Design of a GIS-based System for Landslide Hazard Management

Design of a GIS-based System for Landslide Hazard Management San Antonio del Sur, Cuba, case study

Enrique A. Castellanos Abella May, 2000

Design of a GIS-based System for Landslide Hazard Management San Antonio del Sur, Cuba, case study

By

Enrique A. Castellanos Abella

Thesis submitted to the International Institute for Aerospace Survey and Earth Sciences in partial fulfilment of the requirements for the degree of Master of Science in Applied Geomorphological Surveys.

Degree Assessment Board: Chairman: External supervisor: First supervisor: Second supervisor: Member: Member:

Prof.Dr. J.L. van Genderen Dr. A.C. Seijmonsbergen Dr. C.J. van Westen Drs. R.P.G.A. Voskuil Dr. C. Phol Dr. A. Sharifi

INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES ENSCHEDE, THE NETHERLANDS

ABSTRACT

Abstract This research has a combination of two objectives: landslide hazard assessment and the management of the landslide hazard map in a GIS context. For reaching these two goals many subjects were covered, large data set was produced and analysed and some theoretical issues were also studied. The research was done using as case study an area in San Antonio del Sur, Guantánamo province, Cuba. For the applied geomorphological assessment a terrain mapping was done, surveying 601 units from aerial photographs. Using different remote sensing products as a background the terrain mapping units were digitised on the computer screen. A more detailed mapping was done for those areas, with landslides in order to classify the different landslide types and zones. From this survey a Geomorphological map of the study area was obtained. The different landforms and the causative factors for landsliding were also analysed. Together with the terrain mapping units other data was generated in order to complete a data set for landslide hazard assessment. The variables were: slope angle, internal relief, shape of the slope, geological formation, active faults, drainage, springs, geomorphological subunits and landslide zones. With these variables a hierarchical heuristic model was applied in which three hierarchical levels of weights were designed for classes, variables and criteria. The model combines all the weights into one single hazard value in each pixel of the landslide hazard map. Later, the landslide hazard map was classified using two scales of hazard classes; one with three major hazard classes for disaster managers and one with 10 detailed hazard classes for planners or researchers with more detail information. As a support of the landslide hazard information a Natural Hazard Management System was designed in a CARIS GIS framework linked with an Access database. The system is useful for storing, updating, reporting, analysing, modelling and managing the hazard-related data. The maps are all together in one single file with the features linked by keys to the relational database in Microsoft Access format. The design allows queries in both directions from maps to relational database and viceversa. A customised interface controls the possible options for the user and it is possible to execute macros or external programs or modules. For the design of the management system a theoretical study was done from which was conclude that GIS is a useful tool in disaster management. Among other reasons the GIS advantages include the diversity of types and amount of data, the analysis capability and the customisation and modular design possibilities.

INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

III

ACKNOWLEDGEMENTS

Acknowledgements While finishing the thesis it is difficult to remember how everything was in the beginning and at the time to write the acknowledgements it is difficult to memorise all the people involved in this result. Then, before starting the list of "participants", I would like to apologise for any person I forget to mention. I say that especially because I believe that everybody is important as a part of the whole creativity environment, from the staff members to the receptionist who remember me every night that the building will be closed. Technically some people have to do with the thesis directly. My supervisor Dr. Cees van Westen follows the entire thesis and suffered the fieldwork conditions together with me. He also checked all the spelling for every chapter of my thesis (except the Acknowledgements), as I am not good at all in English. I really would like to express my sincere gratitude to him and I promise to improve my English grammar. Drs. R. Voskuil helped me in some parts of the thesis and he listened to me during the nights at ITC in several discussing points. He and Drs N. Kingma taught me the most valuable parts of my ITC course and I would like to thank them a lot for their friendly guidance. Dr. Simons contributed a lot to the quality of maps and I am very grateful for his helps. Many thanks to Helios Jellema who assist me in several times for any very simple doubts concern with many technical problems. Dr. Sharifí appeared in the last part of the work with very useful comment and information about the hierarchical model, many thanks for his support. In Cuba, during the fieldwork, I received the indispensable support of my organisation, the Institute of Geology and Paleontology (IGP), many thanks for it. The assistance provided by the Ecological Station in Baitiquirí and Civil Defence at municipal, provincial and national level was crucial as well. I would like to express my thanks for their willingness in their assistance. I am very grateful to Dr. S. E. Masry (President and CEO of Universal Systems) for encouraging my research and give me the possibility to use the CARIS GIS package during the thesis. I would like to "glued" in one group to all ITC students and staff friends who made me feel more comfortable during my staying in Enschede out of my country. Thanks to all of them for listening to me and being always agreeing with my complaints as good friends, particularly my group and to the Cuban community in Enschede. Thanks at lot. I give a special word of thanks to my family who keep it touch with me all time, some of them even every day. They encourage me to do my best and specially to "finish" the thesis. I wish they knew I am doing the best I can and I will always appreciate the way they love me.


IV

CONTENTS

Contents Abstract .................................................................................................................................................. iii Acknowledgements .................................................................................................................................iv Contents ...................................................................................................................................................v List of figures ....................................................................................................................................... viii List of Tables ...........................................................................................................................................x 1 Introduction ......................................................................................................................................1 1.1 The problem definition ................................................................................................................1 1.2 Research objetives .......................................................................................................................2 1.2.1 Research topics and questions .............................................................................................2 1.2.2 Research objetives................................................................................................................3 1.2.3 Hypothesis............................................................................................................................3 1.3 Research Methodology ................................................................................................................4 1.4 Study Area location......................................................................................................................6 2 GIS for Natural Disaster Management.............................................................................................8 2.1 Introduction ..................................................................................................................................8 2.2 The nature of disasters .................................................................................................................9 2.3 Mapping natural disasters ..........................................................................................................11 2.4 Management the natural disasters..............................................................................................12 2.5 The GIS technology ...................................................................................................................15 2.6 GIS for natural disaster management.........................................................................................17 2.7 Summary ....................................................................................................................................19 3 Geology and Tectonic Setting of the study area ............................................................................21 3.1 Introduction ................................................................................................................................21 3.2 Previous research .......................................................................................................................23 3.3 The northen ophiolite belt..........................................................................................................23 3.4 Cretaceous volcanic arc .............................................................................................................24 3.5 Paleocene Middle Eocene Volcanic Arc....................................................................................25 3.6 Late Middle-Latest Eocene Piggyback Basin ............................................................................25 3.7 First Transgression-Regression..................................................................................................26 3.8 Second Transgression-Regression .............................................................................................26 3.9 Third Transgression-Regression ................................................................................................27 3.10 General tectonic setting..........................................................................................................27 3.11 Summary ................................................................................................................................30 4 Geomorphology of the study area ..................................................................................................31 4.1 Introduction ................................................................................................................................31 4.2 Previous research .......................................................................................................................31 4.3 Description of the geomorphological complexes.......................................................................33 4.3.1 Coastal Hills.......................................................................................................................34 INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

V

GIS DESIGN FOR LANDSLIDE HAZARD MANAGEMENT

5

6

7

8

VI

4.3.2 Accumulational Slopes.......................................................................................................36 4.3.3 Limestone Hills ..................................................................................................................38 4.3.4 Terrigenous Hills................................................................................................................38 4.3.5 Metamorphic Hills..............................................................................................................39 4.3.6 Alluvial Valleys..................................................................................................................39 4.3.7 Caujeri Depression .............................................................................................................40 4.4 Towards the landform evolution modeling in the study area.....................................................41 4.5 Summary ....................................................................................................................................43 Data preparation and processing ....................................................................................................44 5.1 Introduction ................................................................................................................................44 5.2 The digital elevation model........................................................................................................44 5.3 Image processing........................................................................................................................48 5.3.1 Landsat TM ........................................................................................................................49 5.3.2 SPOT PAN .........................................................................................................................51 5.3.3 JERS-1 SAR .......................................................................................................................52 5.4 Some data fusion ........................................................................................................................55 5.5 The photointerpretation..............................................................................................................57 5.6 Other input data..........................................................................................................................58 5.7 Summary ....................................................................................................................................59 Landslides in the study area ...........................................................................................................61 6.1 Introduction ................................................................................................................................61 6.2 Previous research .......................................................................................................................61 6.3 General overview about the classification .................................................................................62 6.4 Coastal Landslides .....................................................................................................................64 6.5 Landslides in Denudational slopes.............................................................................................66 6.6 Landslides in "Puriales de Caujeri" scarp ..................................................................................68 6.7 Other Landslides ........................................................................................................................71 6.8 Summary ....................................................................................................................................72 Landslide Hazard Assessment........................................................................................................73 7.1 Introduction ................................................................................................................................73 7.2 The main causative factors for landsliding ................................................................................74 7.3 The heuristic landslide prediction model for the study area ......................................................76 7.4 The landslide hazard in the different areas ................................................................................83 7.4.1 Landslides hazards in the coastal areas..............................................................................83 7.4.2 Landslides hazards in the Denudational Slopes boundary.................................................83 7.4.3 Landslides hazards in the Puriales de Caujery scarp .........................................................83 7.5 Flooding hazards in the study area.............................................................................................84 7.6 Towards the risk mapping in study area.....................................................................................85 7.7 Summary ....................................................................................................................................88 Natural Hazard Management System in the study area .................................................................89 8.1 Introduction ................................................................................................................................89 8.2 Data avaliable and database structure ........................................................................................90 8.3 The CARIS system structure......................................................................................................96 8.4 Analysis capabilities attribute database to map .........................................................................97 8.5 Analysis capabilities map to attribute database .........................................................................98 INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

CONTENTS

8.6 Running an example...................................................................................................................99 8.7 Summary ..................................................................................................................................101 9 Conclusions and recommendations..............................................................................................102 9.1 Introduction ..............................................................................................................................102 9.2 Main conclusions about the Landslides problem in study area ...............................................102 9.3 Main conclusions about the Natural Hazard Management ......................................................104 9.4 Recommendations for further investigations ...........................................................................105 9.5 Final remarks for ITC...............................................................................................................105 References............................................................................................................................................106


VII

LIST OF FIGURES

List of figures Figure 1-1. Economic and insured losses by disasters with trends. (Munich Re, 1999) .........................1 Figure 1-2. Flow-chart with the general framework of the study. ...........................................................2 Figure 1-3. Thesis Flow-chart ..................................................................................................................6 Figure 1-4. Location of the Case Study Area San Antonio del Sur. ........................................................7 Figure 2-1. Disaster Management Cycle..................................................................................................8 Figure 2-2. Natural disasters observed during the years 1970-1998 (Ingleton, J., 1999) ........................9 Figure 2-3. Flow chart of different steps and concepts for landslides disasters. ...................................11 Figure 2-4. Relations between external applications and data with CARIS GIS...................................15 Figure 2-5. Illustrative representation of the problem in connecting GIS for NDM. ............................18 Figure 2-6. Three-axis graph representing the GIS applicability for NDM...........................................19 Figure 3-1. Main geological units of Cuba (from Iturralde-Vinent, M. A.; 1994) ................................21 Figure 3-2. Sketch map representing the foldbelt and the neoautochthon in Cuba (Taken from Iturralde-Vinent, 1996)...................................................................................................................22 Figure 3-3 Main Geological units and lineaments in the study area......................................................24 Figure 3-4. Oriente strike-slip fault and the submarine relief at the south of the study area.................28 Figure 3-5. SW-NE, N-S and NW-SE lineaments systems in the study area.........................................29 Figure 4-1. Anaglyph Image of the study area. ......................................................................................32 Figure 4-2. Geomorphological complexes in the study area..................................................................34 Figure 4-3. Anaglyph images and field photographs of different complexes. From top to bottom: Coastal Hills, Limestone Hills and Terrigenous Hills. ..................................................................35 Figure 4-4. Different coastal hills profiles. ............................................................................................36 Figure 4-5. Anaglyph images and field photographs of different complexes. From top to bottom: Accumulational Slopes, Metamorphic Hills and Caujeri Depresion. ............................................37 Figure 4-6. Anaglyph image and field photograph of Alluvial Valley complex. ..................................40 Figure 4-7. Coastal Hills and pouch shape evolution hypothesis (taken from Keijzer, 1945). .............42 Figure 5-1. Flow chart for DEM generation. .........................................................................................45 Figure 5-2. Fragment of vectorization raster lines with SAMI. The black pixels are contour lines in raster and the inner grey lines are new vectors traced by the system. ...........................................46 Figure 5-3. Georeference differences using the Spline method with 43, 36 and 34 points. ..................47 Figure 5-4. Original contour lines and different smoothing order (2, 5, 7). Enlargement 125x............48 Figure 5-5. Spectral ranges (in digital numbers) of different features in Landsat TM bands................49 Figure 5-6. Three-dimensional view with Landsat TM color composite 457 (RGB)............................50 Figure 5-7. Histogram of the original SPOT PAN image. .....................................................................51 Figure 5-8. Two Laplace plus filters designed in ILWIS.......................................................................52 Figure 5-9. Different Sar images processed of a Coastal Hills. Files according to the tables from top to bottom, left to right: sar (original), sar3, sar6, sar7, sar10, sar12, sar14, sar16, sar zoom out by 2, sar zoom out by 4. See text for explanation. ..................................................................................54 Figure 5-10. Data fusion transformation flowchart for DEM-SPOT PAN fusion.................................55 INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

VIII

LIST OF FIGURES

Figure 5-11. SPOT PAN image and DEM fusion. Left, whole study area. Right, window at full resolution (10 meters per pixel). ....................................................................................................56 Figure 5-12. Photocenters of aerial photographs in the study area........................................................57 Figure 6-1. Various Coastal hills landslides. (see text for explanation)................................................66 Figure 6-2. Landslides in Denudational slopes. (See text for explanation)...........................................67 Figure 6-3. Landslides in Caujerí scarp. (See text for explanation) ......................................................69 Figure 6-4. Different profiles in Caujerí scarp ......................................................................................70 Figure 7-1. Lateral groundwater pressure when water table rises. ........................................................75 Figure 7-2. Causative factors for landslide occurrence in the study area. .............................................76 Figure 7-3. Flowchart for heuristic landslide hazard analysis. ..............................................................77 Figure 7-4. Components of the heuristic landslide prediction model....................................................77 Figure 7-5. Caujerí Scarp from Los Jagueyes landslide to the north direction. ....................................84 Figure 7-6. Housing density map in the study area................................................................................86 Figure 7-7. Graph with number of houses per hazard class...................................................................87 Figure 8-1. TMU application initial window.........................................................................................91 Figure 8-2. Access application form for entry of lithology for geological units. ..................................91 Figure 8-3. TMU Access editing form...................................................................................................92 Figure 8-4. Relationship among the tables of the TMU dadabase.........................................................93 Figure 8-5. Landslide Database editing form.........................................................................................94 Figure 8-6. Landslide database tables relationship................................................................................95 Figure 8-7. Example of DEM-related fields of TMU database. ............................................................95 Figure 8-8. CARIS icons suite ...............................................................................................................96 Figure 8-9. Query condition form ..........................................................................................................98 Figure 8-10. Query Condition form example.........................................................................................99 Figure 8-11. Part of query result list. ...................................................................................................100 Figure 8-12. Graphical query result. ....................................................................................................100 Figure 8-13. Query result of the seven landslides................................................................................101


IX

LIST OF TABLES

List of Tables Table 2–1. Example of software and its applicability in disaster management.....................................18 Table 3–1. Geological formations in San Antonio del Sur area and their context in the geological evolution of Cuba. ..........................................................................................................................22 Table 5–1. Satellite metadata summary .................................................................................................48 Table 5–2. Statistics for Landsat TM bands. .........................................................................................49 Table 5–3. Statistics for SPOT Panchromatic image.............................................................................51 Table 5–4. Processing table of speckle suppression. Best results in bold case. ....................................53 Table 5–5. Processing table of Wallis Adaptive filter. Best result in bold............................................53 Table 5–6. Processing table of Luminance Modification. Best result in bold.......................................53 Table 6–1. Landslide types and subtypes using in the geomorphological survey. ................................62 Table 6–2. Landslide zones and subzones surveyed for 1:50,000 mapping scale. ................................63 Table 7–1. Variable used in the prediction model. See text for explanation. (N/A- No Applicable)....78 Table 7–2. Weights for criteria and variables using three methods.......................................................79 Table 7–3. Weights intervals for the three hazards maps ......................................................................79 Table 7–4. Initial and final weights per classes. ....................................................................................80 Table 7–5. Final Landslides Hazard classes its statistics, characterisation and hazards remarks. ........82 Table 7–6. Counting houses per hazard class ........................................................................................86 Table 7–7. Statistics by crossing the roads with hazard classes. ...........................................................87 Table 8–1. Summary of maps or images used in the research. .............................................................90


X

INTRODUCTION

1 Introduction 1.1

The problem definition

Landslides are considered as Natural Hazards, producing serious constraint on economic development and injury or death to people. With time these consequences have been increasing as the statistics shows in the Figure 1-1.

Figure 1-1. Economic and insured losses by disasters with trends. (Munich Re, 1999)

On the other hand, practice has shown that adequate hazard mitigation is possible (ITC/AGS annual report 1998, 1999). To improve the effect of landslides mitigation measures it is necessary to create a landslide hazard map. But even when a landslide hazard map has been created the use of the map information and its combination with other types of information is also a problematic topic. GIS can be an effective approach to design, visualisation, analysis, combine and manage landslide hazard information. The problems to be dealt with include: -

Lack of knowledge and transparency of the transformation process from the geomorphology map, terrain classification map, and landslide hazard assessment to the final landslide hazard map.


1


-

Expert rules that are not well developed and reclassification methods in the landslide hazard assessment to create landslide hazard maps.

-

Access capability to relevant data and combination in a digital version of the landslide hazard map in GIS.

-

The position and role of hazard maps in the disaster management systems context is not well defined.

1.2

Research objetives

1.2.1

Research topics and questions

The thesis was developed in the Applied Geomorphology discipline, specifically in the area of interest of Geomorphological Landslide Assessment and its digital management. The research topics covered in this area were (Figure 1-2): Geomorphological Mapping, Terrain Classification or Terrain Mapping Units (TMU), Aerial Photointerpretation for Landslide Hazard Mapping, Landslide Hazard Map, GIS for Landslide Hazard Assessment, Emergency Management System Development.

Gathering Data techniques Tematic Maps

Socio-Economic Data

Landslide Hazard Assessment Risk Analysis

Landslide Hazard Map Landslide Hazard Management System Disaster Management System

Figure 1-2. Flow-chart with the general framework of the study. The main questions answered are:

•

2

What are the steps and the transformation processes from the geomorphological map up to landslide hazard map? INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

INTRODUCTION

•

How to create expert rules and what are their characteristics to assess the landslide hazard types and degrees?

•

What is the best design for a GIS DataBase on landslide hazard mapping in order to be used by a disaster manager?

•

In which part is located the hazard map in the disaster management systems context and what are the relations with the other part of EMS?

1.2.2

Research objetives

Main Objective: 1. Design of a dedicated GIS based system for Landslide Hazard Management using the CARIS Software. Secondary Objectives:

2. Determine the principal factors, which generated landslide and the most susceptible parameters in the study area. 3. Define the main geomorphological processes and the terrain features occurring in the study area. 4. Study the relationship between the geomorphology and the landslide hazard map with emphasis on the expert rules to build the landslide hazard map. 5. Make a geomorphological landslide hazard map of the case study area San Antonio del Sur. 6. Analyse a framework for an Integrated Disaster Management System and provide an easy

interface for display, queries, reporting and analysing of the landslide and related information. Long-term Objective: Contribute to the development of using hazard maps in the Disaster Management using GIS technology approach. 1.2.3

Hypothesis

This thesis covers two hypotheses. One is on Disaster Management context and one is about the causative factor for the landslides in the case study area. 1. Landslides Hazard Management-Disaster Management System Hypotheses


3


It is possible to integrate in one system all the elements involved in the disaster occurrence and its consequences in order to obtain an “Integrated Disaster Management System”, which can be used more efficiently in the four Disaster phases: Mitigation, Preparedness, Response and Recovery. The elements to be integrated are: • Information technology and communication systems, • Spatial analysis and decision support systems in GIS, • Applied geomorphology for Landslides Mapping, • Mitigation measures for disaster prevent and, • Disaster management. 2. Case Study Area Hypotheses

The most probable factors involved in the Landslide occurrence in the Case Study Area San Antonio del Sur are a combination of: 1) rainfall and groundwater level rising during rainy season, 2) lithology competence differences and 3) a faulting system in combination with uplift movements. Then, when the rainfall becomes intensive the groundwater level rise over the lower parts of the main scarps. The hydrostatic pressure pushes away the rock wall on the scarps, provoking landslides. There are others landslides, seem to be much older, which may be originated by earthquakes. It is also important to note the seismicity as a triggering factor, but with the same value of intensity, because the seismic hazard maps of the CENAIS indicate the same earthquake hazard degrees for the all case study area. 1.3

Research Methodology

The methods applied are mainly descriptive and analytical. The descriptive studies describe objects, situations, factors, events and/or conditions with respect to persons, place and time (Kaewsonthi and Harding, 1992). The descriptive methods in this thesis were used to describe the area, the factors and conditions in relation to the landslides, the processes of the Landslide Hazard Assessment, etc. The analytical studies are designated to investigate hypothesis concerning with relationship between a condition and factors contributing to that condition (Kaewsonthi and Harding, 1992). The analytical studies in this will investigate the two hypotheses explained before. The methods explained by the objectives are as follow: 1) Integrated Hazard Management System

4


INTRODUCTION

This objective has descriptive and analytical parts. The state of art about Disaster Management Systems, GIS use for Disaster Management, GIS use for natural hazard assessment and other topics are described in Chapter 2. After that, the relation between these elements and the effective framework for the design of an Integrated Disaster Management System is analysed in order to test the respective hypothesis. 2) Landslide Factors. The landslide factors are such as groundwater, rainfall, tectonic scarp, etc. and the conditions are

when the groundwater level rise or when the rainfall reach certain values. This objective has descriptive and analytical components. First of all, the different factors are qualitatively described in order to estimate the influence in intensity and extension. Secondly, the relation between the factors and conditions has been analysed, recognising those whom are involve and/or playing priority role in the different types of landslide in the study area. At this point the second hypothesis was tested. 3) Geomorphological Processes. This objective is also descriptive. During the geomorphology and terrain classification mapping a number of geomorphological processes were recognised from the data available. These processes are described as much as possible using the data processed and the fieldwork check. The purpose is increasing the knowledge about the landscape processes, which are going on in the area and are affecting the relief. 4) Landslide Hazard Assessment Processes. This objective is descriptive. The purpose is to describe the relationship (differences, similarities and links) in the processes between the Geomorphological Map-Terrain Classification Map- Landslide Hazard Assessment and the Landslide Hazard Map. The main focus was oriented in how to design the expert rules to build the Landslide Hazard Map. 5) Geomorphological Landslide Hazard Map. This map is descriptive in nature but was generated using the heuristic analysis like Landslide factors/conditions. The map was the combination of Geomorphology Map, Terrain Classification Map and the Landslide Hazard Assessment (Figure 1-2). It was done using aerial photointerpretation and image processing techniques together with GIS modelling of the Digital Elevation Model (DEM) and drainage analysis. In terms of Landslide Hazard Zonation techniques, the methods used were the geomorphological analysis, which is a heuristic analysis (van Westen, 1993). This method is also


5


known as "direct method" where the Hazard is determined directly in the field by the geomorphologist after the photointerpretation was done. The methodology carried out to make the Geomorphological Landslide Hazard Map in this thesis is explained in the Flow Chart on Figure 1-3.

As the Figure 1-3 shows, the Geomorphology, Terrain Classification and Landslide Map will be created using the integrated interpretation of the topomap products, image processing, aerial photointerpretation and geological map interpretation. After these three maps has been created the Geomorphological Landslide Hazard Map was done using a heuristic model.

Geological Map

Landsat TM

Spot PAN

JERS-1 SAR

Aerial Photographs

Image Processing

Faults

Lithology

Topomap Drainage

Other data

DEM

Houses

Shadows

Roads

Landuse

Rainfall Epicenters Landslide

F i e l d w o r k Faults

Lithology

TMU

Landuse

Geomorphology

Geomorphology

Disaster management System Natural Hazard Mapping Hazard Management System

C h e c k i n g Landslide Hazard

Literature Review

Morphometry

GIS Design, Input, Validation, Manipulation and Implementation

Geomorphological Landslide Hazard Map

Landslide Hazard Management System

A priori Landslide Risk Map

Figure 1-3. Thesis Flow-chart

1.4

Study Area location

The study area is located in the eastern province of Cuba: Guantánamo. The province has several landform terrains. Since low lands (swamps, alluvial valleys) like Guantánamo Bay basin to high lands (rocky mountainous areas) like Sierra del Purial. In term of climatic conditions the area is also divide in highly contrasting regions since the province has the highest and the lowest rainy areas in the country. 6


INTRODUCTION

The study area, San Antonio del Sur municipality and its surrounding, also contain such contrasting characteristics. The area is located at 60 kilometres of the capital’s province Guantánamo with an extension of 600 square kilometres. The boundaries in Cuba Sur Coordinate System are: Lower left corner: Lower right corner: Upper right corner: Upper left corner:

700000, 152000 720000, 152000 720000, 182000 700000, 182000

The access to the area is mainly by the road in between the San Antonio del Sur town and Puriales de Caujery town. The most important city close to the area is Guatánamo City, the capital of the province, at 62 km. Figure 1-4.

Caujerí valley

San Antonio del Sur

Figure 1-4. Location of the Case Study Area San Antonio del Sur.


7

GIS FOR NATURAL DISASTER MANAGEMENT

2 GIS for Natural Disaster Management 2.1

Introduction

A Natural Disaster may be defined as the occurrence of a natural phenomenon, which affects the community, its infrastructure and facilities. Due to that effect, the community needs to respond against the phenomenon or its consequences. This implies the fact that wherever the natural event happened in the absence of the population, infrastructure or man-used land, it is not considered a disaster. But, whenever a disaster has happened, the men have to manage with it. Disaster Management is an applied science which seeks, by systematic observation and analysis of disasters, to improve measures relating to prevention, mitigation, preparedness, emergency response and recovery (Carter, N. W., 1991). Knowing the temporal dimension of the disasters, disaster management has been organising in four phases called disaster management cycle: Prevention, Preparedness, Response and Recovery (Figure 2-1).

Figure 2-1. Disaster Management Cycle. Any kind of natural disaster is spatially represented and it can be changing in the temporal space. The forest fires, the flooding, the landslides can be mapped as polygon areas. These areas have an exact location that can be move, grow or decrease with the time. Besides these movements, these areas interact with many other spatial features like natural vegetation, floodplains (natural features) or routes, building, agricultural lands (non-natural features). All these spatial features can become valuable data


8


and information when it is surveyed and digitally represented in computer systems like Geographical Information System (GIS). Taking into account the spatial and temporal component of disasters information, the GIS may cover almost all-current tasks in the different phases of the natural disaster management cycle. The point here is, which data is more important for natural disaster management, how this data should be surveyed and structured, what kind of GIS analysis tools should be used and what is the optimum GIS framework design for NDM? To cover the study, the chapter has been divided in five parts: the nature of the disasters, mapping natural disaster, management the natural disaster, GIS technology and GIS for natural disaster management. The next sections will cover these parts.

2.2

The nature of disasters

A Disaster (ND) is an event, sudden or progressive, which impact with such severity that the affected community has to respond taking exceptional measures (Carter, N. W., 1991). A disaster can be called Natural when the cause of such event is purely natural. The concept is simple and clear but the problem is large and complex. The first step to clarify disaster problems is to study diagnostic parameters such as type, magnitude, frequency, time and space of the natural event to dealing with. 100% 90% 80%

10

3

7 6

26

17

60%

Flood Earthquake

29

51

Others Windstorm

46

70%

50%

3

0.2

66

40% 30%

51

20% 10% 0%

42

28

Number 163

21 Deaths 1.1 milliom Economic Losses 522 US$bn

Insured Losses 104 US$bn

Figure 2-2. Natural disasters observed during the years 1970-1998 (Ingleton, J., 1999) Natural disasters cover a wide range of natural phenomenon from “earth-related” disasters like earthquakes, volcanic activity, and landslides to “meteorological-related” disasters like tsunamis, tropical cyclone, floods, drought and wildfire. Although the dicotomy is not clear-cut since floods and landslides have meteorological causes and volcanic eruptions can have serious climatic consequences (Coppock, J. T., 1995). INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

9


Each of these disasters has particular characteristics, which determine the way they are analysed and mitigated dealt with. But all of them cause thousands of deaths and billion of dollars lost every year. However, the incidence of these phenomenon on the society and infrastructure is not equally distributed. So, the windstorms and earthquakes are the most cause of damage disasters according to a study carried out between 1970-1998 (Figure 2-2). The magnitude of a disaster can be considered as the energy released by the physical event and it expresses the size of that event. The disaster can be also characterised by its intensity: period of time in which the energy is released, as an expression of how concentrated was the event (Martin, J.G. and Dieter, S.B., 1999). On the other hand, the frequency can be defined as a period of time in which an event, with a specific magnitude (or interval of magnitude), uses to occur with certain probability. The magnitude and frequency of disasters vary from one place to another and usually these properties are statistically inversely related in a way that higher magnitude disasters occur at lower frequency values. For example, the active floodplain of a river can be inundated every rainy season, the first river terrace every 5 years and the next terrace level every 20 years and so on. The fact of this relation does not imply that after a disaster of 20-year frequency has occurred, the next one will not happen for another 20 year with the same magnitude. There are also differences in frequency between different disasters. For example, volcanic eruptions occur less often in Central America as the average of five hurricanes, which arrive to Caribbean Sea every year. Another important issue in the disasters is the temporal and spatial evolution. Alexander (1993) points out seven temporal phases of a disaster: 1. 2. 3. 4. 5. 6. 7.

The incubation period which can be monitored for certain phenomenon. The disaster impact properly. The brief unmitigated crisis period. Search and rescue operations. Repairing basic services for fundamental necessities. Restoration-reconstruction period. Developmental reconstruction.

It is important to note that the disaster impact time covers a wide range of time from a few seconds (like earthquakes) up to years (like desertification). Also some disasters like volcanic eruptions can last minutes or years. For that reason, one aspect to be considered is the scale of time for measuring the natural events and their consequences. Closely related with the time property of disaster is the spatial develoment. The first spatial consideration in relation to disasters is whether or not the cause of the disaster is located in the same place of the disaster. For example many flood disasters are originated by intensive rainy upstream far from the area affected. Spatially, some disaster have circular concentration of magnitude (like earthquakes), with other ones the magnitude can reach long distance and is controlled by climatic factors like hurricanes, tornadoes and ash, lavas from volcanic eruptions. There is another group of disasters where damage is more controlled by the current relief, e.g. landslides, floods and tsunamis. 10



All these parameters are important issues for the design of a GIS for NDM, which must take in consideration the spatial characteristic and temporal behaviour of the natural events to be managed. Since the spatial data are spatially referenced, the spatial topology has to be maintained in the updating process (Roshannejad, A.A., 1996). The characterisation of disaster properties in relation with the GIS analysis and management tools is still a research subject to be developed.

2.3

Mapping natural disasters

Nowadays, the Disaster Management Policy of different countries has been directed to improve the technology as tools, techniques and facilities by those who take operational decisions to improve the knowledge and/or response to natural disasters (AGCI Workshop, 1996). One of the most important advances introduced are the new techniques for mapping natural disasters. The theoretical basics for this advance were established by UN-UNESCO (Varnes, D. J., 1984) in 1984. The graphs in the Figure 2-3 represent the concepts and its relations applicable for landslides. Event

Enviromental Parameter

Natural Disaster: Landslides, Floods, Erosive Processes

Geology, Soil, Landuse, Slope, Height Internal relief

Tiggering Factors

Earthquakes and Rainfall

Elements at Risk

Population, Industry, Agriculture, Infrastructure

Vulnerability

Susceptibility

Degrees of loss to a given element(s) at risk resulting from the occurrence of a naural phenomenon of a given magnitude.

Susceptible areas for the ocurrence of a natural phenomenon

Natural Hazard Probability of occurrence within a specified period of time and within a given area of a potential damaging phenomenon.

Specific Risk Expected degree of loss due to a particular natural phenomenon.

Total Risk Definitions from Varnes, 1984

Expected number of lives lost, persons injured, damage to property, or disruption of economic activity due to a particualr natural phenomenon.

Figure 2-3. Flow chart of different steps and concepts for landslides disasters. Although the flow chart shows the general overview for landslide, there are particular differences in correspondence with the type of disaster. Some hazards can thus be a secondary consequence of another, as when an earthquakes triggers landslides (Coppock, J. T., 1995). The first step in taking any mitigation measures is to assess the hazard (Carter, W. N.; 1991), which is the basis for the analysis of vulnerability and risk. All these maps play an important role in all phases of natural disaster management and not only in the mitigation phases.


11


There are different ways to creating each of these maps and the methodologies cover a wide range of approaches. Van Westen, C., (1993) classified the landslides hazard zonation techniques in four types: • • • •

Inventory analysis. The landslide distribution, activity and density are surveyed to get general conclusions about the behaviour of the event. Heuristic analysis. Expert opinions based analysis. Statistical analysis. Many map- parameters are mapped to apply Bivariate o Multivariate statistical analysis Deterministic Analysis. Physical and chemical models are executed to produce probabilistic maps.

This classification can also be valid for most of the other disasters and reach up to the risk step. One of the problems here is that many research and projects finish in the printed or digital hazard or risk map. In that way, it has wasted the opportunity to present the information in a digital-manageable way (like GIS context) and to integrate this information with all other data from the area for an actual disaster management and planning.

2.4

Management the natural disasters

As was mentioned before, the Natural Disaster Management (NDM) can be subdivided into four phases (Figure 2-1), which are in a cycle. Some times the cycle is overlayed by another cycle because some disasters occur when the consequences of the last one were not completely recovered. In the Prevention (Mitigation planning) phase the focus is to develop activities to eliminate or reduce the occurrence of a disaster. Since the natural phenomena cannot be prevents from happening, most of the prevention measures are oriented to reducing the vulnerability of the future disaster. The prevention phase is maybe where the GIS had played the most active role. With the historic disaster inventory, the environmental parameters, triggering factors and elements at risk it is possible to do modelling in a GIS context to obtain the Susceptibility, Vulnerability, Hazard and Risk maps. They are very valuable for NDM since they provide information about the spatial variation of a potential disaster, how it can affect the human environment and which spatial strategy plan can be developed to reduce the potential disaster. The main tasks in prevention phase are: • Susceptibility, vulnerability, hazard and risk mapping. • Mitigation planning. • Building codes, building use regulation and ordinances management. • Resource allocation and control. • Tax incentives/disincentives. • Zoning and Landuse planning. • Public education.

12



The Preparedness phase develops plans to minimise disaster damage whereas mitigation measures can not avoid the disasters. The preparedness phase GIS can be used for the emergency planning and its dissemination to the different organisations involved in NDM. Another important application is to model the loss estimation and produce the emergency plans for different disaster scenarios and intensity. The main tasks in preparedness phase include: • Preparedness and emergency planning. • Forecast, warning and communication systems. • Evacuation plans and training. • Resources inventories. • Public information and education. When the disaster occurs the Response phase is activated to provide emergency assistance for disaster casualties and damaging. The response phase GIS can almost immediately map the extent of the impacted areas and manage the search and rescue operation, the activation of the emergency plan and the distribution of the resources taking into account the best routes available. It is possible to know which population needs to be evacuated, how many and so on. Some tasks in the response phase are: • Activation public warning and notification of authorities. • Mobilise resources, personnel and equipment. • Medical centers and shelters. • Detect affected areas. • Search and rescue operations. Finally the Recovery phase intent in short and long term to return the life to normal levels. In the recovery phase GIS is useful for the damage assessment survey and recovery analysis. In this sense all post-disaster data surveyed is integrated in a GIS and from its statistics will show which areas have been affected and how much it is. Then, it is possible to model a recovery plan and running it in GIS. The main tasks in recovery phase are: • Damage assessment, insurance, loans and grants. • Temporary housing and long term medical care. • Reconstruction and developing programs. • Economic impact studies. The monitoring task is applicable for some disasters and covers all the phases. Another tasks will also appear depending on the type of disaster. For example, for flooding special attention is taken with the construction of dams and bridges. Looking back to these tasks it possible to recognise the potential possibilities of GIS in the whole disaster reduction process. Since the maps generated in the Prevention phase and combined with all other data, it is possible to improve, for several reasons, the NDM.


13


The other dimension of the NDM is the different levels of management. In spite of there being different classifications, it is recognised at least three levels: national, subnational (regional) and local. The OAS (1991) recognise some main GIS applications for each levels that were improved in the following list: In the national level: • Give a reference to the overall hazard situation and help to identify areas that need further studies to assess the effect of natural disasters on natural resources management and development potential. • Used to identify less hazard-prone areas most apt for development activities. • Identify areas where mitigation strategies should be prioritised. • Recognise the number of people or type of infrastructure at risk. • Keep updates of the national inventory of disasters. • Useful for disaster with national magnitudes likes hurricanes. In the subnational level: • Integrate and update the inventory of disasters and the hazards and risk evaluations. • Used to formulate less vulnerable development activities and/or mitigation strategies to lessen vulnerability to acceptable levels. • Determine the conditions under which disasters are likely to occur (susceptibility). • Useful for site location studies for large projects like dams, highways, power plans, etc. • Project integrated regional development plans. In the local level: • Creation and management of susceptibility, vulnerability, hazard and risk maps. • Used in prefeasibility and feasibility sectorial project studies and natural resources management activities. • Landuse planning. • Help planners to identify specific mitigation measures for high-risk investment projects. • Locate vulnerable critical facilities for implementation of emergency preparedness and response activities. • Identify critical resources in high-risk areas and adequately formulate mitigation strategies. • Test "loss estimation models" for different disasters and several scenarios. • Creation of preparedness and emergency plans testing evacuation routes and reallocation places like hospital, shelters. Control mobilising resources and equipment. • Control disaster locations and Search and Rescue operation. • Determinate the most affected areas by damage assessment. • Useful in the designing of reconstruction and development programs. Many other applications can be described and in fact some of these applications are currently carried out on GIS, but in common the lack of integration in the analysis process provoking results overweighting for some disciplines like planning, forestry, geologist, etc.

14



2.5

The GIS technology

The Geographical Information System (GIS) or applications has been growing dramatically in the last ten years. Covering almost all different spatial applications: landuse and cadastral, marine charting and topographic, geology and geomorphology etc. The two main advances of GIS in relation to NDM are the integration and analysis capabilities. GIS data models accept the transferring of real world features to some kind of spatial data structure (like vector or raster). Beside the limitations of the current data models, it is possible to integrate in one system several type of natural features such as geology, soil, landuse (areal maps) or road, power lines, aqueducts (linear features) or well, electric towers, sample points (point features). For Natural Disaster Management combine different sources of information is crucial due to the multidisciplinary and multidimensional characteristic of the problems. Disaster information is needed by decision-makers at many different levels and different scales. The Board on Natural Disasters of National Research Council in USA (National Research Council, 1999) have classified the information resources for decision making on disasters in six principal types: Base data, Scientific, Engineering, Economic, Environmental and Response data. They are starting to survey those who have which data? and in which condition (scale, accuracy, etc. ) the data are? The integration is not only in relation with data, but also with the system itself. The GIS for NDM should be able to couple with many external applications and to exchange (in both in and out directions) data and information. For doing that, the system itself should have a modular design in a way that some modules can be activated or allocated for disaster management depending on a) type of disaster to manage, b) level of decision to handle and c) phase of disaster management. For example, in case of "flooding disaster", can be activated the module for "evacuation zones" at the "local level". The conceptualisation of this modular structure is somehow the same of Open GIS Consortium but at the application level.

C A R I S CARIS EDITOR • Graphical editor • Building topology

CARIS DB INFORMATION MANAGER MANAGER • Graphical Analysis • Interface Design • Model Design

• Queries Design • Project Design • Atribute Queries

ODBC

DB drivers DB Format

Attribute Relational data base (In any DB Format)

Graphical Object’s ID

Input & Editor map

CARIS REPORT

Reports Output

Applications in the DBMS

External Aplications

DDE

External Applications

External Data

Figure 2-4. Relations between external applications and data with CARIS GIS.


15


For an explanation in how it can be done the Figure 2-4 shows some possibilities of the CARIS GIS (Universal System Ltd., 1998). The external applications can be either for graphical and attribute data or for both. The database remains in its own format and it accessible by the ODBC (Open Database Connectivity) application interface for database accessibility. The external applications can be coupled to the system using the Dynamic Data Exchange (DDE) program and transfer either the information for decision making or the results of the processing. They can be Decision Support Systems (DSS), Expert Systems (ES), Environmental and Management Models or any other type required for Disaster Management. All these components can be glued in one nonexpert interface in the CARIS Information Manager where the disaster manager can start up any application and make the decisions depending of the results. In terms of GIS decision tools for NDM was established that "the ultimate aim of is to provide support for making spatial decisions" (Malczewski, 1999). He explains how and in what extent can GIS provide the support required at each of the three stages of decision making: intelligent, design and choice. The summarised applicability is explained in the next three paragraphs. In the Intelligent stage, where the problem is defined, the spatial decision problem is the difference between the desired and existing state of a real-world geographical system, in this case the hazard, the risk or simply the disaster. GIS provide a unique opportunity to solve problems traditionally associated with data collection and analysis more efficiently and effectively. Moreover, it can also effectively present vast amount of information in a comprehensive form to decision makers. In the Design stage, where the action alternatives are designed, the spatial decision alternatives are derived by manipulation and analysis of data and information stored in the GIS. In most of the current GIS, the modelling techniques required for decision makers are not suitable enough and the decision models operate in the background depending on the user' s skills. Due to these problems, there is a need for integrated decision analytical techniques and GIS functions by incorporating analytical models directly into GIS or connecting GIS with an existing decision analysis system. Integrating decision support techniques into GIS was partially cover by some authors (see Eastman, 1995). In the Choice stage, where the alternatives are evaluated and selected, is critical the capabilities of the GIS in incorporating decision maker' s preferences (e.g. weights assignment). However, GIS is very useful in the solution of spatial decision conflict areas. For the choice stage GIS will increase its applicability aggregating the value assignation techniques into the GIS procedures. In general, despite that GIS can provide good support for decision making, the procedures still require spatial analysis skills, which most of decision makers do not have. In many cases the improvement, only provide an easy interface and some functions from where the user can easily solve a spatial decision problem. In particular, for Natural Disaster Management, there is still the necessity of surveying the different types of decision making techniques required for disaster management according to the different tasks

16



in each phase and to the different levels. Later it is also important to research how they can be implemented in a GIS context.

2.6

GIS for natural disaster management

Getting information about the use of Geographical Information System (GIS) for Natural Disaster Management (NDM) is not easy task for two main reasons. One, there are obvious copyright reasons because some companies do not supply additional information about the software they sell and two, there exists many applications at different governmental levels which do not have information available on the Internet or any publication. However, there are a great number of individual GIS applications for the solution of different isolated tasks, mostly in hazard, vulnerability and risk mapping, evacuation planning and loss estimation. Cova (1999) presented an illustrative paper about the roles of GIS in emergency management. In his paper are described phase by phase the different applications already done. However, as was mentioned before, most of these applications only present as result a set of printed maps and a list of recommendations, wasting the opportunity of providing a system for the management of the surveyed data. On the other hand, "the contribution of GIS should therefore be seen in terms of a whole sequence of disasters reduction, from identifying areas at risk and monitoring and forecasting hazards, through warning of their onset and measures (both short- and long term) to minimise loss of life, injury and damage to property, to coping with disaster once it has occurred" (Coppock, 1995). There are a number of systems available for disaster (emergency) management on the Internet, which try to cover the market on this subject. Table 2–1 shows a summary table in relation with the applicability of the software surveyed from the Internet. Some applications are not GIS in strict sense but have somehow a connection with GIS modules or data (Desinventar, Telesafe, and Complete Street). Other applications are the integration of different components: - A customised interface in a GIS with high integration capability. - A properly GIS system. - An (or part of) expert system, or decision support system supporting the models. - A database containing all data and information. Whatever it is, the most peculiar feature of these examples is the fact that they are "activity oriented" and not "tools oriented" like the simple GIS. The implication of this is that many GIS are actually a collection of tools as they were designed for the original functionality and what they should be is a collection of activities (tasks) for Natural Disaster Management.


17


Software Levels Type of Phase Disaster

Tasks

Company or supplier

Desinventar

All

All

Mitigation

Inventory and queries

La red

Complete Street EIS/GEM

Local (urban) All

All

Recovery

CITITECH Systems

All

Preparedness, Recovery

NBC Warning!

All

Mitigation, Recovery

Nhematis

Local

Mitigation

Risk assessment

Nobility Software Inc.

Nobility EM

All

Nuclear, Biological and Chemical Earthquake, Tornados, Floods and Landslides All

Damage Assessment, reconstruction programs Emergency Planning, warning and comm. systems, evacuation plans, resources control, rescues operations and others Vulnerability and risk, damage assessment

Mitigation

Nobility Software Inc.

PlantSafe

Local (industry)

Hazardous Materials

TeleSafe

All

All

Preparedness, Response and Recovery Response

Mitigation planning, monitoring Emergency Planning, damage assessment, evacuation plans Communications systems

Geosphere

Local

Hazardous

Response

Damage assessment,

Essential Technologies

Essential Technologies

Geosphere Emergency Response systems Inc. Geosphere Emergency Response systems Inc. Geosphere Emergency

Table 2–1. Example of software and its applicability in disaster management. As it shows in Table 2–1, although some systems can be applicable for "All" levels and "All" types of disaster they do not cover "All" phases in NDM. This implies that there is not linking between the applications for Prevention phase and the Preparedness, Response and Recovery phases. Subdividing this in two fields, from one side there are GIS applications for Disaster Mapping (Hazard, Risk, etc.) and for the other side there are few GIS applications on Natural Disaster Management. The Figure 2-5 graphically shows the two "walls" or "conflict connecting lines" in this problem. These two "walls" are the main subjects of research nowadays for solver these connectivity problems.

Disaster

Disaster Mapping

Disaster Management

Disaster Management Systems

Figure 2-5. Illustrative representation of the problem in connecting GIS for NDM. 18



There are two erroneous paradigms in relation with this. One, some times it is considered "disaster management" at the time when the disaster had happened and two, there is not full understanding in the potential use of hazard, risk mapping in the disaster reduction cycle life phases. As was mentioned in the abstract, from our point of view, the applicability of GIS for Natural Disaster Management can be represented in a three-axis graph (Figure 2-6). One axis represents the scale of management: local, subnational (provincial, departmental, etc.) and National. Floods Landslides Recovery Response Preparedness Mitigation Local

Subnational

National

Figure 2-6. Three-axis graph representing the GIS applicability for NDM. Some applications can be applicable at any level because the task to solve and, consequently, the decision models are not depending on the scale of the data, but most of the other ones are highly scaledependent. The other axis (the time axis) represents the different phases of Natural Disaster Management. The applicability of GIS should be found in the tasks attached to each phase. There are some GIS applications that can be used in different phases, for example, the Loss estimation-Rapid Damage Assessment models. In the Prevention phase, the model can be useful for planning because it is possible to run the model for different disaster scenarios. In the Response phase the model may be used for a rapid damage assessment with the real data of the disaster. Finally the last axis, the type of disaster, represent the necessity of the applications to take into account the disaster characteristics and behaviours (magnitude, frequency, scale, time, causes, etc.). The box shape of the graph indicates that the GIS for NDM may be designed in a modular way. Meaning that each sub-box (a module) is a GIS application for one type of disaster, at one level and for one phase. The complete NDM implementation can be obtained step by step and whole system may be integrated in one changeable interface within which the specific modules will respond to the specific necessities.

2.7

Summary

As many authors have pointed out, it is clear that the number, size and impact of Natural Disasters is increasing. This is due to a number of reasons that can be grouped into social (population growth, urbanisation, etc.); technological (constructions, designs, etc.) and environmental (sea level rise, climatic change, etc.). As a result, the number of casualties, the amount of damage to the infrastructural INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

19


systems, the disruption of utilities and the number of people displaced from their homes is also on the increase. It is important to note the fact that the worst disasters occur in the poorer countries contributing to the increasing of their poverty since they need to reorient the resources and effort to the recovery. Any type of Natural Disaster is inherently a spatial-temporal problem, since it occurs in “an area” and in “a period” of time. Disaster Management, consisting of several phases such as prevention, preparedness, response and recovery also has a strong spatial component. For this reason Natural Disaster Management (NDM) is mainly dealing with geographical space analysis and decision making, and therefore, this is where GIS appears as an important tool for NDM. The importance of GIS for NDM is relevant in two major aspects: 1) the analytical capability for decision making and 2) the data integration capacity. Both aspects allow the integrated analysis of large amounts of different data in each disaster phase. The use of GIS for NDM can be graphically represent as a three-axis applicability graph. The three-axis in this cases will represent a) the type of disaster (flood, landslides, etc.), b) the level of application (National, Medium, Local) and c) the disaster phases (Prevention, Preparedness, etc.). Taking into account these three axes, new applications of GIS for NDM have appeared in recent years as a result of effort in reducing disaster losses. However, most of these applications are only concerned with one type of disaster at one level and for one or two phases of disaster management. Moreover, some GIS applications aim to finish by producing susceptibility, hazard, vulnerability or risk maps instead of integrating these maps with all other information for the real disaster management. As a result of these problems, disaster managers do not generally use the hazard and risk maps and a lot of effort in disaster mitigation and prevention is actually lost. Therefore, it is concluded the necessity of carry out research to study in detail the applicability of GIS technology for NDM, providing as a main outcome, the design of a framework for a dedicated GIS based system for NDM, able to be handled by non-GIS experts (subnational disaster managers). The research should focus on the optimum ways of producing susceptibility, vulnerability, hazard and risk maps considering that they need to be more understandable and transparent for decision-makers. Another issue should be how to integrate these maps with landuse, cadatral, and social economic information for the operability of a GIS based system in each disaster management phase. The system to be designed should consist of several modules for different phases of DNM, in which a GIS is the central platform. The modules can be integrated and produce a user-friendly interface for Natural Disaster Management. The result may be applicable in developing countries at subnational levels for disaster managers with not GIS expertise. Due the wide range of disaster types the research start focus on one or two main disaster types (e.g. floods and landslides), although there are many general issues, which cover almost all natural disasters. The system intends to provide a unique platform for Disaster Management, useful for several purposes such as integrated disaster mitigation planning.

20


GEOLOGY AND TECTONIC SETTING

3 Geology and Tectonic Setting of the study area 3.1

Introduction

The geology and tectonic setting of the eastern of Cuba is rather complicated and still many "question marks" remains in the complete understanding of the geological and tectonic evolution. The region records, in a relative small area, several geological and tectonic environments, which were going on from the Late Jurassic to the recent time. The different tectonic and structural processes have been overlaping over the geological times in such a way that it is difficult to separate them spatially and temporally. Moreover, the area remains as an active tectonic zone on the north boundary between Caribbean and North America plates, provoking a general uplift and changing the landform with many neotectonic features. In general the geology of Cuba has been subdivided in two principal geological units: a foldbelt and a neoautochthon (Iturralde-Vinent, 1996), which unconformably overlies the foldbelt. The complete subdivision of the subunits is show in the Figure 3-1. Third transgression-regression (Pliocene to recent) Neoautochthon

Second transgression-regression (Lower Miocene to Late Miocene) First transgression-regression (Latest Eocene to Oligocene) Mesozoic Bahamian platform and slopes deposits Continental units

Paleocene-Late Eocene foreland basin Cuban SW terranes (Guaniguanico, Pinos and Escambray)

Foldbelt

Late Middle Latest Eocene piggy back basin Paleocene-Middle Eocene volcanic arc Oceanic units

Latest Cretaceous-Late Eocene piggyback basins Cretaceous (?Aptian-Campanian) volcanic arc Northern Ophiolite belt

Figure 3-1. Main geological units of Cuba (from Iturralde-Vinent, M. A.; 1996) The Figure 3-2 shows the sketch map with the distribution of the foldbelt and the Neoautochthon. It is also represented the main faults system which strongly contributed to the current relief landforms.


21


Figure 3-2. Sketch map representing the foldbelt and the neoautochthon in Cuba (Taken from Iturralde-Vinent, 1996) Specifically, in the study area San Antonio del Sur, there are geological units from the foldbelt and from the neoautochthon. The foldbelt is consisting of the by Northen Ophiolite belt, Cretaceous Volcanic arc, Paleocene Middle Eocene volcanic arc and Late Middle Latest Eocene piggyback basin. The Neoautochthon was called to the recent geological units formed "in situ" and in the study area is totally represented by the three Transgression-Regression steps since the Latest Eocene to recent for some geological formations. Although the neoautochthon is characterised by transgression/regression movements, the main feature throughout this time is the overall uplift of the area. Table 3–1 shows the geological formations mapped in the study area and their classification according to the geological evolution of Cuba. Main geological units Neoautochthon

Main Geological subunits

Geological formations

Age

Third TransgressionRegression

Alluvial Deposits

Holocene

Palustre Deposits Marine Deposits Alluvial-Colluvial Deposits Fm Maya Fm Cabo Cruz

Holocene Middle-Late Pleistocene Pleistocene-Holocene Late Pliocene - Early Pleistocene Middle Upper - Late Miocene

Fm Yateras Fm Maquey

Lower Oligocene - Early Miocene Late Oligocene-lower Early Miocene

Mb Cilindro Fm San luis

Late Oligocene-lowest Early Miocene upper Middle-Late Eocene

Fm San Ignacio Fm Charco Redondo Gp El Cobre

upper Middle Eocene Middle Eocene Paleocene-lower Middle Eocene

Fm Sierra del Purial

Early Cretaceous (Aptian?) - Late Cretaceous (Campanian) Mesozoic

Second TransgressionRegression First TransgressionRegression

Foldbelt

Late Middle - Latest Eocene Piggyback Basin

Paleocene-Middle Eocene Volcanic Arc Cretaceous Volcanic Arc Northern Ophiolite belt

Complejo Ofiolitico

Table 3–1. Geological formations in San Antonio del Sur area and their context in the geological evolution of Cuba. 22



Sedimentary and metamorphic rocks of several lithologies cover the study area. Both, carbonated and terrigenous, sedimentary rocks occur in the study area. The metamorphic rocks are, actually, sedimentary and volcanic rock, metamorphized in very low degree and high pressure. The geological formations are part of the Terrain Mapping Units DataBase (see Chapter 8 Natural Hazard Management System in the study area), meaning that, they can be directed referenced in the relational data base by a simple query creating any subsequent map in the Natural Disaster Management System (NDMS). The following section will explain the previous work done in the study area and the geological formations in the different stages of the evolution.

3.2

Previous research

The area was mapped by two main geological mapping projects. In 1976 (Nagy et al.) finished a 1:100 000 scale geological mapping project for the whole former Oriente province of Cuba. In 1983 (Nagy, et al.) was published the main results of the mapping project establishing most of the current geological formations in the study area. Later in 1981, another geological mapping at 1:50 000 scale was done (Nuñez, et al. 1981), covering 75 percentage of the study area, leaving without mapping 5 kilometres at the western part of the study area. This mapping project introduced new details in the geological mapping, but the geological formations where more or less the same than the previous mapping project. Additional to these two geological mapping projects, many geological researches have been done and published in the study area during last decades. It is important to note the researches developed by Millan, and Somin, (1972, 1981 and 1985) characterising the metamorphic rocks in the area and in Cuba in general. Another important publication is the Stratigraphy Nomenclature of Cuba (Franco, et al. 1992), which explains in detail all the geological groups, formations and members in the whole national territory. From this document was summarised most of the description of the geological units of this chapter. Finally, this thesis was developing at the same time that another student Kenya Nuñez Cambra was doing her M.Sc. in the study area. Her thesis focused on the geology and tectonic setting of the same study area. Because at the time of writing of this thesis, information from her was not available, the author refers to her thesis for further information on the geology and tectonic setting.

3.3

The northen ophiolite belt

The ophiolites are represented in the study area by the hilly zone called "Sierra del Convento" (see Figure 3-3). Although its lithology and structure is very similar to other ophiolites in the rest of the country, its location and position still remains a controversial topic between geologists and geophysics. Using the geophysical data it was recognised (Chang and Suarez, 1998) that the ophiolitic body is larger than its appearance in the surface and dipping to the Southwest. Then, the body is considered as a relic of the basement, which emerged due a spreading of the oceanic crust, pushed from the south by INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

23


tectonic forces. In general, the ophiolites are ultrabasites serpentinized. Its ultramafic rocks are harzburgite, dunite, lherzolite and wherlite.

3.4

Cretaceous volcanic arc

The Cretaceous volcanic arc in the study area is totally represented by the geological formation Sierra del Purial (sp). The Fm Sierra del Purial covers all east and north-east of the study area. This geological formation is consisting of Andesite-basalts, basalts, tuff, lavabreccias, andesite-dacite, polymictic sandstone, sandstone derived from granitoids and limestone lenses. As was mentioned before, all these rocks are metamorphized in very low degree and high pressure.

Figure 3-3 Main Geological units and lineaments in the study area.

24



The rocks underlying Sierra del Purial formations are unknown and It is overlain unconformably, either by piggyback basins formations from Eocene and by neoautochthon units. The thickness is larger than 1000 m and the geological formation was develops in a marine environment of deep to medium waters with submarine volcanism. Age: Lower Cretaceous (Aptian ?) - Upper Cretaceous (Campanian).

3.5

Paleocene Middle Eocene Volcanic Arc

The El Cobre (ec) group represents the Paleocene volcanic arc in the study area. This formation is widely distributed in eastern Cuba. In the study area, El Cobre is located mainly in the north part, together with its piggyback basin formations. The lithology is very variable, from volcanogenic to volcanogenic-sedimentary rocks, with both, gradual and sharp transitions. The most common are tuff, agglomerate tuff, lava and agglomerate lavas. The composition is andesite, andesite-dacite, dacite and rarely rhyolite, rhyolite-dacite and basaltic. There is intercalation of tuffites and limestones, and, less abundant, sandstone. It is also possible to find hypabisal bodies and dikes of different composition. The El Cobre group subdivided in two formations (El Caney and Pilón formations) and an undifferentiated part, however in the study area only the undifferentiated part is found. The thickness of the unit is around 5000 to 6000 m and was developed in a marine basin with variable depth. Age: Paleocene-lower Middle Eocene.

3.6

Late Middle-Latest Eocene Piggyback Basin

The piggyback basin corresponds to Paleocene-middle Eocene volcanic arc. The formations Charco Redondo, San Ignacio and San Luis represent this basin in the study area. These formations cover the Northeast up to central part of the study area (see Figure 3-3). The Charco Redondo (chr) formation consist of compact bioclastic limestones. In the lower part of the formation the breccias are common, where the stratification is thicker. The total thickness is between 50 and 200 m and the formation was deposited in shallow seawater in seashore to sub-seashore environment. Age: Middle Eocene. The San Ignacio (si) formation was located at the border of Sierra del Purial formation. The lithology is mainly polymictic breccias with fragments of green schist, phyllite and serpentine in a clay matrix. The thickness is not more than 700 m. The formation was developed in a submarine bank and slope with probable tectonic origin. Age: upper Middle Eocene. The San Luis (sl) formation was widely developed in the eastern part of Cuba. It is mainly terrigenous of origin with polymictic sandstones, mudstones, marls, clays, limestone clays, bioclastic limestone, sandy limestone and polymictic conglomerates. The stratification is well developed and, in the upper part of the unit, clastic material is more frequent. The San Luis formation is transected by basalt bodies and dikes. The thickness is 700 m and the formation was originated in deep seawater, which later becomes medium to shallow seawater. Age: upper Middle Eocene-Late Eocene.


25


3.7

First Transgression-Regression

The Neoautochthon starts with the first transgression-regression cycle from Late Eocene to Oligocene. In the study area there is one formation (Maquey) and one of its member (Cilindro) representing the first transgression-regression cycle. The Maquey (mq) formation can be found in different parts of the study area, wherever the overlying units are eroded. Then, it is very common in areas close to drainage system or to the scarps. The lithology is mainly alternating layers of sandstone, mudstone and calcareous clay of grey colour. There are white marls, which contain intercalations of biodetritic limestone, sandy limestone and gravel limestone with white yellow and occasional greyish yellow colour. The stratification is slight to medium, less frequently thick and massive. Some layers contain gypsum and lignite. In some parts, Maquey formation has lateral transition to Yateras formation. Age: Late Oligocene-lower Early Miocene. The Cilindro (ci) member belongs to Fm Maquey and is its basal member. This member is developed in areas close to Puriales de Caujery town and it is formed by polymictic conglomerates with lensshape and, sometimes, crossed stratification. It is weakly cemented and has lignite-bearing sandstone lenses. The matrix is bearing-carbonated polymictic sandstone. The thickness is in the order of tens of meters and was develop in deltaic and sub-seashore environment. Age: Late Oligocene-lowest Early Miocene.

3.8

Second Transgression-Regression

The second transgression-regression is represented in the area by two geological formations: Yateras and Cabo Cruz. Because Yateras formation has lateral transition to Maquey formation, there is also transitional relation between the first and second transgression-regression. The Yateras (yt) formation covers almost the entire western part of the area, wherever it is not eroded to show the underlying geological formations. It is composed of alternating layers of clastic, bioclastic and biogenic limestone, with different grain sizes and stratification thickness (even massive). It is hard and has variable porosity with white, rose and brownish colours. The total thickness is between 160 and 500 meters. This formation is mainly reef deposits covering different varieties of reef complex. Age: Early Oligocene-lower Early Miocene. The Cabo Cruz (ccz) formation in the study area covers the top of the coastal hills. It is important to note that there is certain doubt about the mapping of this formation in this area. It is because the formation was mapped by one geological mapping project, but not for the other one; and the holostratatype area for this formation is in the occidental part of western of Cuba, very far from this area. However, Cabo Cruz formation was consider in this paper because it is clear that on top of the coastal hill there is another geological units different from the one in the marine terraces slopes. The lithology is clay bearing bioclastic limestone with reddish coloration. Because weathering process, sometimes they become in secondary marls and in pseudoconglomerates. The total thickness is more than 200 meters and was develop in a sub-seashore environment. Age: upper Middle - Late Miocene.

26



3.9

Third Transgression-Regression

The third transgression-regression in represented in the study area by five formations and deposits from the Pliocene to recent. The Río Maya (rm) formation is composed by algae-bearing bioherm limestones, with corals, generally hard and a fine grain matrix. In the study area Rio Maya is mapped almost totally in the marine terraces. The limestones are frequently dolomitized. There are clays in variable content and intercalation of fine-size polymictic conglomerates with calcareous cement. The colour is white, yellow, rose and greyish. The thickness is between 30 and 80 meters and the formation was created in a combination of reef development and detritus coming for the land. Age: Late Pliocene - Pleistocene Early. The Alluvial-Colluvial Deposits are located mainly in lower side of the Sierra de Caujeri scarp (see Geomorphological Map) and in the intramountain deposits between the Sierra de Mariana and the coastal hills. The deposits are rests of Yateras and Maquey formations and, therefore, contain its materials in way of detritus. The weathering and sliding affect the origin composition of the rocks becomes clays, calcareous gravel, sandy clay, etc. The material has been transported not longer than 3 km from the source in these deposits. The Marine Deposits in the study area are located accord the current coastline. They are bioclastic and biogenic limestone, calcareous sandstone with grey and greyish colour. The deposits and weakly consolidated. These deposits were called Jaimanitas formation in the past. Age: Middle-Late Pleistocene. There are Palustre Deposits in the inland part around the Bays of Baitiquirí and Sabanalamar. The origin is due the concentration of organic material coming from drainage system and accumulates close to mouth of the river. The material are either carbonate, terrigenous or peat and there are swamp mangrove. Age: Holocene. The Alluvial deposits in the study area are mainly located in the floodplain of the main river Sabanalamar. The lithology is grey and greyish mud, sandy mud and sandy clay. Some materials are currently eroded and others are re-erosion of the Alluvial-Colluvial Deposits. Age: Holocene.

3.10 General tectonic setting Taken into account the latest structural classification of eastern Cuba (Flores, et al., 1998), the study area can be divided in three main structural zones: Metamorphic Terrain Purial-Asuncion, Paleogene Magmatic Arc and San Luis-Guantánamo Basin. Although, this classification does no consider areas like ophiolites in Sierra del Convento, can be linked with the general geological evolution explained before (see Figure 3-1 and Table 3–1).

The Metamorphic Terrain Purial-Asuncion corresponds to Cretaceous Volcanic Arc rocks. The type of metamorphism (low degree-high pressure) is different from the one, which is produced in the front


27


part of subduction zones of the volcanic arcs. Meaning that the cause of metamorphism should be another one than the subduction movement in the volcanic arc itself. The Paleogene Magmatic Arc is the upper part of the Cuban foldbelt and is integrated by the Paleogene volcanic arc and its piggyback basin. As is concluded this volcanic arc and part of his basin, was cut through and displaced, the entire southerner part since the Oligocene-Miocene, to what is now, La Española (Haiti and Dominican Republic Island). The reason of this movement is the sinitral strikeslip Oriente fault, which pass through the south part of the study area (in the sea) and has many secondary faults reflecting inland. The three transgression-regression steps of the neoautochthon compose the San Luis-Guantánamo Basin. This basin was also cut by Oriente fault and has a subhorizontal stratification. Since from the Oligocene, the Caribbean plate changes the direction and originated the strike-slip Oriente fault, many secondary faults were originated and present remarkable features in the current landform.

Figure 3-4. Oriente strike-slip fault and the submarine relief at the south of the study area. In the study area three main lineaments systems can be distinguished: SW-NE, N-S and NW-SE. All three have geomorphic expressions recognisable in the aerial photographs, satellite images and digital elevation model. The SW-NE system seems to be the most active one and displaced the N-S system. This system may be a series of normal faults in a sequential distension stress dipping to SE. The sequence changes the strike from south to north, from SW-NE to W-E. The distentional stress is more evident on the Sierra de Mariana where at least four lineaments can be clearly tracing. 28



The southward lineaments represent large scarps where landslides like occurring mainly in the SW part of the study area, at the north of Baitiquirí town. The southernmost lineament is the limit between the Sierra de Mariana and the accumulational slopes at the north part of Baitiqurí and San Antonio del Sur towns. This scarp has strong geomorphic expression in the relief in several parts and supposedly can be trace even over the metamorphic rocks. In the northward lineaments of the SW-NE system, are more W-E oriented and some of them can even be trace on the colluvial and alluvial deposits of the Caujeri valley.

1 3

2 2

3 1

Figure 3-5. SW-NE, N-S and NW-SE lineaments systems in the study area In the coastal hills, lineaments are more doubtful because in southern slopes are shaped by the marine terraces and in the northern slope the sliding materials covers up the bottom. However, it is recognisable that the coastal hills are lined up approximately from SE to NW in three SW-NE lineaments: Baitiquirí-San Antonio del Sur, Baitiquirí-El Naranjo and San Antonio del Sur-Macambo respectively. The N-S system controls the main scarp of Sierra de Caujeri with height differences around 500 meters. This system turns to the ENE in the north part and presents some parallel lineaments in the Yateras formation. The lineament may be normal fault dipping to east and is displaced in the south part INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

29


by the SW-NE system. There is not clear indication about the connection of N-S and the SW-NE systems, although the area where they seem to be connected, at the north of Baitiquirí, is abundant in landslides. Finally, the NW-SE system seems to be the oldest one because is displaced by the N-S system. In previous publications this system is part of the large lineament Miraflores-Riito-La tinta (Flores, et al., 1998), however in the study area seem to be the third-order system and its continuation inside of the ophiolites is not very clear. This system has a branch point in the northwest of the area, at the north of Mameyal region. In the branch point is clear how the N-S system displace the NW-SE system.

3.11 Summary The geology of Cuba has been subdivided in two principal geological units: a foldbelt and a neoautochthon, which are separated at the end of the Eocene. The study area presents geological formations in both units. The geological formations, which belong to the foldbelt, are mapped in the eastern part of the study area while the neoautochthon formations cover the west and the south (coastline) part. The formations of the area can be grouped in seven geological events: the ophiolites, Cretaceous volcanic Arc, the Paleogene Volcanic Arc and its piggyback basin (relate to the folbelt) and three transgression-regressions (relate to the neoautochthon). Both volcanic arcs present volcanic and sedimentary rocks. The piggyback basin is mainly composed of terrigenous materials with bioclastic limestone, polymictic breccias, sandtone, mudstone, etc. The transgression-regression of the neoautochthon start with terrigenous materials in the first transgression, changing to mainly biogenic carbonates deposits in the next two transgressions. Tectonically the area can by subdivide in three main systems: SW-NE, N-S and NW-SE being the displaced one each other in the same order respectively. The first two systems show large scarps with abundant landslides.

30


GEOMORPHOLOGY

4 Geomorphology of the study area 4.1

Introduction

The geomorphology of the study area, as well as of the whole island, is conditioned by: • The Caribbean-North American inter-Plate zone, • Climatic conditions, • The paleoclimatic oscillations during the Quaternary period. In general, the landforms were created by the combination of horizontal and vertical movements. In the Meso-Cenozoic Era, before The Middle Eocene, the movements were predominant horizontal and compressional acting, with folding and thrusting of several types of rocks into parallel belts along the island. After the Middle Eocene the movements were predominant verticals and characterised by strike-slip fault and extensional movements, sedimentation and blocks structures. The thrust-belts made large orographic units, but, even when they are important, the morphostructures are controlled by the strong neotectonic development, especially in the study area. Figure 4-1 shows the Anaglyph image of the study area where the relevant recognisable uplifted landforms can be distinguished. Due to climatic, tectonic and lithological factors, the study area has different landforms from the coast to the north. The coastline presents the more intensive semiarid region in Cuba (annual average temperature of 26 degrees Celsius and annual precipitation less than 600 mm) while inland, around Puriales de Caujeri town, tropical rain forests are predominant (annual average temperature of 23 degrees Celsius and annual precipitation more than 2000 mm). Tectonically, the area is the convergence of three principal systems, which seem to be still active. The lithology differences are also extraordinary, since the area presents several alternating layers of sedimentary rocks and the eastern part is almost totally covered by metamorphic rocks, which belong to a volcanic arc with many rock types.

4.2

Previous research

There are not many papers on the geomorphology of the study area at detailed scale. One important publication was the last edition of the Nuevo Atlas Nacional del Cuba (Gutierrez, 1989), where the relief of the island was described according to many genetic characteristics. The atlas presents several maps regarding the relief and its genesis, with many general characteristics. Another important publication (Magaz et al., 1991) describes the southern (coastal) parts of the study area and gives an inventory of the different landslides from Baitiquirí to Maisi (the easternmost corner of Cuba).


31


Figure 4-1. Anaglyph Image of the study area.

32


GEOMORPHOLOGY

4.3

Description of the geomorphological complexes

The applied geomorphological survey in the study area was done with the photointerpretation and later, the field checks. In the interpretation the area was subdivided in 603 Terrain Mapping Units (TMU). A terrain mapping unit groups the following aspects (Meijerink, 1988): • geomorphologic origin and physiography • lithology • morphometry • soil geography According with the specific interests of this research each unit was differentiated in the stereo image on the basis of one or more of the following criteria: • geomorphology origin • specify origin • morphometry • lithology Additionally two more aspect was recorded: the geomorphological process and the complex. The first three criteria were store as origin, main unit and subunit in the database. With them, a geomorphological map was created (see map appendixes) at 1:50 000 scale. In the map the boundaries of the original terrain mapping units remain, although some of them have the same geomorphological origin and therefore the same color. Another complementary and linked database was designed specially for the units, which are landslides of part of it. In the landslides database an extra 8 data was recorded for 296 units. The explanation of the landslides is on the Chapter 6, dedicated to the landslide problems in the study area. There was recorded also in the data base many statistics for each unit such as average, maximum, minimum, standard deviation, etc. of the morphometry variables slope angle, slope shape, internal relief and drainage density. The whole database is described in the section 8.2. All the data and the maps are part of a geographic information system application able to update, edit, analysis and manager the information with emphasis in disaster management applicability. The complexes are a geomorphological regionalisation considering the genesis, lithology and geography of the different landforms in the study area. Taking into account these factors and after photointerpretation and image analysis, the area was subdivided in the following Complexes: - Coastal hills - Accumualtional slopes - Limestone hills - Metamorphic hills - Terrigenous hills - Caujeri depression


33


Figure 4-2. Geomorphological complexes in the study area. The distribution of the complexes is shown in Figure 4-2. This chapter will describe the characteristics of each complex as a description of the geomorphology of the study area. 4.3.1

Coastal Hills

The coastal hills in the study area are isolated hills parallel to the coastline (Figure 4-3). The length is variable, depending of the mouths of the rivers, as the coastal hills are cut by the drainage system. The width is also variable and is between one or two kilometres. In the study area there are three coastal hills: between El Naranjo and Baitiquirí bay, between the Baitiqurí bay and Sabanalamar bay (Loma Los Aposentos) and between Sabanalamar bay and Macambo 34


GEOMORPHOLOGY

town. As was explained in the section 3.10, these three hills are located paralel to the coastline and separated by three different lineaments. Consequently, Loma los Aposentos in the northernmost, El Naranjo-Baitiguiri bay is in the middle and Sabanalamar Bay-Macambo is the Southeast.

Figure 4-3. Anaglyph images and field photographs of different complexes. From top to bottom: Coastal Hills, Limestone Hills and Terrigenous Hills. One distinguished feature of the coastal hills is their top, which is almost horizontal and covered by a more resistant layer, probable Santa Cruz formation. It is possible to see local karst forms due to dissolution in the top layer of the coastal hills. The hills have different altitude and profile (Figure 4-4). The top of El Naranjo-Baitiquirí hill is around 190 meters; of Loma los Aposentos between 220 to 280 and of Sabanalamar bay-Macambo hill is between 100 to 130 meters. Also the isolated hill Pan de INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

35


Azucar has similar altitude, of 220 meters. Figure 4-4 shows different Coastal Hills profiles for the three main coastal hills in the area. 300 250 200 150 Baitiquirí 100

Los Aposentos Macambo

50 0 0

200

400

600

800

1000 1200 1400 1600 1800 2000

-50 -100

Figure 4-4. Different coastal hills profiles. Geologically, the Coastal hills are composed of four main geological formations, which play an important role in the current form of the hills and in the process going on. The lithology is different in the southern (coastal) side and in the northern side. The north side is totally covered by the Maquey formation, which are mainly terrigenous rocks and are susceptible to landslides. The coastal slope, characterised by marine terraces is composed of the Maya formation, except for the lowest terraces, which are composed of recent (Holocene) marine deposits. These recent marine deposits act as “rings” of the coastal hills and are uplifted between 5 and 10 meters from the current sea level. They are useful to recognise if coastal landslides were pre-Holocene or Holocene, depending of the conservation of these deposits. The vegetation is rare, although there are some endemic species typical from semiarid zones. The soil is also scarce due the high erosion, slope steepness and lack of humidity. The drainage system is limited to few channels because the proximity to the sea and the relative small area of the coastal hills. 4.3.2

Accumulational Slopes

The Accumulational Slopes are located between the Coastal Hills and the Limestone Hills (Figure 4-5), geographically between San Antonio del Sur - Baitiquirí and Sierra de Mariana. In addition, there is another area to the east of the Sabanalamar River and south of Sierra del Covento. In fact, it is an intra-mountainous fluvio-marine plain with deltaic origin. This complex is lightly sloping to the south (sea) side with angles between 5 and 15 degrees as average. The area is elongated and curved with a width of around 2 kilometres. The Accumulational Slopes are composed of recent (Quaternary) deposits transitioning from colluvial, in area close to the mountains, to Alluvial in the southern part of this complex. The materials are 36


GEOMORPHOLOGY

gravels, yellow and reddish sandy clays, which are generally compacted. The lithology also varies in the Southwest part and in the east part of this complex where the main rocks are from formations Cilindro, Maquey and San Luis.

Figure 4-5. Anaglyph images and field photographs of different complexes. From top to bottom: Accumulational Slopes, Metamorphic Hills and Caujeri Depresion. This complex seem to belongs to an old planation surface, which collected all the sediments coming from the upward area, what is now Sierra de Mariana, during the Pleistocene. The extension and volume of the quaternary sediments reveals that the rainfall at that time was larger than currently. This might be true considering the fact that in the northern border of Sierra de Mariana the drainage seems to be cut-off due to large mass movements. In both the western and eastern sides of the area, the PleisINTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

37


tocene sediments are not present. In the western side the drainage system was sufficiently strong to erode the sediments to the Baitiquirí bay, besides this part appears to be slightly more uplifted. In the eastern part the pre-Quaternary formation overly the Ophiolites and the area has an irregular relief. The planation surface is raised 10 to 30 meters above the current erosion levels, creating around ten new channels, which eroded the old plain and generated erosional scarps with the same high differences. These channels control the poor drainage in the area. Due the proportion of the terrigenous material, the water is infiltrated and the channels are being lost in the lower parts. The vegetation is sparse with isolated shrubs being more common in the Southwest part. The land is mainly fallow, although in the Southwest there are some cattle farms. 4.3.3

Limestone Hills

The Limestone Hills include Sierra de Caujeri and Sierra de Mariana (Figure 4-3). This complex is located in the entire western side of the study area, although Sierra de Mariana also covers until the central part. The Limestone Hills is a monoclinal plateau with an average altitude of 500 meters. It is composed mainly of limestone of Yateras formation. Karst dissolution is present and generally in relation with tectonic processes. Then, there are karst depressions and isolated karstic hills. Wherever the upper limestone layers were eroded, coinciding with the drainage system, the underlain layers belonging to the Maquey formation are outcropping. Therefore, the erosional process is increasing due to the low competence of Maquey formation rocks. The tectonic structure plays an important role in this complex since the area in the south and east is limited by large fault scarps. The south scarp has around 100 meters high difference and gives the impression of a normal fault as a consequence of extensional stress. The east scarp is the boundary of Sierra de Caujeri and also seems to be a normal fault. Both scarps are where most of the landslides are located in the study area. It is remarkable also the subsidence of a large rectangular block in the area of La Tinaja, at the west of El Mije. This block was subsiding and tilted to the south by two main faults, which even can be traced into the Caujeri valley. The interruption of Sierra de Caujeri scarp by these two faults is an indication of the dependence of one fault system to the other one. Due to the karstic and tectonic processes in the limestone hills, the fluvial system has a combined pattern where rivers follow fault lines and karst-dissolved zones. Moreover, there is a significant the number of rivers and streams, which start close to the Sierra de Caujeri scarp to the west at such altitude as 700 meters. The Limestone hills are mainly covered by forest, although some parts are burned. The area is almost unpopulated and, therefore the terrain is underexploited. 4.3.4

Terrigenous Hills

The Terrigenous Hills are located in the central and north-central part of the area (Figure 4-3). They are composed of terrigenous and volcanogenic rocks belonging to the Cilindro, Charco Redondo, San Ignacio, San Luis and El Cobre formations. Due to the large number of rock types, the landforms are also diverse, but in general they differ from moderate to severely dissected hills. They can be classi38


GEOMORPHOLOGY

fied as folded and dissected mountains (500-600 m) and hills (300-500). The area shows diverse altitudes between 200 and 500 meters with extreme exceptions. Structurally, the Terrigenous hills are the limit between the Foldbelt and the Neoautochthon in the study area in an approximately north-south direction. The drainage system is well developed and also follows this north-south direction, collecting the water and sediments from the metamorphic rocks in the east, and from the terrigenous hills itself in the west. The water is collected to the main drainage, which is the Sabanalamar River and is stored in the Plamarito Dam. After the dam, the river also receives all the water coming from the west belongs to Caujeri valley. The vegetation is abundant and varied. Mainly coffee plantation and secondary individual farmers use the area. The rill and gully erosion is present and there are isolated landslides 4.3.5

Metamorphic Hills

The Metamorphic hills are present in the east and north-east part of the area (Figure 4-5). They are composed of the Sierra del Purial formation and the Ophiolites. Although the lithology is different in both geological units, from a geomorphological point of view, there are not many differences between these two units in the study area. This is because, the competence of the rocks and the tectonic processes going on are more or less the same. The volcanogenic and metasedimentary rocks of Sierra del Purial formation and the Ophiolites rocks are both included in the Metamorphic hills complex. This complex also presents an isolated rounded outcrop in the north part of the Caujeri valley, east of Mameyal and Letreros. This outcrop, belonging to the Sierra del Purial formation, seem to be a "window" of the metamorphic rocks where the young material were complete eroded. The Metamorphic hills are the most dissected, from moderate to severely dissected, rocks in the study area. The area also has the highest parts reaching up to 1,060 meters high. They constitute folded blocks in relation with intensive neotectonic uplift. The new displacements are clearly in the design of the fluvial network, which are shaped in the combination of the neotectonic and metamorphic rock type. The area is mainly forest with abundant vegetation and there are some parts, which are used for coffee plantations. 4.3.6

Alluvial Valleys

The Alluvial valley complex is related to the recent sediments accumulated close to the principal river systems (Figure 4-6). Sabanalamar river floodplain, Macambo river floodplain and the most recent fluvial channels in the Caujeri valley are part of this complex. They are composed of alluvial and swampy deposits. The alluvial deposits are mainly grey and greyish mud, sandy mud and sandy clay. The swampy deposits are accumulations of mangrove rest, silting up fluvial lacustrine accumulation rich in carbonates, clays and peat.


39


Figure 4-6. Anaglyph image and field photograph of Alluvial Valley complex. This complex is essentially a fluvial plain with a combination of erosive and accumulation processes. Accumulation is prevailing in the Sabanalamar and Macambo floodplains and erosion in the Caujeri valley. Three floodplain levels are recognisable: active, occasionally submerged and exceptionally submerged floodplains. The area is regularly inundated during intensive rain. It is not totally recognised the fluvial terraces levels. Close to Sabanalamar bay and in the surrounding of San Antonio del Sur town, there are brackish water swamp lagoon and swamp deposits. The mangrove vegetation is abundant only in Sabanalamar river mouth where the brackish water is accumulated. The channels in Caujeri valley are the widest in the study area, between 2-5 meters; and in some places even 10 meters. They are dry most of the time and are used for local transportation. The bottom of the channels is totally covered by pebbles with a regular diameter of 10 centimeters and smaller. The channels in Sabanalamar river are more narrow and permanently cover by water. The pebbles are around the same size being smaller in the mouth of the river. 4.3.7

Caujeri Depression

Caujeri depression complex was called to Caujeri valley and the colluvial deposits of Sierra de Caujeri scarp including the main scarp (Figure 4-5). The area is a sequence of Scarp-colluvial-alluvial deposits, where in some parts it is difficult to distinguish the boundary between one type of deposits and the other. In the geological maps, San Luis formation has been mapped between the colluvial deposits of the scarp and the alluvial deposits of the valley. However, during the fieldwork this formation was not recognisable and all de material in this area seems to be terrigenous material originated by the combination of landslides and the weathering process. The valley can be considered as erosive fluvial plain, relatively high (between 200 and 300 meters above sea level), presenting frequent small hills. The origin of the valley seems to be tectonicstructural (see next section) and the main process going on is fluvial erosion. The scarp and colluvial deposits are a consequence of several multiple and successive landslides. The slope' s shape is very

40


GEOMORPHOLOGY

variable due to overlapping of the mass movement processes. Therefore, many steps, including concave and convex forms, can be found in the slope profile, but three main steps can be recognised. The lower or front part of the slope was delimited by mapping the end of the landslide transport zone. But, in many landslides, this boundary is not exactly recognisable, especially in the mud flow or debris flows where the material slides down until it is mixed with the alluvial deposits. The drainage is actually starting in the main scarp from where many springs originate. There are many small streams in the main scarp or in the colluvial deposits down hill, which are not in the topographic map and they are used for minor agricultural use. As soon as the drainage enter into the valley, only the main streams remains since the small ones infiltrate into the soil. This means the existence of high infiltration capacity in the valley. In some areas, the erosion creates large gullies in the valley and it is possible to see a lower layer with large unsorted materials belongs to previous mass wasting processes.

4.4

Towards the landform evolution modeling in the study area

The landform evolution in the study area can be explained using in two main models. One to explain the Caujeri valley and one to explain the Coastal hills and the pouch-shape bays. The reason of these two models is because in certain time both areas were developing independent landform due to different causes. The Caujeri valley is an inland depression of the graben-type with high elevation differences up to 500 meters. The valley is limited in the west by a large scarp of Sierra de Caujeri, with constantly active west-retrogressive movements due to landslides. On the other hand, the valley is also surrounding by major fault scarps, which argued the depression form. Therefore, the origin of Caujeri valley can be interpreted as a combination of tectonic and mass wasting processes. The main fault systems (Mariana and Caujeri) started to generate a graben depression after the Second Transgression-Depression (Lower Miocene to Late Miocene). Later, a chain of landslides were moving back the Sierra de Caujeri scarp with an average of two kilometres, some times even three kilometres. Regarding the Coastal hills, more literature is available. Talking about the marine terraces as early as 1945 the Dutch geologist F. G. Keijzer interpreted them "as a sign of intermittent uplift of the land". Thereafter, the last paper published about the marine terraces (Peñalver et al., 1998) characterised the lithology and other characteristics of the different terraces levels.


41


One of the issues is the pouch-shape bays. They are distributed not only in the study area (Baitiquirí bay) but also in the entire coast along the eastern corner of Cuba. The better interpretation for this phenomenon is the eustatic change of sea level. Another possibility is the tectonic downward movement, which is not very possible since marine terraces show exactly the opposite: upward tectonic movement. However, another aspect to find an answer for the isolation and linear shape of the Coastal hills. It may be possible (Keijzer, 1945) by the tectonic movement either slightly folding or faulting, by unconformable cover of "protective" rocks and by both. In the study area there are no signs of anticline of syncline forms, but there are tectonic lineaments (see 3.10 General tectonic and 4.3.1 Coastal Hills), which limited the coastal hills in the north side. Unfortunately these lineaments are not so clear in the field due the competence of the Maquey formations rocks. Finally, three aspects are important to explain and they seem to be genetically close related: the pouchshape bays, the marine terraces and the isolated and linear shape of the coastal hills. The recent evolution relief can be explained as an intermittent uplift, together with erosional process in the north side of the coastal hills, during the Pleistocene (Figure 4-7A).

Figure 4-7. Coastal Hills and pouch-shape bay evolution hypothesis (taken from Keijzer, 1945).

42


GEOMORPHOLOGY

The tectonic setting and the upper layer of the coastal hills generated its the isolation and linear shape. Later, in the last glaciation the erosive levels were much lower than the actual sea level and the mouth of the river were more dissected than as are now (Figure 4-7B). When the sea level goes up, these areas were occupied (Figure 4-7C) and finally differential movements now lifted the land out of the sea again generating the pouch-shape bays and the marine terraces (Figure 4-7D). Wherever the recent sediments were enough, they filling up the bays as probable happened in Sabanalamar bay with the large amount of sediments coming from the Caujerí valley.

4.5

Summary

The relief of the study area was created by the combination of horizontal and vertical movements. The horizontal ones predominated up to Middle Eocene and later mainly vertical movements took place. The landforms in the area are a result of climatic, tectonic and lithological factors, which, due to their variability, present different geomorphological regionalization. The area can be subdivided into six complexes: coastal hills, accumulational slopes, limestone hills, metamorphic hills, terrigenous hills, and mass movements complex. The coastal hills are isolated, monoclinal, karstic, tectonic-structural hills with marine terraces parallel to the coastline. The limestone hills are a monoclinal plateau controlled by the tectonic and the lithology with karst dissolution. Terrigenous hills are folded, dissected small mountainous and premountainous. The Metamorphic hills are folded blocks of metamorphic rocks with intensive neotectonic uplift. The alluvial valleys are erosive and accumulative fluvial plains with alluvial and swampy deposits. The Mass movement complex are the colluvial deposits of Caujeri scarp and the erosive fluvial plain of Caujeri valley with tectonic-structural origin. The Caujeri valley was originated by a combination of tectonic and mass wasting processes. The coastal zone was affected by intermittent uplift with intensive erosion controlled by the tectonic lineaments. Later, eustatic movements during the last glaciation moved down the erosive levels and when the sea level rose again, the pouch-shape bays were created.


43

DATA PREPARATION AND PROCESSING

5 Data preparation and processing 5.1

Introduction

The data preparation and processing stage took a long time and a lot of effort. Since there was hardly any existing data nearly all data had to be surveyed, created or processed from the beginning. The general idea was to produce a number of products that can be used later during the interpretation. The digital elevation model (DEM) is one of the most important products for landslide hazard assessment. With the DEM many relief-derivative maps can be produced including slope, internal relief, aspect, shadows, etc. It is possible to combine many of these derivative maps and the DEM itself with satellite image and vector information. The DEM is also useful for threedimensional representation of any other data (Satellite Images, vector maps, etc.) which is very useful in the interpretation and in the understanding of the study area. Various image data and aerial photographs were used during the data preparation and processing. Landsat TM, SPOT PAN and JERS-1 SAR were the satellite products to be processed. Their georeference, radiometric, spatial correction, and enhancement are explained in this chapter. On the other hand, for the research two-scale aerial photographs were used with a total of 101 aerial photographs. Additionally, other information such as a geological map, infrastructure, drainage information, etc. was surveyed and included in the analysis. The input of these data and the preprocessing will be described in this chapter too.

5.2

The digital elevation model

The digital elevation model (DEM) was the most time consuming part on the data preparation; it took one full month to complete. The procedure was done scanning the contour lines and digitizing them using a "raster tracing line" system called SAMI (Semi Automatic Map Input). The standard procedure followed different steps and, in addition to that, some technical problems provoked a longer procedure and with less accuracy. The general flow chart is shows in Figure 5-1. The contour lines were drawn with a 2-millimeter ink-pen over an acetate overlay from the topographic map (1:50,000). The next step was the scanning of the overlay, but the scanner was not large enough for scanning the whole area at one time (in one file). As a consequence, the area was scanned in three pieces (three files) that later were georeferenced and glued in ILWIS. The pieces were glued in ILWIS after georeference by control points located in the overlay. It is important to point out that normally it is possible to have access to original contour layer in acetate of the topographic map and INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

44


also to scanning in only one piece. Then, two errors were introduced: the drawing of the contours and the scanning in pieces.

Drawing contour lines

Scanning overlay Glue area’s parts in ILWIS Import contour into CARIS Thin raster data in CARIS Vectorize contour lines in SAMI Create control files Register contour map in CARIS Cleaning contour map in CARIS Smoothing the contour lines in CARIS Create DEM in ILWIS

Figure 5-1. Flow chart for DEM generation. For scanning an A3-size DUOSCAN T2000XL version 1.5 was used, produce by AGFA. Several options were tested in order to get the optimum contour separation and line width. The line width was tested changing the number of dots per inch (dpi) in 100, 300, and 500 dpi. The color separation was tested changing the gray threshold in 10, 25 and 50 percentage. Finally, the optimum combination was 300 dpi with 25 percentage of threshold. Once the image was in one file (TIFF format), it was imported into CARIS GIS. The next procedure was to get the lines as thin as possible into CARIS. The algorithm used reduces from the borders to the center the number of pixels for each line until pixel line is one pixel wide. After that, the SAMI package was used to vectorize the raster contour lines. The procedure is simple and fast, depending on how clear the image is. Clipping on a pixel and the program starts to trace the line until the finish or till it has found "a decision point" to continue tracing. Once it finishes to trace one line, the program asks for the contour line value and the vector line is stored. Figure 5-2 shows a small fragment of raster lines converted in vector lines by the system.

When all the lines were digitized, the next step was to georeference the segment contour map to the actual coordinate system. To do that, it was necessary to create two control point files, one with the INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

45


map coordinates and one with the real coordinates. Because the image was already scanned with 20 control points, the map coordinates of these points were recorded. However, when the contours map was georeferenced and overlain with a Spot PAN image, large positional differences were noted between the image and the contours map. It was assumed the number of aggregational errors introduced from the beginning deformed the current contour map.

Figure 5-2. Fragment of vectorization raster lines with SAMI. The black pixels are contour lines in raster and the inner grey lines are new vectors traced by the system. To solve this problem more control points were located in those areas where there were not many points before. A small Microsoft Excel application was developed to calculate the sigma error interactively and different interpolation methods were tested. The original 20 points presented a SIGMA error of 5.7 pixels. Additional 23 points (43 in total) were located and the SIGMA was reduced to 3.2. After that some control points were eliminated interactively to continue reducing the SIGMA error, with 36 points the error went down to 2.1, and finally, removing two more points the SIGMA was 1.51. There was no more possibility to remove points and consequently reduce the SIGMA error. Figure 5-3 shows as an example the difference in georeference with the Spline method using 43, 36 and 34 points in the worst part of the contour map. The maximum displacements between the contour lines are around 50 meters. Even when the SIGMA was reduced to 1.5, meaning 75 meters at 1:50,000, the contour map still looks deformed in some areas after the georeference. In order to correct this deformation, different interpolation methods were tested. First at all, 15 extra control points were taken with the actual coordinates from the topographic map. These 15 points were taken to test the interpolation differences according with the interpolation methods.

46



Figure 5-3. Georeference differences using the Spline method with 43, 36 and 34 points. The methods include Projective, Orthographic, Triangle, Polynomial (up to 5th order), Similarity and Spline (CARIS, 1998). The best result obtained was with the Spline method getting a SIGMA error of 0.52 for these 15 points. The reason because the Spline method gave better results can be because the Spline method uses a surface fitting spline to register data and made a “rubber sheeting” transformations by holding the ground control points fixed. This way avoids the possibility that the errors in one area take effect in other areas far away (see Figure 5-3). After the contours map was georeferenced the next step was to “clean” the contours as much as possible. The cleaning step was subdivided in three tasks: 1. Check junctions or lines elevations 2. Check duplicate lines 3. Check contour consistency The first two tasks were done automatically in CARIS since the system has specific tools for that. In Check Junctions tools the program checks if there is some height differences in the points of the same contour line. For checking the duplicated lines the data need to have the network topology, get reported an error for duplicated lines and applies the tool Remove Duplicate data. Check contour consistency means to verify if the Z-values of contour lines are in the corresponding sequence. There is not a specific tool for that task in CARIS. A solution was to make a temporal DEM with its shadow images and check visually in the shadow images if there were anomalous shadow pixels corresponding to contour inconsistency.


47


After the cleaning of the contour map, the next step was smoothing the contour lines. A CARIS tool was used for this purpose testing different smoothing order. The smoothing order 2 was the most suitable since the contour lines are smooth but the line shape is not lost. Figure 5-4 shows the original contour line and the smoothing using different smoothing order.

Figure 5-4. Original contour lines and different smoothing order (2, 5, 7). Enlargement 125x. Finally, the contour map was exported in DXF format to be imported in ILWIS 2.2 to create the Digital Elevation Model (DEM). Because the ILWIS 2.2 has problem to import lines in DXF format with Z-values, the DXF file was first imported in ArcView 3.2 and then exported as a shape file. The shape file with the contour lines was later imported in ILWIS 2.2 and the DEM was created successfully. For further research it would be necessary to compare this method with manual digitising.

5.3

Image processing

For carrying out this research different types of remote sensing data were used for various purposes. As was mentioned in section 5.1, the Landsat TM, SPOT PAN and JERS-1 SAR data were used in the research. A satellite metadata summary is presented in Table 5–1, all imagery data cover the entire area. The idea was to combine different sensors type with the purpose of comparing their applicability for applied geomorphology and landslide mapping. Satellite Landsat SPOT JERS-1

Sensor TM PAN SAR

Date 01/15/1985 28/12/1994 01/05/1994

Time 14:51:03 15:40:45 15:27:39

Spectral resolution Multispectral 7 bands Panchromatic Radar HH

Spatial resolution 30 x 30 meters 10 x 10 meters 12.5 x 12.5 meters

Table 5–1. Satellite metadata summary Then, the multispectral, panchromatic and radar data was collected and processed. The next three subsections will explain the processing for each data type specifically. 48



5.3.1

Landsat TM

The Landsat Thematic Mapper (TM) data was georeferenced with 7 ground control points and the SIGMA error was 0.668 meaning 20.04 meters for the 30 meters image resolution. TM Band 1 2 3 4 5 6 7

Min 0 0 0 0 0 0 0

Max 200 129 185 158 255 166 241

Mean 73.36 29.02 29.48 57.00 70.56 133.87 26.98

Median 70 27 25 60 68 132 23

Mode 71 19 21 5 3 132 0

Std. Dev 12.00 8.55 14.84 23.10 39.46 8.01 19.76

Table 5–2. Statistics for Landsat TM bands. One positive factor in the processing of the TM bands was the total absence of clouds. The purpose of the TM bands processing was to use for visual interpretation. Then, the bands where first visually characterised without streching. The statistics show (Table 5–2) that most of the bands have low digital numbers and consequently they look in general with dark tones. Figure 5-5 shows ranges of digital numbers for different features of the study area.

0

100 53

TM1 13 22

TM2 5

TM3 TM4 TM5 TM7

16

0

10

0

8

58 30

80

255

200

129

60

185 158

62

0 4 16

Sea

65

200

120

255

60

Forest

Valley

241

Recent material

undifferentiated

Figure 5-5. Spectral ranges (in digital numbers) of different features in Landsat TM bands.

Analysing the histograms of the six bands (excluding band 6) it is possible to recognise and spectrally separate only a few natural features including: INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

49


-

The sea. The forested area. The Caujeri valley. The non-vegetated area north of San Antonio del Sur.

The sea contain the lowest digital numbers in all TM-bands except in TM-Band 1, where the forest contain the lowest digital numbers. The minimum value for all the images is different and never start from 0, except for the band 7, where few pixel contain zero values. After the sea, the next feature with lower digital numbers is the forest. Its values vary due to the vegetation differences but in general, they are dark pixels. In some parts there are few pixels with slightly lighter values due to non-forested areas or intensive denudational processes like landslides.

Figure 5-6. Three-dimensional view with Landsat TM color composite 457 (RGB). The Caujeri valley appears in general with higher digital numbers than the forest, except in the TMband 4 where its digital numbers can not be differentiated from the forest. The Caujeri valley is mainly agricultural area, then, the digital numbers present more differences due to the fact that some areas are cultivated and other ones not. An area containing contrasting digital numbers are the accumulational slopes in the northern part of San Antonio de Sur up to Baitiquirí. Due the lack of vegetation, the type of material this area have very high digital numbers, usually the higher digital number in the images. For enhancement the Landsat TM images the general procedure was to made the atmospheric correction bringing all the bands to the lowest value 0 and then, to execute different stretching methods. The better stretching method was linear stretching using the standard deviation statistics. This method will be explained in the processing of the SPOT PAN image. 50



Finally, with the Landsat TM image was tested several color composite in order to get better information from the image. The better color composites were 457 and 321 (Red, Green and Blue). With these color composites several three-dimensional views were created and the main landforms in the study area were analysed visually. Figure 5-6 shows a three-dimensional view using the 457 (RGB) color composite. The central part of the figure is the Caujerí valley and in the lower part are located the Accumulational slopes. The upper left part in red is the metamorphic hills. 5.3.2

SPOT PAN

The SPOT panchromatic (PAN) image was more useful for the thesis objectives due to the higher spatial resolution (10 meters), although the image has about 2 percentage clouds cover in the study area. As the statistics shows (Table 5–3) the range of digital numbers in the SPOT PAN are from 5 (the sea) to 255 (the clouds). Actually the "sea" digital numbers start in 20, but from 5 to 20 there are only few pixels. On the other hand, starting in 100 the feature "clouds" take all the digital numbers until 255. Min

Max 5

255

Mean 44.179

Median

Mode 43

24

Std. Dev 15.54

Table 5–3. Statistics for SPOT Panchromatic image. Taking into account these problems a standard deviation stretching was apply following some steps: 1. The few pixels below 20 were moving up to 20, remaining all other pixel in the images as the same. 2. All the pixels in the images were moved up to 0, subtracting 20 digital number. 3. The standard deviation stretch method was applied to enhance the image.

Figure 5-7. Histogram of the original SPOT PAN image. The Figure 5-7 shows the histogram of the original image over which the steps were executed. The standard deviation stretch method is simply linear or equalised stretch but using minimum and maximum values instead of percentage. The minimum should be the mean minus two times the standard deviation and the maximum should be the mean plus two times the standard deviation. In the SPOT INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

51


image case, after calculate the new statistics the minimum value was 13.43 and the maximum value was 74.87. With the new stretching the image looks much better. In addition to that, different stretching methods were testing and visually compared. As a conclusion the standard deviation stretched image was accepted. After stretching the SPOT PAN image, different high pass filters were also visually tested to improve the contrast in the image. Two Laplace plus filters were designed (Figure 5-8) in ILWIS. 0 -1 0

-1 5 -1

0 -1 0

0 -1 0

-1 7 -1

0 -1 0

Figure 5-8. Two Laplace plus filters designed in ILWIS The filters were applied to several stretched images and to the original image. In general many filters produce a loss of details and some filters create black and white areas. Due the needs of good quality images for data fusion, the standard deviation stretched image was used. 5.3.3

JERS-1 SAR

The radar image used was from the Japanese satellite JERS-1, sensor SAR. The format of data is 16 bit per pixel with polarization HH, wavelength 23.5 (L-Band) at 35 degrees as incidence angle. This 16 bit data is a combination of the cosine component (I or in-phase component) and the sine component (Q or quadrature component) of backscattered radar return signal. For pre-processing and enhancement the radar data ERDAS Imagine 8.3.1 image processing system was used, later a georeference was created in ILWIS 2.2. Once the data was into the ERDAS system two mains tools were used in the RADAR module (ERDAS Imagine 8.3.1, 1998): 1. Speckle suppression 2. Image enhancement As is well known most of radar images have a problem called speckling due to signal out of phase producing interaction between radar waves. The speckle noise generates light and dark pixels making the interpretation difficult. For speckle suppression different filters were used changing some parameters of the algorithms. Table 5–4 shows the different parameters used for speckle suppression. In general the speckle was never suppressed completely all and in some cases the result produced worse images to be interpreted. The best results visually evaluated were the images sar3, sar6 and sar7 (Table 5–4). With these results the image enhancement procedures were carried out. Speckle Suppression tool, RADAR module, ERDAS 8.3.1, 1998 Input file Output file Coef. of Coef. of var. Window Filter variation multiplier size sar sar1 0.280340 0.5 3x3 Lee-sigma sar1 sar2 0.219507 1 5x5 Lee-sigma sar2 0.214156 2 7x7 Lee-sigma sar3

52



sar sar4 sar5 sar sar

sar4 sar5 sar6 sar7 sar19

0.26 0.26 NA 0.280340 0.280340

1 1 NA NA NA

3x3 5x5 5x5 3x3 3x3

Lee-sigma Lee-sigma Local region Frost Gamma map

Table 5–4. Processing table of speckle suppression. Best results in bold case. For image enhancement the Wallis Adaptive Filter and Luminance Modification (ERDAS Imagine 8.3.1, 1998) tools were used. The Wallis Adaptive filter was used in the bandwise filtering option, because there was only one band. Wallis Adaptive Filter, Image Enhancement tool, RADAR Module, ERDAS 8.3.1, 1998 Input file Output file Window size Multiplier sar sar8 3x3 2 sar3 sar9 3x3 2 sar6 3x3 2 sar10 sar7 sar11 3x3 2 sar6 9x9 2 sar12

Table 5–5. Processing table of Wallis Adaptive filter. Best result in bold. Table 5–5 show the different processing carried out. The best results (sar10 and sar12) were obtained with the image "sar6" either with 3x3 or 9x9 window size. Additionally to this processing for image enhancement the Luminance Modification tool was used to try to improve the radar image. The Luminance Modification tool is "an adaptive enhancement filter which separates the original image in two parts- the scene luminance and the scene contrast. These two parts are modified and recombined to created the enhanced output" (ERDAS Imagine 8.3.1, 1998). The tool was executed over different images changing the parameters. Luminance Modification, Image Enhancement tool, RADAR module, ERDAS 8.3.1, 1998 Input file Output file Objective Multiplier Window size Local Luminance intercept sar3 sar13 Undegraded 2 5x5 100 sar6 Undegraded 2 5x5 100 sar14 sar14 sar15 Undegraded 2 3x3 100 sar13 Undegraded 2 3x3 100 sar16 sar6 sar17 Undegraded 2 5x5 200 sar19 sar20 Undegraded 2 3x3 100

Table 5–6. Processing table of Luminance Modification. Best result in bold. Table 5–6 shows the results of processing the former processed images with the Luminance Modification enhancement filter. In some cases, as the procedure recommends, some images were passed two times with the same filter in order to get better results.


53


After finishing the enhancement processing, the selected images were exported from ERDAS to ILWIS in order to georeference. The georeferencing in ILWIS was rather difficult due to the impossibility to find out control points. A georeference tie points with 10 points only was able to bring the SIGMA to 4.367 pixels, meaning 54.5875 meters. For that reason the image was not used for combination with other data. Finally all SAR images were compared to evaluate visually the different landforms. Figure 5-9 shows the seven images selected from the processing. The images correspond to the coastal hills located Southwest of Baitiquirí. The area covers about 2 km by 2.5 km and it is oriented with east upward as the arrow shows. The resolution is the original 12.5 meters except for two small windows (lower-right corner) where the zoom is reduced by 2 and by 4, meaning 25 and 50 meters respectively. The names of the images correspond to the names in the processing tables 5-4, 5-5 and 56.

Figure 5-9. Different Sar images processed of a Coastal Hills. Files according to the tables from top to bottom, left to right: sar (original), sar3, sar6, sar7, sar10, sar12, sar14, sar16, sar zoom out by 2, sar zoom out by 4. See text for explanation. As the Figure 5-9 shows seven the selected images are not clear enough to make a geomorphological landform interpretation. Moreover, the scale is not appropriated for landslide interpretation since in the image it is difficult to recognise the marine terrace levels. However, when image is zooming out 54



by 2 o by 4 times, the speckle noise start to disappear and the different landform at more regional scale start to become clear. As a conclusion of the radar SAR satellite image processing this data is not totally suitable for geomorphological landform interpretation at detailed and medium scale (1:50 000 and 1:25 000) due to speckling. The radar SAR sattellite images can by used for regional interpretation especially when no other data is available. There is a potential capability of airborne radar images, but these data still remain very expensive, especially for developing countries.

5.4

Some data fusion

After the main data from original satellite was prepared the next step in the processing was the combination or "fusion" some of these data. In this research four main data fusion were develop: 1. DEM - SPOT image, bidimensional 2. DEM - SPOT image, tridimensional, Anaglyph image 3. Geology - SPOT image 4. Landsat - SPOT image The DEM-SPOT image was combined using the RGB-IHS transformation. The Figure 5-10 shows the general flowchart for data fusion. The procedure start with the SPOT PAN image, the digital elevation model DEM and a "dummy" channel with only one digital number for the whole image. With the digital elevation model is create an image in which each pixel correspond to RGB color (0-255,0- 255,0255) of the color table or color representation according with the altitude (Z-value in the DEM). Then, this image was convert to RGB color representation system creating three files: red, green and blue components. With these three files the IHS (Intensity, Hue and Saturation) color representation system is created generating three new files. From this result only the Hue image is taken.

SPOT PAN

Input

DEM

Dummy Channel

Map Color Table

Use as substitute of intensity

Convert To RGB

Covert To IHS

Use as substitute of saturation

Intensity

Hue

Convert To RGB

Output

Saturation

Figure 5-10. Data fusion transformation flowchart for DEM-SPOT PAN fusion. The reverse conversion, IHS to RGB, is executed but now the Intensity image is the SPOT PAN image and as Saturation image is used an dummy channel image with only one digital number. The digital number to be selected depends on the previous saturation of the SPOT image and it is an interactive process changing the digital number until the desirable results are reached. In this case a suitable digital number was 230.


55


The result of this fusion is represented in Figure 5-11. The study area can now be interpreted through the SPOT image also taking into account the relief differences. The red and brown colors represent the highest parts whereas the green and grey-blue areas are the lowest part. The window on the right shows the Jagueyes landslide where the white pixels correspond to the scarp and non-vegetated areas. To produce the Anaglyph image (Figure 4-1) the original digital elevation model (DEM) and the improved SPOT PAN image were used in StereoPair tool of ILWIS 1.4. The procedure made a linear stretching of the SPOT image between two defined values, define the height reference at which points having this altitude will appear to be on the "screen level" and the angle of the stereo view meaning the shift between both images in the stereo pair. With these parameters the Anaglyph image was generated and then imported in the ILWIS 2.2.

Figure 5-11. SPOT PAN image and DEM fusion. Left, whole study area. Right, window at full resolution (10 meters per pixel). The Anaglyph image was very useful for the initial interpretation, the fieldwork and for creating the finals maps, using digitising on screen. It is important to note the digitising on screen over the Anaglyph introduces an error because the boundaries to be traced may be "behind" or "ahead" the actual computer screen. However the tool is very useful to recognise height differences and trace boundaries that may be spatially corrected with another background image like the original SPOT PAN image. The Figure 4-1 shows the Anaglyph image for the whole study area and the figures 4-3, 4-4, 4-5 show windows of different areas at more detail scale. Another data fusion develop was geology raster map and SPOT image with the purpose of enhancement the visual interpretation and to compare the geology with the different landforms in the study area. Similar to this data fusion was the fusion between Landsat TM bands and SPOT PAN image. In 56



this case the purpose was to improve the spatial resolution of the Landsat TM bands for the visual interpretation. Both data fusion follows similar procedure as DEM-SPOT PAN fusion (Figure 5-10). In one case the DEM was changed by the Geology raster color map and in the other by TM bands color composite. The results were also very useful in the interpretation and in the general understanding of the different landforms in the study area.

5.5

The photointerpretation

The phointerpretation in the study area was done using aerial photographs at 1:25,000 (22 photos) and 1:37,000 (79 photos) scale approximately. The photos were in 18 x 18 centimetres format with relative good quality. Figure 5-12 shows the centers of the aerial photographs with their number. The black circles are at 1:37,000 and the black boxes are at 1:25,000 scale. As the Figure 5-12 shows some areas were not completely covered by the photos. In these cases the SPOT PAN image was used.

Figure 5-12. Photocenters of aerial photographs in the study area Initially, a general photointerpretation of the central part of the area was done during the Terrain Mapping Classification ITC course together with the colleagues Juan Carlos Rojas Calabria y Ruben Vargas Franco. During the thesis period this interpretation was rechecked, the landslide areas were interpreted in more detail and the areas not covered before were completed now. Together with the second INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

57


photointerpretation two Photo-checklists were filled: one for the Terrain Mapping Unit (TMU) and one additional for landslides. These two photo-checklists correspond somehow with the database explained in Chapter 8 and the characteristics of the interpretation are explained in Chapter 4 and Chapter 6. During the field period the areas photointerpreted were rechecked in order to determine the actual characteristics of the different areas. After the field the boundaries were transfered from the photos to the computer with on screen digitising using as a background different image products including the SPOT PAN and the Anaglyph image.

5.6

Other input data

Other types of information useful for the thesis purposes were surveyed during this research. Data such as drainage, springs, houses and roads were digitised from the topographic map at 1:50,000 scale. All data was digitised in ILWIS with a CALCOMP 3400 A0 Format digitizing table. The drainage was digitised taking into account the different types of streams, and then each drainage line receive one of the following codes in a segment (vector) map: • "CHANNEL" for artificial drainage channels • "PUMPINGLINE" for pumping line from spring to mini-hydroelectric power plant. • "SALINE" for boundaries of the salt pans in the study area. • "STREAM1" for streams of 1st order. • "STREAM2" for streams of 2nd order. • "WACL" for the coast line border. • "WALK" for lake boundaries. • "WARV" for rivers. • "WASW" for swamp boundaries. After digitising a representation was designed according to the meaning of each line. In the case of springs, there was no clear topographic symbol to identify if the heads of the stream are permanent springs or not. In this case was assumed that all heads of streams were springs and a point map with its coordenate was created. This resulted in 123 spring points. The houses in general were also digitised as a point map making differences between three main features: • "HOUSES" for any real house or small industry. • "CEMENTERY" for small cemeteries in the study area. • "SCHOOL" for small schools located in the study area. In the study area there are 3317 "houses", 88 small schools and 4 small cemeteries. The appropriated representation was also created after all data was into the computer.

58



Finally the road system was digitised in the same way as the drainage system. Consequently different features codes were created: • "NATROAD" for the national road crossing the study area. • "PROVROAD" for the provincial road in the study area. • "PAVEDROAD" for secondary paved roads. • "RURALROAD" for non-paved roads. • "ROAD" for small roads. The representation of the road system was created according to the meaning of each line in the segment (vector) map. Once these maps were created, they were exported to CARIS GIS for the final analysis and for the Management system explained in the Chapter 8. In CARIS GIS the data was also coded in a sense to have appropriate cartographic representation.

5.7

Summary

The data preparation and processing was the longest stage during this research. The digital elevation model and its derivatives, the satellite products and their fusion, the aerial photographs and other topographic data were the main data to be surveyed and/or processed. The digital elevation model was created by interpolating the contour lines and the contour lines were entered using a "raster tracing line" system called SAMI (Semi Automatic Map Input). The procedure take varies steps to be followed in order the get the final contour lines. Reducing the steps is useful for reducing the final error. After finished the contour map, still some steps needs to be follow in order eliminate mistakes committed during the input. In the image processing the Landsat TM bands, SPOT PAN and JERS-1 SAR images were used. The landsat TM bands were first analysed statistically, spectrally enhanced and later some color composite and 3D views were created. With SPOT panchromatic image the standard deviation stretch method was applied for enhancement and later the image was very useful to combine with others images and in the interpretation. For the radar JERS-1 SAR image different speckle reduction and image enhancement tools were applied without an actual successful result. After the data was prepared and processed some data fusion product were developed in order to generate more products for the interpretation. The basic method was RGB-IHS and reverse conversions, changing the IHS components. The best results were obtained with DEM and SPOT and the Anaglyph image. Also, the photointerpretation was done using 1:37,000 and 1:25,000 scales aerial photographs for the geomorphological and landslides mapping. The photointerpretation was checked in the field and entered in the computer using the data fusion prepared before.


59


Another information was also surveyed mainly by digitising. This information is related to the drainage, roads and houses.

60


LANDSLIDES IN THE STUDY AREA

6 Landslides in the study area 6.1

Introduction

The occurrence of landslides in the study area is well known since in 1963 during hurricane Flora a large landslide occurred in Jagüeyes, Sierra de Caujerí, with three casualties, 5 houses totally buried and some hectares of agricultural land totally lost. This landslide occurred in a large scarp (Caujerí scarp) where a large number of multiple and successive landslides can be interpreted from the aerial photographs and checked in the field. However, after starting the photointerpretation it was realised that the landslides were not only in the Caujerí scarp but also in the coastal hills and in the northern part of the Baitiquirí area (Accumulational slopes). In both areas several landslides were mapped and they seem to have different characteristics and causative factors. For that reasons the description of landslides in this chapter was subdivided in three separate sections according to these three areas: Coastal hills, Accumulational slopes and Puriales de Caujeri Scarp. A number of factors, which have influence over the landslides have been identified and they can be more or less ranked in order of influence. But, the statistical relationship between the landslides and the factors/conditions is completely unknown. The main reason for that are the lack of information recorded and the range of time between the occurrence of one landslide and the other, especially during the last century. Nevertheless, factors controlling landslides occurrence are analysed and discussed as an intention for characterising the landslide processes in the different regions of the study area. For the photointerpretation, the boundaries of different landslide parts were surveyed. During photointerpretation it was realised that in the study area the landslide problem had been continuously occurring one after another during a long period of time generating many multiple and successive landslides. It is particularly true for the Caujerí scarp and the Accumulational slopes areas, since the coastal hills seem more isolated landslides. Because the scale of the aerial photographs was between 1:37,000 and 1:25,000 and the final map was at 1:50,000 scale, there was a problem to determinate how much detail should be mapped. For example there are five landslide areas with less than 1 hectare in the final map. It means that still the current areas can be subdivided into smaller landslide parts specially if the final mapping is at more detail scale.

6.2

Previous research


61


There is no specific record about any landslide in the study area, although is known the disaster in Jagüeyes de Caujerí during the hurricane Flora in 1963. It is known in 1991 when a book was published containing one chapter about an inventory of the landslides in the coastal hills (only sea -southside) develop by the Institute of Tropical Geography of Cuba (Magaz, 1991). In this research the different geomorphological characteristics of the area are revealed and the landslide types occurring in the marine terraces are described. For the landslides surveyed they also recorded six morphometric measurement including: length of the head scarp, maximum depth, average width, average slope, total volume and number of marine terraces destroyed. As a conclusion they associate the landslide occurrence in the coastal hills to Pliocene-Quaternary earthquakes, without explaining why. In 1998 a research project was developed at 1:100,000 scale in the eastern of Cuba for the mapping of the principal hazardous geological processes (Castellanos, 1998). In the framework of the project the main types of landslides were recognised and an inventory and hazard map were produced. The hazard map was made using a multivariate statistical analysis (Turner, 1996) as a "data driven method".

6.3

General overview about the classification

For the landslides survey in the study area the same photointerpretation as for Terrain Mapping Units was used. For those areas, which were recognised as landslide parts, a different checklist was filled recording data for type, subtype, zone, subzone, age, depth, name and phase. The landslide types and subtypes are according to the classification described by Varnes (1958, 1978 and 1984). The subtype actually is classified by type of material: rock, debris or soil. In some classifications the term "soil" is changed by "earth", but in general all means the same (Table 6–1). Type Unknown Fall Topple Slide, Rotational Slide, Translational Spread, lateral Flow Creep

Subtype Rock fall, debris fall, soil fall Rock topple, debris topple, soil topple Rock slide, debris slide, mud slide Rock slide, debris slide, mud slide Rock spread, debris spread, soil spread Rock flow, debris flow, soil flow Soil creep

Complex Table 6–1. Landslide types and subtypes using in the geomorphological survey. The zone and subzone are in relation with the landslide parts. For this case specific classification was not used but the zones and subzones names were completed according to the specify areas during the survey. The Table 6–2 shows the landslides zones and subzones surveyed for a final scale of 1:50,000 using the aerial photographs at 1:37,000 scale. Because in some cases it was difficult to subdivide different landslides zones it was necessary to create zones like "Scarp-Body", Body-Transport" and even, 62



"All-Mixed" for those areas where the boundaries are not totally recognisable at this scale or where there is not an exact boundary to be mapped. As the Table 6–2 shows for almost all the subzones there was also surveyed a qualitative criterion about the activity. For do this; the amount of vegetation in the SPOT PAN image was used. The areas without vegetation were classified as "with recent activity". The Scarp subzones were classified according to the spatial location although there was also surveyed the "Tectonic scarp" as a one possible subzone. The tectonic scarp in the study area is a scarp created by a tectonic movement, specifically a normal fault. The foot-wall fault block goes up creating an scarp. Some areas have a combination of "Tectonic scarp", which is a genetic classification and any of the other: Back, Side or Intermediate scarp. In a case like this the predominant was used. Zones Unknown Scarp

Transport

Body

Depression Initiating Scarp-Body combination Body-Transport combination All-mixed (undifferenciated)

SubZones and activity Back scarp, with recent activity Back scarp, with no-recent activity Side scarp, with recent activity Side scarp, with no-recent activity Intermediate scarp, with recent activity Intermediate scarp, with no-recent activity Tectonic scarp, with recent activity Tectonic scarp, with no-recent activity Upper part, with recent activity Upper part, with no-recent activity Lower part, with recent activity Lower part, with no-recent activity Side slope on the body, with recent activity Side slope on the body, with no-recent activity Blocks on the body Normal body Remain body surface With transversal cracks Slightly undulate Slope on depression With recent activity With no-recent activity With recent activity With no-recent activity With recent activity With no-recent activity

Table 6–2. Landslide zones and subzones surveyed for 1:50,000 mapping scale. The transport zone was only possible to classify in "Upper" and "Lower" parts and this zone is strongly related to the flow landslides where the transport zone is still recognisable. "Blocks on the body" subzone was called where close to the head of the landslide an area with large blocks is recognisable and can be separated from the rest. In others parts there are some areas which belong to former INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

63


landslides bodies, but because a number of successive and multiple landslide cover the same area, a small surface remain and was called "Remaining body surface" (Table 6–1). It is important to note that these classified areas are highly dependent on the scale of mapping, especially because at more detailed scale other landslides may be recognised within areas that now have other classification. In the study area there was a relation between the type and subtype with the zones and subzone of landslides surveyed. In this sense, the Scarp zones were more characterised by for Falls, Topples and Slides while the Transport zones was usually related to Slides and Flows. The age classification was very subjectively surveyed in the study area, since, as was mentioned before there are no records about when the landslide had occurred in the past, Only one landslide date is known, in 1963 during the hurricane Flora. For that reason the possibilities were reduced to "Very old", "Old" and "Historic". "Very Old" landslides are as old as Pleistocene while "Old" term was used where the landslide landform remain but either new landslides had occurred on it or the activity seem to have ceased. The "Historic" landslide was called to those areas where landslides seem to have occurred less than 100 year ago approximately and some remarkable features remain as a signal of its activity. The estimation of the depth was also subjective and only two possibilities were considered "Deep" and "Shallow" as an inference of height differences between the head and the lower part of the Landslides. Finally, two more fields were available in the checklist and in the database, the "Name" of the landslides and the "phase". Both were not totally filled in the database due to the scale of survey and the lack of time. The "Name" field was designed with the purpose of identify each landslide following geographical location features. But in some parts there are around 20 landslides in a relative small area. "Phase" is referring to multiple generation of landslides where the dependence between the landslides are recorded by temporal phases I, II, III, etc. All these information is attached to each landslide part in a GIS context where the Terrain Mapping Units (TMU) map is linked with the Landslide database (in Microsoft Access format). The managing and analysis was done in CARIS GIS in both directions: from map to database (spatially) and from database to map (by the attributes). An explanation about the database and its management can be found on Chapter 8.

6.4

Coastal Landslides

The coastal landslides are located on both sides of the coastal hills with 34 terrain mapping units. Since the type of material and the landforms are different on both sides, the landslides are also different in both sides. In the coastal hills four main landslide types can be mapped: rock fall, rotational rock slides, debris slides, and possibly translational rock slides. Rock falls are typical in the marine terraces and they seem to be originated by a combination of karstic dissolution and physical weathering process and then, triggered by earthquake or intensive rainfall. Large blocks can be found in the lower marine terraces, which clearly have been transported 64



from the upper terraces. Dating the rock falls is not easy task, since there is not any report about blocks rolling down hills. On the upper part of the slope there are many blocks with high susceptibility to falling down hills. Most of the rock fall zones were classified as scarps. Due to wave erosion on the cliff at the coastline, large blocks also fall down into the sea time by time (Figure 6-1D). The block dimensions are about 5 to 8 meters height by 3 to 5 wide. On top of the cliff various cracks delimited the boundaries of new block falls. The cracks has different widths depending on the how advanced the "breaking process" is.


65


Figure 6-1. Various Coastal hills landslides. (see text for explanation) Debris flows are common in the North slope of the coastal hills where the terrigenous materials of the Fm Maquey are sliding very often. Some debris flows can also be recognised in the east and west side of the Coastal hills (Figure 6-1C). On this sides the debris are fragment of carbonated material. The occurrence seems to be continuous and in small proportions rather than sporadically in large debris slides. In some parts also seem to occur small debris slides in the body of former large landslides. The depth is variable since some are from the top to bottom of the slope and other only take place in one part of the slope. They are associated with non-vegetated areas in the slope. For some debris slides only the transport zone remains for the rest the entire sequence is still recognisable. Rotational and translational rock slides were mapped in the marine terraces by Magaz (1991). They found five landslides, three rotational and two translational and calculated the morphometric parameters. In this thesis, theses five landslides were confirmed and classify according to the standard international classification (Varnes, 1984). The rotational landslides are large and usually cover from the top to the bottom the coastal hills. In the area is remarkable the Landslide in Los Aposentos coastal hill (Figure 6-1B). Los Aposentos landslide is more recent than others are since the lower terrace (Holocene) was also destroyed. Other landslides are pre-Holocene (Baitiquirí west and San Antonio del Sur east- Figure 6-1A) because the lower terrace was deposited on top of its slid material. Inside of Los Aposentos slide two more small rotational rock slides occurred. In the eastern coastal hills of the study area some landslides give the impression to be translational. Due to lithology differences and the dipping of the rocks some intermediate more terrigenous layers seem to be the sliding surface on which the upper layer slid down. This movement is still going on in the west part of Macambo and in other areas out of the study area boundaries. Figure 6-1 shows different samples of landslides in the coastal hills of the study area. In upper part the coastal hills and the locations of different landslides be recognise from a Landsat TM band 4 image. The Landslide "A" is a pre-Holocene rotational rock slide in a aerial photograph at 1:37,000 scale. The dating pre-Holocene is possible due to the Holocene material (very white areas between the sea and the landslide) deposited after the landslide occurrence. The landslide "B" is the located at Los Aposentos hill. The aerial photograph at 1:37,000 scale (B-upper picture) shows the different small landslides in its body and the field picture (B-Lower picture) shows the scarp and the body of the western part of Los Aposentos landslide. The landslide "C" is a debris flow sliding down hill in one the side of Baitiquirí bay.

6.5

Landslides in Denudational slopes

The Landslides in the denudational slopes occur actually in the area between the denudational slopes and the limestone hills (Figure 6-2). The landslides are more concentrated in the north part of Baitiquirí and El Naranjo. As the area is strongly affected by active tectonic faults the landslides are located in three main "steps" from the limestone hills toward the coast. In the upper part the landslides occur in the limestone rocks from Yateras formation (Figure 6-2A). The next two steps the landslides 66



occur over terrigenous materials form the Maquey, Cilindro and San Luis formations (Figure 6-2B and C).

Figure 6-2. Landslides in Denudational slopes. (See text for explanation) Figure 6-2 shows the main landslide area. In the upper left image the Terrain Mapping Units boundaries are over the SPOT PAN image. Pictures A and D are landslides on limestone hills. The picture D is specifically the block rocks that may be a topple landslide type. Both pictures B are different views of El Naranjo Landslide. In the left picture part of the body with small blocks are shown and the right


67


picture the scarp shows the limit between different rock layers. The picture C has small landslides originated in the slopes of the denudational slopes. Towards the south-west, the head of the landslides is almost exactly located in the main scarps of the active faults and some tectonic cracks can be recogned in the aerial photographs. In the north part of Baitiquirí the area is different in the sense that the head of the landslides do not follow the main fault line system SW-NE (Figure 6-2). In this part the area has a rounded shape. However, the area is located exactly in the south part of the N-S lineament system (see Figure 3-5) and another small lineament can be see in the W-E direction. That can be interpreted, as the north part of Baitiquirí is actually a branch point where at least two main fault systems coincide. This generated a weakness area due the convergent activity of both fault system and therefore, this weakness may play an important role in the landslide occurrence. Due that the upper layers are composed of compact limestone rocks there are many landslides where large rock blocks were broken and displaced a few meters from their its original position. This is particularly most common in the upper or last step of this landslide area (Figure 6-2D). The landslides in the lower steps are more difficult to delimit because they are in a multiple and successive way, meaning that they had been occurred one on top of the others during a long period of time and consequently the most recent landslide overlap and destroy the previous landslides. The main landslide types in this area are rotational rock and debris slides. The rotational landslides were mapped in the upper parts and the debris slides are more common in the lower parts. It is possible to recognise rill and gully erosion on top of the current bodies. In term of the landslide zones there is a repetition of zones due to the multiple occurrence. Then, the normal sequence scarp-bodytransport zones is usually transformed to scarp1-scarp2-scarp3-body-transport or many other possible combinations. The differentiation of various scarps can be done because the changes of the slope shape. In some parts it is very straight and in other parts different concave slope shapes occur. The same problem with dating was encountered in this area due the lack of record. Most of the landslides were dated as old or very old in the database. The most recent known landslide was in north part op the locality El Naranjo dated in 1997 during an intensive rainy season (Figure 6-2B). The landslide occurred just on top of a mini-hydroelectric power plant and the cause seems not only natural. The local people describe there was a tube from an upper spring, which brought the water to the power plant. This tube was broken for many months and the water was leaking in the slope of the hills during a long period of time. Once the rainy season came the hill was already semi-saturated with water and the landslide occurred. This occurrence was in terrigenous rocks belongs to the Maquey and San Luis formation.

6.6

Landslides in "Puriales de Caujeri" scarp

"Puriales de Caujerí" scarp is well known as a landslide area as was mentioned before. However, there is not a specify study in relation with the landslide problem in this area. The area is just down slope to the Sierra de Caujerí (Figure 6-3). It is almost a straight large scarp with height differences between 300 to 400 meters in N-S direction along 15 kilometers approximately. At the north the scarp ends 3 to 5 kilometers to the north of Mameyal with less recent landslides. At the south the scarp ad68



joins to Sierra de Mariana scarp creating another area with large landslides. The branch point is not sharp as in the north but completely rounded where many landslides show more recent activity.

Figure 6-3. Landslides in Caujerí scarp. (See text for explanation) INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

69


Figure 6-3 shows different views of the landslides in Caujerí scarp. The upper left image is a Landsat TM true-color composite 321 (RGB) with the location of the other images. The picture A, B and C correspond respectively to the upper, middle and lower part of the scarp. The picture C is a gully or ravine erosion in coluvial deposits of former landslides. The gully is 2 meters depth, 15 meters long and 3 to 4 meters width. The picture E is a 1:25,000 scale aerial photograph showing the details of the Jagueyes landslides. The landslides in Caujerí scarp follow more or less the same pattern in the whole area. At the highest part there is a large scarp, about 100 meters height, which cut almost vertical the limestone layer of Yateras formation. In some parts the underlain rocks of the Maquey formation can also be recognized. This scarp is actually the back-scarp of multiple landslides and the main lineament of the N-S lineament system. It is difficult to subdivide this long scarp into the individual landslide scarps due to the very linear behavior of the main scarp. At the lower part of this scarp the slope changes shape. In some parts the slope is straight steep, where landslides are more recent, and some parts is more concave without being an exactly limit between the scarp and the higher step. After, the main scarp presents different profiles (Figure 6-4). The north part (Figure 6-3A) presents three well-recognized steps with about 50 to 100 meters height differences, which in some parts are interrupt by smaller landslides going across the main scarp to the Caujerí valley. These smaller landslides made a mini-basin shape and are active in term of rill and gully erosion. The middle part (Figure 6-3B) is the more active and many multiple and successive landslides can be mapped. The landslides can occur from the main scarp like the 1963 landslide. However most of them actually happen from higher step up to the valley presenting many multiple landslides, which can be temporally ordered. The south part (Figure 6-3C) is where the retrogressive processes have been stronger and therefore it is in general wider. Here the landslides present several forms but in general are as in the middle part. 8 00

7 00

6 00

5 00 No r t h 4 00

Ce n t e r S out h

3 00

2 00

1 00

0 0

500

1 00 0

1 50 0

2 0 00

Figure 6-4. Different profiles in Caujerí scarp 70



In Caujerí scarp were mapped around 150 landslide zones, which have reasonable size on the 1:50,000 mapping scale. But in more detailed scale many other landslide features can be recognized. Due to this scale problem many zones were categorized as mix of "scarp-body", "body-transport" or even "all mixed". There are also many "remnants of body surface". According to the classification used above (Table 6–1) in the Caujerí scarp the landslides are mainly rotational rock slides, although there are also various debris flows. The sizes of the slide vary from very small (few meters) to large (2 kilometers) from the head to lower-recognizable part of the transport zone. Due to the long period of the occurrence of landslides in the area, occasionally can be found in current agricultural land large gully and ravine erosion (Figure 6-3D). This happens because the colluvial deposits are not totally consolidated and during the raining season sub-surface streams eroded until remove the upper material and created large gullies. The flow occurrence is a consequence of removal material already slid. In relation with the time many landslides were classified as "Old" and "very Old", although some are "Historic". The most recent landslide occurred in 1963 during the hurricane Flora in Jagueyes locality (Figure 6-3E). With Jagueyes landslide three people died, five houses were buried, many cattle were lost and some agricultural land was also lost. After visiting the area, interviewing a survivor of the disaster and studying in detail the aerial photographs it was conclude that in fact there were two landslides which occur in a time lap of 45 minutes. It seems that after few days raining the water level went up the start to push the large block in the main scarp. Once the force was enough a first huge crash fell down massive rock blocks from the scarp. After that, the hydraulic and gravitational energy was accumulated in a relative short period of time generating the final and longest landslide that reached about 3 kilometers from the head. This hypothesis is also confirmed by the aerial photographs, which show both landslides coming from a single main back scarp. Close to Jagueyes landslide there are few small landslides, which also occurred during the Flora hurricane.

6.7

Other Landslides

In the study area other landslides located in different areas than Coastal hills, Denudational Slopes and Caujerí scarp were also mapped. These landslides are either related to the metamorphic hills or to the terrigenous hills. They do not follow a specific spatial pattern and they are not associated to a specify reason such as tectonic weakness lineaments, lithological contact, etc. There is not complete certainty that some of the disperse landslide are in fact landslide and it was not possible to check all of them in the field due to the inaccessibility of the areas. For locating these landslides was used the aerial photographs where they show a spoon-shape as typical rotational landslides. In the study area was also mapped some Terrain Mapping Units classified as "Steeply to Very Steeply Face on the Slopes" in the legend. In many of these areas intensive denudational processes are also occurred and most probable at more detailed scale they can be classified as certain landslide type. It is INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

71


important to note also that due to the scale of mapping many very vertical slopes could not be mapped as they actually are. The rivers cut most of these areas and some of them are up to 50-meter height.

6.8

Summary

Landslides in the study area not only occurred in the Caujerí scarp but also in the coastal hills and in the north part of Baitiquirí area. In theses areas several landslides were mapped and they seem to have different characteristics and causative factors. The boundaries of different landslide parts were surveyed using photointerpretation. It was realised that landslide occurrence had been continuously happening one after other during long period of time generating many multiple and successive landslides. For survey of the landslides in the study area the same photointerpretation than for Terrain Mapping Units was used. For those areas, which were recognised as landslide parts, a different checklist was filled recording data for type, subtype, zone, subzone, age, depth, name, phase. For better understanding of the landslides they were subdivided according to the three main occurrence areas: Landslides in the coastal hills, in the acumulational slope and in Caujerí scarp. In the coastal hills four main landslide types can be mapped: rock fall, rotational rock slides, debris slides, and possibly translational rock slides. The rockfall are typical in the marine terraces. The debris flows are common in the North slope of the coastal hills where the terrigenous material of the Fm Maquey sliding very often. The rotational landslides are large and usually cover the entire side of the coastal hills. In the area the landslide in Los Aposentos coastal hill is remarkable. The Landslides in the denudational slopes occur in the limited area between the denudational slopes and the limestone hills. As the area is strongly affected by active tectonic faults the landslides are located in three main "steps" from the limestone hills towards the coast. The main landslide types of this area are rotational rock and debris slides. The rotational landslides were mapped in the upper parts and the debris slides are more common in the lower parts. The landslides in the Caujerí scarp follow more or less the same pattern for the whole area. At the highest part there is large scarp, about 100 meters height and the lower part of this scarp the slope changes in shape. In Caujerí scarp areas were mapped around 150 landslide zones scarp. The landslides are mainly rotational rock slides, although there are also various debris flows. The most recent landslide occurred in 1963 during the hurricane Flora in Jagueyes locality. In the study area was also mapped another landslides randomly located in different areas These landslides are either related to the metamorphic hills and to the terrigenous hills.

72


LANDSLIDES HAZARD ASSESSMENT

7 Landslide Hazard Assessment 7.1

Introduction

Mapping landslide hazard has become a complicated matter when the purpose is to do it seriously. Although many methods have been implemented (see Section 2.3) a reliable determination of areas with different probabilities of landslide occurrence is always affected by the availability of the data. Most of the time the available data are not enough or the accuracy is not appropriate. On the other hand, the optimal hazard-mapping model should be able to map areas with certain probability of occurrence for certain period of time and with certain magnitude. In other words, combine spatial, temporal and magnitude probability in one model considering also the possibility of more than one triggering factor. Such models are still under research and may take some time to be really tested. Most often the landslide hazard maps show areas with qualitative classes as LOW, MODERATE and HIGH and some considerations regarding the landslide expectancy and the land use development in such areas. To consider that these maps are not useful may be a wrong approach since these maps somehow present expert knowledge on the subject. To reach the hazard-mapping goal different ways may be used as were listed in Section 2.3. In this research a Heuristic Analysis method was used in order to obtain the final hazard map. A discussion about heuristic analysis will be given in Section 7.3. The selection of this method was done considering that: 1. Once a TMU map has been created the use of this map in the statistical analysis methods will produce a biased results because of the strong spatial correlation between the landslide inventory map and one (or some) class(es) in the TMU map. 2. The use of any deterministic analysis methods needs more detailed scale maps and some engineering parameters. Both of them were not available in this research. Due to the geological and geomorphological characteristics of the evolution of Cuba, the territory presents a "mosaic-type" configuration (Magaz, 1996). Meaning that in relative small areas the physicgeographic conditions have a high variability and therefore, the morphogenetic processes and landforms have great spatial and temporal diversity. Large landslides are not predominant in Cuba, which is different from many other countries. Large landslides are not expected to be of regional importance because there are not enough weak materials like volcanic ashes. Also the thickness of the weathering crusts and the layers of detritus are not large enough. In Cuba denudational surfaces are predominant with a shallow weathering crust. These surfaces are related with a strong "erosional period", which occurred since early Paleogene during the three transgression-regression phases. For that reason landslides in large proportions causing significant damage or changing abruptly significant landforms are less possible.


73


However, small landslides associated to slopes are more common in mountain regions where weathering processes have played an important role. Although the average size of individual landslides is small, in some parts many landslides have occurred related to tectonically weak areas generating regional or large problems. This is especially true for the study area where a large tectonic scarp (Caujerí Scarp) together with other factors have produced a large landslide zone. Once the landslide hazard map was done, the importance of the flooding hazards was also considered in order to get a more comprehensive hazard map for disaster management. Consequently the flooding hazard areas were mapped and added to the final hazard map. How it was done is discussed in Section 7.5.

7.2

The main causative factors for landsliding

The study area presents particularities in relation with the landslide occurrence in the sense that most of the landslides are concentrated in specific regions. The existing landslides can be found in the regions already discussed in Chapter 6: Coastal hills, denudational slope boundary and Caujeri scarp. From this geographic behaviour it can be inferred that the causative factors for the landslides are also located in these three areas. After analysing the three existing landslide areas it is recognised that the extensional faulting has played an important role in the location of the current landslides especially in the Denudational slope boundary and in the Caujerí scarp. Both areas present large tectonic scarps. In denudational slope boundary instead of one fault scarp there are a number of scarps (at least three) due to a sequential normal faulting. Although tectonic features are present in the coastal hills it does not appear to be the most influencing factor. Dating these fault systems requires a detailed structural tectonic study, which will be also useful for describing in more detail its influence on the landslide occurrence. In relation with the tectonics the general and recent uplifting of the area is important. This vertical movement seems to have created an inbalance between the landform generation (by the uplift) and erosional processes. As a consequence the instability of the slopes generated gravitational movement as landslides. This phenomenon happens especially in the coastal hills. The lithology plays an important role in landslide occurrence especially the limestone layers which are near horizontally overlaying the terrigenous material of the Maquey formation. Karst processes are present in the limestone rocks with different intensities following the joint and fault directions. When the karst processes have dissolved enough limestone the surface water start to have direct influence on the underlying terrigenous material. Groundwater is also an important factor in the study area. Due to the active tectonics the area can be subdivided in different blocks with different groundwater levels. The groundwater in the Limestone Hill area is affected by the karst processes but in general the water table is much higher than in the surrounding: the Caujerí Depression and the Denudational slopes. It can be recognised by the positions of the springs along the Caujerí scarp and even in the Limestone Hills itself. It seems when the 74


LANDSLIDE HAZARD ASSESSMENT

water table rises in the rainy season the lateral hydrostatic pressure in the Limestone Hills can generate landslides (Figure 7-1), especially during intensive rainfall usually associated to the hurricane season in October and November.

Figure 7-1. Lateral groundwater pressure when water table rises. The influence of the precipitation can be analysed in two ways: in the short term, during intensive rainfall and in the long term, during the annual seasons. The precipitation is recorded as one triggering factor for landslides in the study area. In fact intensive rainfall during a hurricane triggered the only two landslides with known dates in the study area. During the year the area has extreme different conditions. In the dry season it is very dry (few centimeters rainfall per month) and in the rainy season it is rainy almost all days. This situation, together with the high temperature and humidity contributes strongly to the chemical and physical weathering fragmenting the rocks into blocks, which later fall down slope. As was mentioned before earthquakes have been recognised as one of the triggering factors and for that reason they are also considered in the research. However, there are no landslide records that allow to establish a relationship between earthquakes and landslides in the study area. Summarising the main causative factors it is possible to separate them into two groups. The triggering factors, which act suddenly, and the intrinsic factors, which during a long term period "prepare" the landforms conditions for landslide occurrence. Figure 7-2 shows a diagram in which these factors are separated in these two groups. As can be seen, there is a certain relation between both groups because the rainfall has a strong influence on karst processes, the seasonal climatic changes and the groundwater table and earthquakes are related with the general uplift and the tectonic faulting.


75


Causative factors Triggering factors

Intrinsic factors

Rainfall

Karst processes

Earthquakes

General uplift Seasonal climatic changes Groundwater table Lithological control Tectonic faulting

Figure 7-2. Causative factors for landslide occurrence in the study area.

7.3

The heuristic landslide prediction model for the study area

For evaluating the areas where landslides can occur a heuristic analysis was used. The method was classified as a Qualitative Weighted method (van Westen, 1993). The general idea is to assign weights to a number of maps, which are considered as important variables in the occurrence of Landslides. After assigning weights a combination formula is used to integrate all the weights and produce a final map. The final map is classified into a number of classes and the hazard areas are mapped according to the expert opinion. The general procedure is shown in Figure 7-3. The different steps follow more or less the decision support system (DSS) methodology (Saaty, 1996). The first step was to select the components of the model and characterised them. The components of the used model for landslide hazard mapping in the study area are shown at the lower level of the tree structure in Figure 7-4. The components were organised in a tree-shaped structure and ranked according to their importance to generate landslides. The upper level of the components was called criteria (in ranked order): 1. Geomorphology 2. Topography 3. Geology 4. Tectonic 5. Hydrology

76



Select the components Design the hierarchical relation Assign weight for each level Standarise the maps Running the model Classifying the final map Final map

Figure 7-3. Flowchart for heuristic landslide hazard analysis. The variable (components) with sub-ranking per criteria are as follows: For geomorphology: 1.1 Landslides zones 1.2 Geomorphological subunits For topography: 2.1 Slope 2.2 Internal relief 2.3 Shape For hydrology: 5.1 Spring 5.2 Drainage distance

Components of the Heuristic Landslide Prediction Model

Topographic Slope

Slp1

Slp2

Internal Relief

Slp3

Geology Shape

Slp4

Tectonic

Formation

Active Faults

..etc...

Spring1

Hydrology Springs

Spring2

Drainage Density

Spring3

Geomorphology Subunits

Spring4

Landslides Zones

..etc...

Criteria

Variable

Classes

Figure 7-4. Components of the heuristic landslide prediction model.


77


Because the other criteria only have one variable they do not need to be ranked. The variables used are described in Table 7–1. The description “relation” can be either benefits or costs depending on the relation of the variable with the possibility of landslide occurrence. Consequently benefits means "the higher the better" (for example: high slope, high possibility of landslide occurrence) and costs "the higher the worst". The class boundaries in case of numerical variables were selected taking 25 cumulative percentage of the histogram. The classes are shown in Table 7–4. The description "Scale" is referring to the scale of measurement (Bonham-Carter, 1996). Variable Slope Internal Relief Shape

Geology Faults Springs Drainage Distance Geomorphological Subunits Landslides subzones

Origin From the original DEM using the methodology of ILWIS From the original DEM using the methodology of ILWIS From the original DEM using the methodology of ILWIS

Scale Interval

Degrees

Relation Benefits

Ratio

Meters/hectares

Benefits

Ratio

Benefits

By reclassifying the TMU map Calculating a distance from the fault map and classifying in four classes Calculating a distance from spring points and classifying in four classes Calculating a distance from the drainage map and classifying in four classes By reclassifying the TMU map

Categorical Ratio

No meaning. >o concave, 0 straight, >0 convex. N/A Meters

N/A Costs

Ratio

Meters

Benefits

Ratio

Meters

Benefits

Categorical

N/A

N/A

Categorical

N/A

N/A

By reclassifying the TMU map

Units

Table 7–1. Variable used in the prediction model. See text for explanation. (N/A- No Applicable) The methodology for assigning weights in different levels is according to the Analytic Hierarchy Process (AHP) developed at the Wharton School of Business by Thomas Saaty (1996). The advantages of this method include: • Allow the application of data, experience, insight and intuition in a logical and thorough way. • Enables derive ratio scale priorities or weights as opposed to arbitrarily assigning them. • Accept to incorporate both objective and subjective considerations in the decision process. • The heuristic model is better structured and easily to compare group of elements. • Avoids more human' s mistakes because the assigning weights consider less elements. Once all the variables were characterised and classified in four classes the weights were assigned in the three corresponding levels: the criteria, the variable and the classes. For assigning the weights a Microsoft Excel application was created. In the application all the criteria, variable and classes with their weights were listed in tables and a simple weight summation formula was applied to test how the weights will result in the final landslide hazard map. Changing weights interactively permits to have an idea how the model will run for a single pixel and it is possible to check the extreme values and the average conditions. Weights were assigned by expert opinion, which is called Direct Method in decision support system jargon. For checking the weight assignment a decision support system called Definite was used (Janssen, 1994). The idea was to use two more weight methods and compare those

78



methods with the expert opinion. The methods are the pairwise comparison matrix and the ranking methods (Janssen, 1994). Topography Slope Internal Relief Shape Geology Formation Tectonic Active faults Hydrology Spring Drainage density Geomorphology Subunits Landslides zones Total for criteria

Direct Method 0.3 0.7 0.2 0.1 0.2

Pairwise Matrix 0.224 0.7 0.2 0.1 0.131

1 0.05

1 0.040

1 0.05

1

0.5 0.5

1 0.065

0.5 0.5 0.566

0.4 0.6 1

1 0.065

0.038

0.4

Ranking method 0.257 0.7 0.2 0.1 0.157

0.5 0.5 0.457

0.4 0.6 0.999

0.4 0.6 1.001

Table 7–2. Weights for criteria and variables using three methods The pairwise comparison matrix is a matrix where each variable (or criteria) is compared to all other variables in order denote whether they are equally significant, or whether one of them is somewhat more significant / better than the other for the goal concerned. The ranking method simply means that the variables are ranked. In addition, it is assumed that theses ranking can be considered as units on a cardinal scale. Consequently the weights can be easily found by standardising the rank order (Voogd, 1983). Table 7–2 shows a comparison of the three methods, as can be see the results are very similar. For the pairwise comparison matrix method the inconsistency value was 0.08%, demonstrating that the weights are reliable enough. The inconsistency is a parameter to measure how randomly the expert judgements are. The value has a range from 0 to 100% and normally, inconsistency values below 10% area acceptable. As a conclusion the initial weights assigned by expert opinion were taken for the analysis. Classes

4 classes Intervals

1 2 3 4 5 6 7 8 9 10

0.500-12.275 12.275-24.050 24.050-35.825 35.825-47.100

Weights intevals for the Hazard map 6 classes 10 classes Intervals Intervals Area 0.500-8.350 8.350-16.200 16.200-24.050 24.050-31.900 31.900-39.750 39.750-47.100

0.500-5.160 5.160-9.820 9.820-14.480 14.480-19.140 19.140-23.800 23.800-28.460 28.460-33.120 33.120-37.780 37.780-42.440 42.440-47.100

TMU

14189.310 4176.3 18106.85 18360.87 2013.13 1702.47 968.78 718.59 291.42 439.36

88 40 125 97 85 87 77 59 36 24

Table 7–3. Weights intervals for the three hazards maps Because the weights were assigned in three levels and the GIS used only can process one level at the time, the two upper hierarchical levels (criteria and variable) were multiplied by the lowest level (the classes) in order to get one final weight per class. The ranges of the weights were different. In the criteria and variable levels the weights ranged for 0 to 1 and the total sum must be equal 1within the level and the criteria. As a result, to sum of all criteria weights must be equal to 1 and the sum of INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

79


weight within the Topography criteria must be also equal to 1. The weights for the classes level were ranged from 1 to 100 and the also the total weight values within a variable must be 100. Table 7–4 show the initial weights assigned to each class and the final weights after multiply by its correspondent upper levels. For example, the slope class shp4 (>20.37 degrees) have 50 multiply by Slope variable (0.7) and by Topographic criteria (0.3) is equal to 10.5, which is the final weight. When all the weights were assigned to each class the model was completed and executed. As a result several areas received weights in the range from 0.500 to 47.100. The next step was to re-classify the final hazard map in order to get an understandable number of classes. Variable

classes

Intervals

Slope

Slp4 Slp3 Slp2 Slp1 Inre4 Inre3 Inre2 Inre1 Shp4 Shp3 Shp2 Shp1

>20.37 >15.81,=10.41,=0.00,=19.34 >14.41,=9.49,==0.00,=0.60 >0.00,=-0.70,=-14.00,=1000 100 100-500 500-1000 >1000 >735 >304,=0,= (as operator for "greater than") • In Combo box Value: 2500 (as value for landslides polygon areas)


99


Step 6. Click on the Add button and the query will be added for select the landslide units, which area is larger than 2500 hectares. Step 7. Click on OK button (Figure 8-10) and the query is executed.

Figure 8-11. Part of query result list. The system selects 86 records in the attribute database and sends the keys of the corresponding polygons to the graphical interface. Figure 8-11 shows part of the results in the table in CARIS DB manager module for landslides with more than 2500 hectares. When the keys of the polygons are in the graphic interface module the corresponding polygons can be filled, highlighted or outlined. Figure 8-12 shows the polygons selected filled in another color only for the graphic area of Puriales de Caujerí scarp. The polygons filled are landslides or part of landslide with more than 2500 hectares. This selection can be converted in a raster zone or accumulated for other analysis. It is also possible to overlap this result with any other map in the system. For example it is possible to combine the result with the farm houses to see how many of them occur on the larger landslides. The analysis can be done also from the map to the attribute database. For example, if the central part of Puriales de Caujerí scarp is relevant from the previous selection it possible to re-select specific polygons in order get their attributes. In this case, instead of selecting data for the TMU project, the data will be selected for the Landslide project with the idea to record the landslide characteristics of these polygons. For that purpose the first step is to change the current project from TMU to landslides and the User Defined Area MapQuery tool can be used to delimit a box in the central part as Figure 8-12 shows. Figure 8-12. Graphical query result.

100



Once the area has been selected the polygon keys are sent to the CARIS DB manager, which looks in the attribute database for the corresponding records and report them.

Figure 8-13. Query result of the seven landslides. Figure 8-13 shows the result of the MapQuery for Landslide database. The different characteristics of the 7 landslides can be analysed in this table. As a conclusion the data related to landslide hazards can be managed in both the maps and the attribute databases, which are linked. The Landslide Hazard Management System can be improved with the addition of many other data and increasing the customisation. Besides, new maps and attribute databases can be made at more detailed scale and the system is able to perform a "vertical database" (explained in section 9.3) (Universal System, Ltd.1998).

8.7

Summary

The Landslide Hazard Management System (LHMS) is a system in which for certain areas the different hazards are mapped in digital format together with many other important types of information in a way that it is allowed to analyse, combine and model this information with two main purposes improve the disaster management in all phases and improve the regional and local planning. The system counts with a set of maps all together in a file and accessible thought one customised interface in CARIS GIS system and a set of attribute database in an application in Microsoft Access' 98. The Access application is a group of forms for entry and editing the attribute data with a minimum of possible mistakes. The LHMS has complete and direct access to the attribute database for querying and analysis. The system has two interfaces that have a customised menu in which the options can run macros and each macro may be as complex as is needed for executing a number of actions within the system or even running external programs. Both interfaces have an Auto-send and Auto-retry option, which allow making queries in the attribute database and send the keys to the map or make graphical selections and send the keys to the database. The results will also appear in the CARIS report template.


101

CONCLUSIONS AND RECOMMENDATIONS

9 Conclusions and recommendations 9.1

Introduction

During the research process many points were discussed and analysed in order to derive some conclusions but also raising many new "questions marks". Besides, some subjects were not covered as deep as was planned and for others there are clear needs for more detailed study. Nevertheless, the research has arrived at various conclusions and recommendations to be taken into account either for the disaster management of the study area and for landslide hazard assessment within a GIS context. The main problems initially considered were handled somehow during the research. The transparency in the landslide hazard assessment process was obtained through the Heuristic Model in which all the criteria, variable and classes were ranking according to their relevance with the landslide problem in the study area. In the same sense, the reclassification of the hazard map in classes was also explained in detail statistically and qualitatively. Once the hazard map was made, a management system in a GIS-context was designed in order to provide full accessibility to the hazard map and all other spatial information relevant for hazard management. In this sense, the landslide hazard-related information is available for natural disaster management planners. Therefore the hazard maps can be easily used in the different disaster management phases and not only in the mitigation phase, where is known it is more useful until today.

9.2

Main conclusions about the Landslides problem in study area

The landslide hazard problem in the case study area was slightly known before this research was carried out. Just because the landslide disaster occurred in 1963 the existence is known of a landslide in the Caujerí valley. A detailed explanation about the existing landslides was made in Chapter 6. Different steps need to be covered from the initial applied geomorphological survey to the final landslide hazard map. They were explained in detail in the Chapter 7: Landslide Hazard Assessment, particularly in section 7.3 where the heuristic prediction model was explained. The main conclusions can be summarised as follows: •

In general it is important to recognise the main criteria for the prediction model and to rank them according to their influence on the landslide occurrence. The ranking and weighting of the criteria, variables and classes is strongly depending of the study area conditions and it is very difficult to make standards in this sense. In relation with the hazard prediction methods it is important to


102


note that once a detailed geomorphological survey has been done and a Terrain Mapping Unit (TMU) map obtained, the statistical methods are less useful because the conceptual data model is so mature that the statistical model will be biased. •

The Heuristic Analysis method used called Analytic Hierarchy Process (AHP) is one among different methods for predicting landslide locations. For example it is also possible to assign the weights directly to the classes in relation with the landslides hazard instead of made a hierarchical structure. It has some advantages and disadvantages. For actually finding to most appropriate heuristic model in landslide hazard assessment still further research needs to be done comparing various methods.

•

The causative factors for landslide occurrence in the study area are a combination of two triggering factors (rainfall and earthquakes) and a number of intrinsic factors, which play an important role even at the time when the landslide is triggered (Karst processes, general uplifting, seasonal climatic changes, groundwater table, lithological control and tectonic faulting). In relation with time the most common processes are: 1. On the long term the general uplifting and the tectonic faulting created certain instability, 2. On the medium term the karst processes and the lithological control dominate the erosional processes and increased the dangerous conditions for landsliding. 3. On the short term the seasonal climatic and the groundwater table changes increase the hazard. 4. Very short term a heavy rainfall or an earthquake can trigger landslides in some parts of the study area by the continuos accumulations of the energy due to the previous factors.

•

The study area is more affected by different erosional types and processes rather than accumulative ones. Even the Caujerí valley is being affected by erosion from sheet wash to large gully formations. This problem has direct influence on the agricultural development of the study area and should be studied in more detail. Various karst processes, which affect in general the landforms, also influence the limestone rocks and should be taken into account for local and regional planning.

•

The existing landslides in the study area are mainly located in to three areas: the denudational slopes boundary, the coastal hills and the Caujerí scarp, with increasing importance. The landslide occurrence seems to continue without any apparent measure will be able to controled. Then, as an effective disaster prevention measure the HIGH hazard areas must not be invaded by any condition that increases the existing element at risk already present. The Landslide hazard map was designed in a way that can be useful either for disaster managers with a simple legend (three major hazard classes) and for planners or researchers with a more technical information (10 detailed hazard classes).


103

CONCLUTIONS AND RECOMMENDATIONS

9.3

Main conclusions about the Natural Hazard Management

For the natural hazard management a Landslide Hazard Management System was designed using the GIS as a framework (Chapter 8). Although the system is not in an operational stage many useful analyses can already be done and at the end, with some additions the system can become an operational tool for disaster management in the study area or even in any other area when changing the data set. The theoretical background about the designing of a Disaster Management System, as a next step, was also covered on this research with many practical recommendations in Chapter 2. After this research some conclusions were arrived including: •

The use of GIS as basic platform for a Disaster Management System is possible and highly recommended. The reasons are based on: 1. The diversity and amount of data types that can be handled in a GIS context. 2. The analysis and modelling capability of the many existing GIS systems. 3. The modular and customisable GIS structure allowing many external connections and data transfer. These reasons become more evident every year with the new advances of the information technology like the Open GIS philosophy.

•

The customisation of a GIS for Disaster Management should be done in a way that the system changes from the tools-oriented approach to the task-oriented approach. With the task-oriented approach the system will be able to execute several disaster management tasks organised by disaster phases and by the scale of the disaster. On the other hand, the system should also be able to execute several external modules, which facilitate the use of many well-developed independent programs. For this facility the system needs to have efficient data transfer protocols.

•

The CARIS GIS system is a suitable GIS for developing a Disaster Management System because it already contains many of the properties needed for the purpose. Besides, the system is able to implement horizontal and vertical databases. With the horizontal database the study area can be extended without loosing any setting and the spatial area for management can grow gradually. In vertical databases capability, the system is capable to handle in one set-up different scales of mapping with its corresponding attribute database and therefore, different scales of disaster management. Nevertheless, some additional capabilities are desired like more raster analysis and in general the handling of the raster data. Another mainly raster GIS like ILWIS is very suitable for analysis and modelling but does not complete satisfactory the other two reasons for being used in disaster management.

•

The management of natural disasters requires a better understanding of: 1. The disasters itself and the data requirements. 2. The disasters mapping methods and mapping requirements. 3. The disasters management and user requirements. 4. The GIS technology and technology requirements. 5. The GIS linked to natural disaster management and operational requirements.

104



The survey of these aspects should be adapted to the local conditions were the system will be operated. In this sense, the system is more useful at the level where the managers make decisions rather than where the decisions are executed.

9.4

Recommendations for further investigations

Some recommendations have already been done and stated early in the document. They are related to the subject of each Chapter. Additionally two more are explained here: •

In the study area there is a need to continue the landslide hazard mapping at more detailed scale with the order of importance from Caujerí scarp, Coastal hills to Denudational Slope boundary. The new mapping should be focused on the different landslide parts and their temporal relations, the type of landslide materials and the relief features. For a geomorphological point of view, the study area requires further study of the different erosional processes and their strong influence on the agricultural development especially in the Caujerí valley.

•

It is recommended that the designed Landslide Hazard Management System is completed, improved and implemented in an operational way in order to continue this case study in the implementation phase.

9.5

Final remarks for ITC

For ITC, especially for the Applied Geomorphology Division, the following is recommend: •

The creation of a comprehensive attribute database system for data acquisition for geomorphological mapping, TMU mapping and natural disaster mapping. The data base structure should be able to be linked to a GIS. The system can be created through M.Sc. theses by different modules.

•

To increase the teaching skills in how to management natural hazards, additionally to the natural hazard mapping techniques. The purpose is to be closer to the final user: the disaster managers.


105

REFERENCES

References Alexander, David. E. 1995. A survey of the field of natural hazard and disaster studies. In Geographical Information Systems in Assessing Natural Hazards: Select contribution from an international workshop held in Perugia on September 20-22, 1993 / ed. by A. Carrara and F Guzzetti. : Klumer Academic, 1995. -353 pp. Aspen Global Chance Institute (AGCI). 1996. Workshop on Natural Hazards and Global Change, URL: http://www.gcrio.org/agci-home.html. Bonham-Carter, G.F. 1996. Geographic Information Systems for Geoscientists: Modelling with GIS. Pergamon, Elsevier Science Ltd. New York, 398 pp. Carter, Nick. W. 1991. Disaster Management: A Disaster Manager' s Handbook. Asian Development Bank (ADB), 417 pp. Castellanos, E., Carrillo, D., Diaz, O., Pérez, R. and Garcia, J. 1998. Manejo de amenazas geológicas: apuntes para su implementación en el ejemplo del proyecto CARTAGEO. Memorias Geología y Mineria' 98, La habana, 1998, Centro Nacional de Información Geológica, Instituto de Geología y Paleontología. Vol. I, pp 109-112. Castellanos, E., Carrillo, D., Díaz, O., Alfonso, W. 1998. Consideraciones sobre el diseño de una DB-SIG para la localizacion, prediccion y manejo de procesos geologicos amenazantes. Memorias Geología y Mineria' 98, La habana, 1998, Centro Nacional de Información Geológica, Instituto de Geología y Paleontología. Vol. II, posters section. Chang, J.L. and V. Suarez. 1998. Fuentes magneticas anómalas profundas y su implicación en el modelo tectonico de Cuba oriental. p. 169-172, In Memoria del tercer congreso cubano de geología y mineria, 1998. Vol. 1, pp. 737, Sociedad Cubana de Geología, Coppock, J. Terry. 1995. GIS and Natural Hazards: an overview from a GIS perspective in Assessing Natural Hazards: Select contribution from an international workshop held in Perugia on September 20-22, 1993 / ed. by A. Carrara and F Guzzetti. Netherlands: Klumer Academic Publishes, 1995. -353 pp. Cova, T.J. 1999. GIS in Emergency Management in Geographical Information Systems Vol.2 Management Issues and Applications (Second Edition) Ed. by Longley, Paul. A. et al. 580 pp. Eastman, J.R., W. Jin, P.A.K. Kyem, and J. Toledano. 1995. Raster procedures for multicriteria/multi-objective decisions. Photogrammetric Engineering and Remote Sensing 61(5); 539-547 pp. ERDAS Imagine production tour guides. 1991. ERDAS imagine version 8.3.1`for windows. Atlanta ERDAS, 162 pp. Federal Emergency Management Agency (FEMA). URL: http://www.fema.gov/. Flores R., G. Millan, J.L. Chang, C. Pérez, E. Castellanos, K. Nuñez. 1998. Tectonica de Cuba Oriental. In. Geología y Minería, Memorias I (III Geological and Mining Congress). Havana. p 240-243. Franco, G. L. et al. 1992. Lexico Estratigrágico de Cuba. Instituto de Geología y Paleontología. La Habana. 410 pp. INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

106


Gutierrez, G. O. et al. 1989. Nuevo Atlas Nacional de Cuba. Instituto de Geografía, Instituto de Geodesia y Cartografía. 85 p. ILWIS-PCI. 1999. ILWIS User' s Guide 2.23, Enschede, 511 pp. Ingleton, J. Ed. 1999. Natural Disaster Management. Leicester: Tudor Rose Holdings Ltd. 320pp. ITC/AGS. 1998. ITC/AGS Annual report, 1998, 15 pp. Inturralde-Vinent, M. A. 1996 (Ed.) Cuban Ophilites and Volcanic Arcs. Project 364: Geological Correlation of Ophiolites and Volcanic Arcs Terranes in the Circum-Caribbean Realm. Contribution No.1. 256 pp. Janssen, R. and M. van Herwijnen. 1994. Multiobjective decision support for environmental management + DEFINITE Decisions on an FINITE set of alternatives. Dordrecht. Klumer, 132 pp. Kaewsonthi, S. and Harding, A. G. 1992. Starting, Managing and Reporting Research. Chulalongkorn University Press, Thailand, 135 p. Keijzer, F.G. 1945. Outline of the Geology of the Eastern Part of Province of Oriente, Cuba. Ph.D. dissertation at Utrecht University, Utrecht, 234 pp. Magaz, A., et al. 1991. El complejo de formas del relieve gravitacional en la franja costera Baitiquirí- Punta de Maisí provincia de Guantánamo, Cuba. In Morfotectonica de Cuba Oriental, colectivo de autores, Ed. Academia, La Habana 1991. Magaz, A. 1996. Personal communication about ' The Particularity of the Gravitational Morphogenesis in Cuba". Malczewski, J. (1999) GIS and multicriteria decision analysis. New York: John Wiley & Sons, Inc., 390 pp. Martin, J.G. and Dieter, S.B. 1999 Atmospheric trigegring and geomorphic significance of fluvial events in high-latitude regions in Magnitude and frequency in geomorphology: proceedings of the 4th int. conf. on geomorphology, Bologna 1997 Vol. II/ ed. by M. Crozier and R. Mausbacher, Gebruder Borntreager, 1999, p. 87-111. Meijerink, A.M.J. 1988. Data Acquisition and data capture through terrain mapping units. ITC Journal, ITC, Netherlands, p. 23-44. Millán, G. and M. Somin, 1985. Contribución al conocimiento geológico de las metamorfitas del Escambray y Purial: Reportes de Investigación (2): 1-74, Academia de Ciencias de Cuba. (Published in 1987). Munich Re. 1999. Munich Re Group. Annual Report 1998. http://www.munichre.com/ Nagy, E., K. Brezsnyánszky, A. Brito, D. Coutín, F. Formell, G.L. Franco, P. Gyarmaty, P. Jakus y Gy. Radócz. 1976. Texto explicativo del mapa geológico de la provincia de Oriente a escala 1:250 000, levantado y confeccionado por la Brigada Cubano-Húngara entre 1972 y 1976. Archivo IGP, La Habana. Nagy, E., K. Brezsnyánszky, A. Brito, D. Coutín, F. Formell, G.L. Franco, P. Gyarmaty, P. Jakus y Gy. Radócz. 1983. Contribución a la geología de Cuba Oriental, Edtiorial Científico-Técnica, La Habana, 273 p. National Research Council. 1999. Board on natural Disasters, Commission on geosciences, environment and resources. Reducing disaster losses through better information. National Academy Press, Washington, D.C. 25 pp. Nuñez, A., et al. 1981. Informe geologico sobre los trabajos de levantamiento, busqueda a escala 1:100 000 y los resultados de los trabajos busqueda a escala 1.50 000 Y 1:25 000 ejecutados en la parte este de la provincia de Guantánamo, ONRM. MINBAS, La Habana. INTERNATIONAL INSTITUTE FOR AEROSPACE SURVEY AND EARTH SCIENCES

107

CONCLUTIONS AND RECOMMENDATIONS

Open GIS Consortium. 1998. Disaster Management and Public Safety Special Interest Group. URL: http://www.opengis.org/disaster/index.htm. Organization of American States (OAS). 1991. Primer on natural hazard management in integrated regional development planning. Whashington, D.C. 480 pp. Prague, C.N. and M. R. Irwin. 1997 Access 97 bible. Foster City: IDG Books, 1119 pp. Roshannejad, A. Asghar. 1996. The management of spatio-temporal data in a national geographic information system. PhD. Thesis University of Twente; ITC 1996, 188 pp. Saaty, T.L. 1996. The Analytic Hierarchy Process, New York, McGraw Hill, 1980, reprinted by RWS Publications, Pittsburgh. Somin, M. and G. Millán. 1972. The metamorphic complexes of Pinos, Oriente, Escambray and Oriente in Cuba and its ages: Izvestia Akad Nauk, SSSR, Geology 5:48-57. Somin, M. and G. Millán. 1981. Geología de los complejos metamorficos de Cuba (in Russian): Edit. Nauka, 219 p. Moscow. Turner, A., K. and Schuster R.,L. (Ed)1996. Landslides: Investigation and Mitigation. Special report 247, 671 p. Universal Systems Limited. 1998. CARIS GIS: User' s Guide. Universal System Ltd. Fredericton. 276 p. Van Westen, C. J. 1993. Application of geographic information systems to landslide hazard zonation. ITC publication number 15, ITC, Enschede 245 p. Varnes, D. J. 1958. Landslides types and processes. In Landslides and Engineering Practice. Eckel, E.B., (Ed.), HRB, Special Report 29, pp. 20-47. Varnes, D. J. 1978. Slope Movement types and processes. In Landslides: Analysis and Control. Schuster, R. L. and Krizek, R. J. (Ed.), Special Report 176, Transportation Research Board, National Academy of Sciences. Whashington, D.C., pp.11-33. Varnes, D. J. 1984. Landslides hazard zonation: a review of principles and practice, Comision on landslides of the IAEG, UNESCO, Natural Hazards No. 3, 61 p. Voogd, H. 1983. Multicriteria evaluation for urban and regional planning. London: Pion. 367 pp.

108
