665-017-6958-1 Mineo, S., Pappalardo, G., Mangiameli, M., Campolo, S., Mussumeci, G., 2018. Rockfall analysis for preliminary hazard assessment of the cliff of taormina Saracen Castle (Sicily). Sustain. 10, 1–18. doi:10.3390/su10020417 MonaLisa, Khwaja, A.A., Jan, M.Q., Yeats, R.S., Hussain, A., Khan, S.A., 2009. New data on the Indus Kohistan seismic zone and its extension into the Hazara-Kashmir Syntaxis, NW Himalayas of Pakistan. J. Seismol. 13, 339–361. doi:10.1007/s10950-008-9117-z Nagarajan, R., Roy, A., Vinod Kumar, R., Mukherjee, A., Khire, M. V., 2000. Landslide hazard susceptibility mapping based on terrain and climatic factors for tropical monsoon regions.

122

References ______

Bull. Eng. Geol. Environ. 58, 275–287. doi:10.1007/s100649900032 Nandi, A., Shakoor, A., 2010. A GIS-based landslide susceptibility evaluation using bivariate and multivariate statistical analyses. Eng. Geol. doi:10.1016/j.enggeo.2009.10.001 Nilsen, T.H., Wright, R.H., Vlasic, T.C., Spangle, W., 1979. Relative slope stability and land-use planning. Selected examples from the San Francisco Bay region, California. [WWW Document]. US Geol. Surv. Prof. Pap. URL http://www.scopus.com/inward/record.url?eid=2- s2.0-0018731268&partnerID=tZOtx3y1 (accessed 7.11.17). Ocakoglu, F., Gokceoglu, C., Ercanoglu, M., 2002. Dynamics of a complex mass movement triggered by heavy rainfall: A case study from NW Turkey. Geomorphology. doi:10.1016/S0169-555X(01)00094-0 Ohlmacher, G.C., Davis, J.C., 2003. Using multiple logistic regression and GIS technology to predict landslide hazard in northeast Kansas, USA. Eng. Geol. 69, 331–343. doi:10.1016/S0013-7952(03)00069-3 Oppikofer, T., Hermanns, R.L., Horton, P., Sandøy, G., Roberts, N.J., Jaboyedoff, M., Böhme, M., Yugsi Molina, F.X., 2014. Consequence assessment of large rock slope failures in Norway. Geophys. Res. Abstr. 16, 11286. doi:10.13140/RG.2.2.12340.55682 Park, S., Choi, C., Kim, B., Kim, J., 2013. Landslide susceptibility mapping using frequency ratio, analytic hierarchy process, logistic regression, and artificial neural network methods at the Inje area, Korea. Environ. Earth Sci. 68, 1443–1464. doi:10.1007/s12665-012-1842-5 Peila, D., Guardini, C., 2008. Use of the event tree to assess the risk reduction obtained from rockfall protection devices. Nat. Hazards Earth Syst. Sci. 8, 1441–1450. doi:10.5194/nhess- 8-1441-2008 Perret, S., Stoffel, M., Kienholz, H., 2006. Spatial and temporal rockfall activity in a forest stand in the Swiss Prealps - A dendrogeomorphological case study. Geomorphology 74, 219–231. doi:10.1016/j.geomorph.2005.08.009 Petley, D., 2016. Landslide damage to the Karakoram Highway [WWW Document]. Voice Earth Sp. Sci. Community. URL https://blogs.agu.org/landslideblog/2016/04/13/karakoram- highway-1/ (accessed 6.12.18). Pierson, L., Davis, S., Van, Vickle, R., 1993. Rockfall Hazard Rating System. Federal Highway Administration, Washington, D.C. Pourghasemi, H. Reza, Beheshtirad, M., Pradhan, B., 2016. A comparative assessment of prediction capabilities of modified analytical hierarchy process (M-AHP) and Mamdani fuzzy logic models using Netcad-GIS for forest fire susceptibility mapping. Geomatics, Nat. Hazards Risk 7, 861–885. doi:10.1080/19475705.2014.984247 Pourghasemi, H.R., Pradhan, B., Gokceoglu, C., 2012. Application of fuzzy logic and analytical hierarchy process (AHP) to landslide susceptibility mapping at Haraz watershed, Iran. Nat. Hazards 63, 965–996. doi:10.1007/s11069-012-0217-2 Pourghasemi, H.R., Rossi, M., 2017. Landslide susceptibility modeling in a landslide prone area in Mazandarn Province, north of Iran: a comparison between GLM, GAM, MARS, and M-

123

References ______

AHP methods. Theor. Appl. Climatol. 130, 609–633. doi:10.1007/s00704-016-1919-2 Pradhan, B., Lee, S., Buchroithner, M.F., 2010a. Remote Sensing and GIS-based Landslide Susceptibility Analysis and its Cross-validation in Three Test Areas Using a Frequency Ratio Model. Photogramm. - Fernerkundung - Geoinf. 2010, 17–32. doi:10.1127/1432- 8364/2010/0037 Pradhan, B., Youssef, A.M., Varathrajoo, R., 2010b. Approaches for delineating landslide hazard areas using different training sites in an advanced artificial neural network model. Geo- Spatial Inf. Sci. 13, 93–102. doi:10.1007/s11806-010-0236-7 Pritchard, M., Porter, M., Savigny, W., Bruce, I., Oboni, F., Keegan, T., Abbott, B., 2005. CN Rockfall Hazard Risk Management System: Experience, Enhancements, and Future Direction. 2005 Int. Conf. Landslide Risk Manag. 18th Annu. Vancouver Geotech. Soc. Symp. 14. Qi, S., Xu, Q., Lan, H., Zhang, B., Liu, J., 2010. Spatial distribution analysis of landslides triggered by 2008.5.12 Wenchuan Earthquake, China. Eng. Geol. doi:10.1016/j.enggeo.2010.07.011 Rahim, I., Ali, S.M., Aslam, M., 2018. GIS Based Landslide Susceptibility Mapping with Application of Analytical Hierarchy Process in District Ghizer, Gilgit Baltistan Pakistan. J. Geosci. Environ. Prot. 06, 34–49. doi:10.4236/gep.2018.62003 Rahman, G., Miandad, M., Collins, A.E., n.d. Spatial analysis of Landslide Susceptibility using Failure Rate approach in the Hindu Kush region , Pakistan Spatial analysis of Landslide Susceptibility using Failure Rate approach in the Hindu Kush region , Pakistan 1–30. Regmi, A.D., Cui, P., Dhital, M.R., Zou, Q., 2016. Rock fall hazard and risk assessment along Araniko Highway, Central Nepal Himalaya. Environ. Earth Sci. 75, 1–20. doi:10.1007/s12665-016-5905-x Reichenbach, P., Rossi, M., Malamud, B.D., Mihir, M., Guzzetti, F., 2018. A review of statistically- based landslide susceptibility models. Earth-Science Rev. 180, 60–91. doi:10.1016/j.earscirev.2018.03.001 Reis, S., 2008. Analyzing land use/land cover changes using remote sensing and GIS in Rize, North-East Turkey. Sensors 8, 6188–6202. doi:10.3390/s8106188 Restrepo, C., Alvarez, N., 2006. Landslides and their contribution to land-cover change in the mountains of Mexico and Central America. Biotropica 38, 446–457. doi:10.1111/j.1744- 7429.2006.00178.x Riaz, M., Khattak, A.G.A., 2002. Earthquakes in Nanga Parbat Area, North Pakistan: Causes and Concerns with particular reference to Basha Dam. Geol. Bull. Univ. Peshawar 35, 139–150. Rickenmann, D., 2016. Debris-Flow Hazard Assessment and Methods Applied in Engineering Practice. Int. J. Eros. Control Eng. 9, 80–90. doi:10.13101/ijece.9.80 Roslee, R., Mickey, A.C., Simon, N., Norhisham, M.N., 2017. Landslide Susceptibility Analysis ( Lsa ) Using Weighted Overlay Method ( Wom ) Along the Genting Sempah To Bentong Highway , Pahang. Malaysian J. Geosci. 1, 13–19. doi:10.26480/mjg.02.2017.13.19

124

References ______

Rozos, D., Bathrellos, G.D., Skillodimou, H.D., 2011. Comparison of the implementation of rock engineering system and analytic hierarchy process methods, upon landslide susceptibility mapping, using GIS: A case study from the Eastern Achaia County of Peloponnesus, Greece. Environ. Earth Sci. 63, 49–63. doi:10.1007/s12665-010-0687-z Ruff, M., Czurda, K., 2008. Landslide susceptibility analysis with a heuristic approach in the Eastern Alps (Vorarlberg, Austria). Geomorphology 94, 314–324. Rwanga, S.S., Ndambuki, J.M., 2017. Accuracy Assessment of Land Use/Land Cover Classification Using Remote Sensing and GIS. Int. J. Geosci. 08, 611–622. doi:10.4236/ijg.2017.84033 Saaty, R.W., 1987. The analytic hierarchy process-what it is and how it is used. Math. Model. 9, 161–176. doi:10.1016/0270-0255(87)90473-8 Saaty, T.L., 1990. How to make a decision: The analytic hierarchy process. Eur. J. Oper. Res. 48, 9–26. doi:10.1016/0377-2217(90)90057-I Sandersen, F. (NGI), Bakkehøi, S. (NGI), Hestnes, E. (NGI), Lied, K. (NGI), 1996. The influence of meteorological factors on the initiation of debris flows, rockfalls, rockslides and rockmass stability. Publ. - Norges Geotek. Inst. 201, 97–114. Santi, P.M., Hewitt, K., VanDine, D.F., Cruz, E.B., 2011. Debris-flow impact, vulnerability, and response. Nat. Hazards 56, 371–402. doi:10.1007/s11069-010-9576-8 Sarkar, S., Kanungo, D.P., 2004. An integrated approach for landslide susceptibility mapping using remote sensing and GIS. Photogramm. Eng. Remote Sens. 70, 617–628. doi:10.14358/PERS.70.5.617 Sarro, R., Mateos, R.M., García-Moreno, I., Herrera, G., Reichenbach, P., Laín, L., Paredes, C., 2014. The Son Poc rockfall (Mallorca, Spain) on the 6th of March 2013: 3D simulation. Landslides 11, 493–503. doi:10.1007/s10346-014-0487-8 Schuster, R.L., Fleming, R.W., 1986. Economic losses and fatalities due to landslides. Bull. Assoc. Eng. Geol. 23, 11–28. Searle, M.P., Khan, M.A., Fraser, J.E., Gough, S.J., Jan, M.Q., 1999. The tectonic evolution of the Kohistan-Karakoram collision belt along the Karakoram Highway transect, north Pakistan. Tectonics 18, 929–949. Shahabi, H., Hashim, M., 2015. Landslide susceptibility mapping using GIS-based statistical models and Remote sensing data in tropical environment. Sci. Rep. 5, 9899. doi:10.1038/srep09899 Shit, P.K., Bhunia, G.S., Maiti, R., 2016. Potential landslide susceptibility mapping using weighted overlay model (WOM). Model. Earth Syst. Environ. 21, 1–10. Shroder, J.F., Bishop, M.P., 1998. Mass movement in the Himalaya: New insights and research directions. Geomorphology 26, 13–35. doi:10.1016/S0169-555X(98)00049-X Sletten, K., Stalsberg, K., Rubensdotter, L., 2015. DEebris slides and debris flows [WWW Document]. Geol. Surv. Norw. URL https://www.ngu.no/en/topic/debris-slides-and-debris-

125

References ______

flows (accessed 11.10.18). Soeters, R., van Westen, C.J., 1996. Slope instability recognition, analysis, and zonation. Landslides Investig. Mitig. Süzen, M.L., Doyuran, V., 2004. A comparison of the GIS based landslide susceptibility assessment methods: Multivariate versus bivariate. Environ. Geol. 45, 665–679. Taherynia, M.H., Mohammadi, M., Ajalloeian, R., 2014. Assessment of Slope Instability and Risk Analysis of Road Cut Slopes in Lashotor Pass, Iran. J. Geol. Res. 2014, 1–12. doi:10.1155/2014/763598 Tahirkheli, R.A.K., 1982. Geology of the Himalaya, Karakoram and Hindukush in Pakistan, Geological bulletin / University of Peshawar. National Centre of Excellence in Geology, University of Peshawar. Tahirkheli RAK, Jan, M., 1979. Preliminary geological map of Kohistan and adjoining areas, Northern Pakistan. Toppe, R., 1987. Terrain models — A tool for natural hazard mapping. Proc. Davos Symp. Sept. 1986 629–638. Treloar, P.J., Petterson, M.G., Jan, M.Q., Sullivan, M.A., 1996. A re-evaluation of the stratigraphy and evolution of the Kohistan arc sequence, Pakistan Himalaya: implications for magmatic and tectonic arc-building processes. J. Geol. Soc. London. 153, 681–693. doi:10.1144/gsjgs.153.5.0681 Ulbricht, K.A., Teotia, H.S., Civco, D.L., 1993. Supervised Classification to Land Cover Mapping in Semi-Arid Environment of NE Brazil Using Landsat-TM and SPOT Data [WWW Document]. Int. Arch. Photogramm. Remote Sens. URL http://www.isprs.org/proceedings/XXIX/congress/part7/821_XXIX-part7.pdf (accessed 2.16.16). USGS Earthquake Catalog, 2017. USGS Earthquake Catalog [WWW Document]. Earthq. Cat. URL https://earthquake.usgs.gov/earthquakes/search/ (accessed 7.23.17). Vakhshoori, V., Zare, M., 2018. Is the ROC curve a reliable tool to compare the validity of landslide susceptibility maps? Geomatics, Nat. Hazards Risk 9, 249–266. doi:10.1080/19475705.2018.1424043 Vallejo, G.L., Ferrer, M., 2011. Geological Engineering, 1st ed. CRC Press/Balkema, AK Leiden. Van Westen, C.J., van Asch, T.W.J., Soeters, R., 2006. Landslide hazard and risk zonation - Why is it still so difficult? Bull. Eng. Geol. Environ. 65, 167–184. doi:10.1007/s10064-005-0023-0 Varnes, D.J., 1978a. Slope movements types and processes. In: Schuster R.L., y Krizek R.J. (ed.). Landslides analysis and control. Washington D. C. doi:In Special report 176: Landslides: Analysis and Control, Transportation Research Board, Washington, D.C. Varnes, D.J., 1978b. Landslide hazard zonation : a review of principles and practice. 1-63, Paris. Wang, Q., Li, W., Chen, W., Bai, H., 2015. GIS-based assessment of landslide susceptibility using certainty factor and index of entropy models for the Qianyang county of Baoji city, China. J.

126

References ______

Earth Syst. Sci. 124, 1399–1415. Wang, W. neng, Wu, H.L., Nakamura, H., Wu, S.C., Ouyang, S., Yu, M.F., 2003. Mass movements caused by recent tectonic activity: The 1999 Chi-chi earthquake in central Taiwan. Isl. Arc 12, 325–334. doi:10.1046/j.1440-1738.2003.00400.x Williams, M.P., 1989. The Geology of the Besham area, North Pakistan: Deformation and Imbrication in the footwall of the Main Mantle Thrust. Geol. Bull. Univ. Peshawar 22, 65–82. Wrachien, D. de, Mambretti, S., 2011. Assessment of debris flow magnitude in small catchments of the lombardy alps: the val gola case study. Agric. Sci. 02, 9–15. doi:10.4236/as.2011.21002 Xu, C., Xu, X., Yao, Q., Wang, Y., 2013. GIS-based bivariate statistical modelling for earthquake- triggered landslides susceptibility mapping related to the 2008 Wenchuan earthquake, China. Q. J. Eng. Geol. Hydrogeol. 46, 221 LP – 236. doi:10.1144/qjegh2012-006 Yalcin, A., 2008. GIS-based landslide susceptibility mapping using analytical hierarchy process and bivariate statistics in Ardesen (Turkey): Comparisons of results and confirmations. Catena 72, 1–12. doi:10.1016/j.catena.2007.01.003 Yamamoto, H., Rehman, H.U., Kaneko, Y., Kausar, A.B., 2011. Tectonic stacking of back-arc formations in the Thelichi section (Indus valley) of the Kohistan arc, northern Pakistan. J. Asian Earth Sci. 40, 417–426. doi:10.1016/j.jseaes.2010.09.006 Yang, Z.Q., Zhu, Y.Y., Zou, D.H.S., 2011. Formation conditions and risk evaluation of glacial debris flow disasters along International Karakorum Highway (KKH) 1031–1037. doi:10.4408/IJEGE.2011-03.B-112 Zhiquan, Y., Yingyan, Z., Shanguang, Q., Kepeng, H., 2016. Types and Space Distribution Characteristics of Debris Flow Disasters Along China-Pakistan Highway. Electron. J. Geotech. Eng. 21, 191–200. Zhou, S., Chen, G., Fang, L., Nie, Y., 2016. GIS-based integration of subjective and objective weighting methods for regional landslides susceptibility mapping. Sustain. 8, 1–15. doi:10.3390/su8040334

127