• No results found

Pixel Based and Object Oriented Multi-spectral Remotely Sensed Data Analysis for Flood Risk Assessment and Vulnerability Mapping.

N/A
N/A
Protected

Academic year: 2021

Share "Pixel Based and Object Oriented Multi-spectral Remotely Sensed Data Analysis for Flood Risk Assessment and Vulnerability Mapping."

Copied!
162
0
0

Loading.... (view fulltext now)

Full text

(1)

Institutionen för datavetenskap

Department of Computer and Information Science

Final thesis

Pixel Based and Object Oriented Multi-spectral

Remotely Sensed Data Analysis for Flood Risk

Assessment and Vulnerability Mapping

By

Waqar Ul Hassan

Email: {waqch723}@student.liu.se

LIU-IDA

/FFK-UP-A--11/001--SE

(2)

Pixel Based and Object Oriented Multi-spectral Remotely

Sensed Data Analysis for Flood Risk Assessment and

Vulnerability Mapping.

By

WAQAR UL HASSAN

Student No. 780826-3292

LIU-IDA/FFK-UP-A--11/001--SE

The Project Submitted to the Department of Computer and Information Sciences, Linkoping University Sweden for the Partial Fulfilment of the Requirement for the Degree of Master of Science in Geoinformatics

Department of Computer and Information Sciences Linkoping University, Sweden

(3)

Pixel Based and Object Oriented Multi-spectral Remotely

Sensed Data Analysis for Flood Risk Assessment and

Vulnerability Mapping.

By

WAQAR UL HASSAN

Supervisor: Dr Åke Sivertun

(4)

1.0. INTRODUCTION 13

1.1. Background of Study 13

Indus Basin and Natural Hazard 15

Flooding as Natural Hazard 16

Definition of Natural Hazard 16

Type of Floods in Lower Indus Basin 17

History of Worst Natural Disaster in Pakistan 17

Natural Hazard Assessment 19

Indus Flooding Mechanism 19

1.2. Problem Definition 20

1.3. Motivation 20

1.4. Aim of Study 22

1.5. Conceptual and Methodological Framework 23

1.6. Data Required for this Research 24

1.7. General Information about Remotely Sensed Data 24

Electromagnetic Energy and Electromagnetic Spectrum 25

Digital Image Characteristics 26

Spectral Information in Digital Image 26

Philosophy of 2-Dimensional and 3-Dimensional Feature Space of Digital Imagery 26 Distance Calculation in 2-Dimensional and 3-Dimensional Feature Space of Digital Imagery 27

Landsat MSS Data 27

Landsat TM and ETM Imagery 28

MODIS Imagery 29

Aster Imagery 29

2.0 INTRODUCTION 31

2.1.1. GIS and Disaster Management 31

2.1.2. Definition and Types of Disasters 31

2.1.3. Data Required For Disaster Management 32 2.1.4. Remote Sensing and GIS for Disaster Management 33

2.1.5. GIS General Management Strategies 34

2.1.6. Disaster Management Process Flow 34

Planning 35

Mitigation 36

Preparedness 36

Response 37

Recovery 37

2.2. Pre Disaster Risk Mapping and Management Model 38 2.3. Role of Remote Sensing and GIS during Planning Phase 39 2.4. Role of Remote Sensing and GIS for Rehabilitation 40 2.5. Basic Principal of Object Oriented and Pixel Based Satellite Image Analysis 41

(5)

Image Extraction and Conversion into Compatible Format 42

Object Based Image Analysis 43

Region Growing Technique 45

Content Based Technique 45

2.5.1. Segmentation 45

2.5.2. Segmentation Algorithm of SPRING 47

2.6. Object Oriented Image Segmentation Evaluation Philosophy 48

2.6.1. Supervised Segmentation Evaluation 49

2.6.2. Unsupervised Segmentation Evaluation 50

2.7. Object Based Data Modelling Levels 50

2.7.1. Conceptual Level 50

2.7.2. Implementation Level 51

2.7.3. Relationship Between Conceptual and Implementation Level 52

2.8. Digital Image Classification 52

2.8.1. Object Oriented Classification 53

Attribute Extraction Methodology in Object Oriented Classification 54

2.8.2. Basic Theory of Pixel Based Image Analysis 55

Pixel Based Supervised and Unsupervised Classification Method 56

Unsupervised Classification 56

Supervised Classification 56

Training Stage 56

Classification Stage 57

Accuracy Assessment Stage 57

3.0. STUDY AREA, DATA AND RESEARCH METHODOLOGY 59

3.1. Indus Flood Forecasting Mechanism 59

3.2.Temperature and Precipitation Data of Pakistan (1931 - 2000) 60

3.3.Location of Study Area 68

3.4.Topographic Map and Elevation Data of Southern Pakistan 70

Geography of Indus River Lower Basin 71

Climate of Indus Lower Basin and Indus Vellay Region 72

Hydrology 73

Agro- Climatic Ecological and Geo-Morphological Zone of Indus 74

Lower Basin of Indus River and Delta’s Shelf Structure 75

Meteorology (Tides and Strom Surges in Indus River Delta) 75

Some Facts Lower Basin of Indus and Indus Delta Degradation 75

3.5. Material and Method 76

Analysis Procedure / Methodology 78

3.6. Basic Data For Research 79

Remotely Sensed Data 79

Maps of Study Area 79

Data Preparation 79

(6)

Geometric Correction of the Data 80

Atmospheric Correction of the Data 81

Overview of the Methodology for this Research 81

General Object Oriented and Pixel Based Analysis Hierarchy 82

4.0. RESULTS OF OBJECT ORIENTED AND PIXEL BASED ANALYSIS 83

4.1. Introduction 83

4.2. Image Classification 84

4.2.1. “Pixel to Pixel Classifier” 84

4.2.2. “Segmented Image Classifier or Region Growing Classifier” 84

4.3. Pixel based Classification 85

MAXVER Classifier (Maximum Likelihood Classifier) 85

MAXVER-ICM (Integrated Conditional Mode) Classifier 85

Euclidian Distance Classifier 86

K-Median Classifier 86

4.4. Region Growing Classification 86

4.4.1. Unsupervised Classification 87

IsoSeg Classifier 87

4.4.2. Supervised Classification 87

Bhattacharya Distance Classifier 87

ClaTax Classifier 88

4.5. Analysis Results of Landsat MSS Data (1977 Imagery) 88

Reults of Principal Component Analysis (Sukkur and Dadu District) 88

Segmentation Results (Sukkur District) 90

Segmentation Results (Dadu District) 90

Region Growing Classification 91

Object Based Classification Results 91

Pixel Based Classification of Landsat MSS Images 92

4.6. Analysis Results of Landsat TM Data (Post Disaster- 1992 Flood) 95

Analysis of Post-Disaster Landsat TM Data 95

Segmentation Results (Sukkur District) 95

Segmentation Results (Dadu District) 96

Results of Principal Component Analysis of TM Data (Post-Disaster) 97

Region Growing Classification (Segmented Image Classification) 98

Pixel Based Classification Results 100

4.7. Analysis Results of Landsat ETM Data (Pre Disaster 2000 Imagery) 103

Reults of Principal Component Analysis (Sukkur and Dadu District) 103

Segmentation Results (Sukkur District) 104

Segmentation Results (Dadu District) 105

Pixel Based Classification Results 106

Region Growing Classification Results 108

Object Based Classification 108

4.8. Analysis Results of Landsat ETM Data (2006 Imagery) 110

Segmentation Results (Sukkur District) 111

Segmentation Results (Dadu District) 112

Object Based Classification Results 112

Pixel Based Classification of Landsat ETM 2006 Imagery 113

4.9. Analysis of Post Disaster Landsat TM and ETM Data (August 2010) 115

Results of Principal Component Analysis (Sukkur and Dadu District) 115

Segmentation 116

Segmentation Results (Sukkur District) 117

Segmentation Results (Dadu District) 117

(7)

Object Oriented Classification of Landsat TM Bands 117

Pixel Based Classification Results 119

4.10. Analysis of ASTER Data 121

ASTER Post Disaster (August 2010) Data Analysis Results 121

Analysis Results of ASTER 122

Results of PC Analysis ASTER (Sukkur District) 122

Principal Components Analysis of Aster Images Result 122

Region Growing Classifier 123

Object Oriented Classification 123

Pixel Based Classification Results 124

4.11. Accuracy Assessment 125

Error Matrix 125

Error Matrix of Region Growing Object Oriented Classification (Dadu District Pre-Disaster) 126 Error Matrix of Region Growing Object Oriented Classification (Dadu District Post Disaster) 129 Error Matrix of Pixel Based Classification (Sukkur District Post Disaster) 133 5.0. COMPARISON OF OBJECT ORIENTED AND PIXEL BASED IMAGE

ANALYSIS METHOD 137

6.0. CONCLUSION 140

6.1. Limitations of the Research 142

6.2. Recommendations and Future Work 143

7.0. REFERENCES: 145

8.0. APPENDIX 158

(8)

List of Figures

FIGURE: 1.1 ELECTROMAGNETIC SPECTRUM SPECIFICATIONS 25

FIGURE 1.2: WAVELENGTH COVERAGE RELATIONSHIP OF ASTER AND LANDSAT IMAGERY CREATED BY DR

ANDREAS KAEAEB DEPARTMENT OF GEOGRAPHY UNIVERSITY OF ZURICH SWITZERLAND 29

FIGURE 2.2: STAGES OF DISASTER 33

FIGURE 2.3: DISASTER MANAGEMENT PROCESS FLOW 35

FIGURE 2.4: GIS AND REMOTE SENSING PRE-DISASTER RISK MAPPING AND MANAGEMENT MODEL 39

FIGURE 2.5: MOLENAAR LINGUISTIC DYNAMIC MANAGEMENT MODEL FOR SPATIAL DATA MODELING IN

GEO-INFORMATICS 41

FIGURE 2.6: CONCEPTUAL LEVEL OF GIS DATA MODEL 51

FIGURE 2.7: OBJECT BASED DATA MODELLING HIERARCHY WITH DIFFERENT DATASETS 52

FIGURE 2.8: DIGITAL IMAGE ATTRIBUTES EXTRACTION METHODOLOGY 55

FIGURE 2.9: STAGES OF SUPERVISED IMAGE CLASSIFICATION 56

FIGURE 3.1: DETAILED MAP OF PAKISTAN. 60

FIGURE 3.2: DETAILED MAPS OF PRECIPITATION RECORDS ( 1931 – 2000 ) 65

FIGURE 3.3: DETAILED MAPS OF TEMPERATURE RECORDS ( 1931 – 2000 ) 67

FIGURE 3.4: MAP OF STUDY AREA. 68

FIGURE 3.5: (A) AREA OF INTEREST - DISTRICT SUKKUR (B) AREA OF INTEREST - DISTRICT DADU 69

FIGURE 3.6: DIGITAL ELEVATION MODEL OF SOUTHERN PAKISTAN 70

FIGURE 3.7: TOPOGRAPHIC MAP WITH1000FT CONTOUR INTERVAL DERIVED FROM ETOPO 2V2 ELEVATION

DATA OF SOUTHERN PAKISTAN 71

FIGURE 3.8: COMPLETE RESEARCH METHODOLOGY PROCESS FLOW DIAGRAM 79

FIGURE 3.9: GENERAL OVERVIEW OF PIXEL BASED AND OBJECT ORIENTED ANALYSIS 82

FIGURE 4.1: PRICIPAL COMPONENT ANALYSIS RESULTS OF STUDY AREA ( DADU AND SUKKUR ) 1977 MSS DATA 90

FIGURE 4.2: SEGMENTATION RESULTS OF LANDSAT MSS 1977 DATA (SUKKUR DISTRICT) 90

FIGURE 4.3: SEGMENTATION RESULTS OF LANDSAT MSS DATA 1977 (DADU DISTRICT) 91

FIGURE 4.4: OBJECT BASED CLASSIFICATION RESULTS OF LANDSAT MSS DATA 1977 92

FIGURE 4.5: PIXEL BASED CLASSIFICATION RESULTS OF LANDSAT MSS DATA 1977 95

FIGURE 4.6: SEGMENTATION RESULTS OF TM 1992 BANDS 1, 2, 3, 4 AND 5.(SUKKUR DISTRICT) 96

FIGURE 4.7: SEGMENTATION RESULTS LANDSAT TM 1992 (DADU DISTRICT) 97

FIGURE 4.8: PRINCIPAL COMPONENT ANALYSIS OF LANDSAT TM 1992 DATA ( SUKKUR AND DADU DISTRICT) 98

FIGURE 4.9: OBJECT ORIENTED CLASSIFICATION RESULTS LANDSAT TM 1992 DATA 99

FIGURE 4.10: PIXEL BASED CLASSIFICATION RESULTS OF LANDSAT TM 1992 DATA 102

FIGURE 4.11: REULTS OF PRINCIPAL COMPONENT ANALYSIS (SUKKUR AND DADU DISTRICT) LANDSAT ETM

2000 DATA 104

FIGURE 4.12: SEGMENTATION RESULTS OF LANDSAT ETM 2000 DATA (SUKKUR DISTT.) 105

(9)

FIGURE 4.14: PIXEL BASED CLASSIFICATION RESULTS OF LANDSAT ETM 2000 DATA. 108

FIGURE 4.15: OBJECT BASED CLASSIFICATION RESULT OF LANDSAT ETM 2000 DATA 109

FIGURE 4.16: PRINCIPAL COMPONENT ANALYSIS RESULTS OF LANDSAT ETM DATA (2006 IMAGERY) 110

FIGURE 4.17: SEGMENTATION RESULTS OF LANDSAT ETM 2006 DATA (SUKKUR DISTT.) 111

FIGURE 4.17: SEGMENTATION RESULTS OF LANDSAT ETM 2006 DATA (DADU DISTT.) 112

FIGURE 4.18: OBJECT BASED CLASSIFICATION RESULTS OF LANDSAT ETM 2006 DATA 113

FIGURE 4.19: PIXEL BASED CLASSIFICATION RESULTS OF LANDSAT ETM 2006 DATA 115

FIGURE 4.20: RESULTS OF PRINCIPAL COMPONENT ANALYSIS (SUKKUR AND DADU DISTRICT) 116

FIGURE 4.21: SEGMENTATION OF LANDSAT TM BANDS 1, 2, 3, 4 AND 5 ( SUKKUR AND DADU DISTT.) 117

FIGURE 4.22: OBJECT BASED CLASSIFICIATION RESULTS OF 2010 IMAGERY. 119

FIGURE 4.23: PIXEL BASED CLASSIFICIATION RESULTS OF 2010 IMAGERY 121

FIGURE 4.24: PRINCIPAL COMPONENT ANALYSIS OF ASTER 2010 DATA 122

FIGURE 4.25: SEGMENTATION OF THE ASTER BANDS 1, 2 AND 3A 123

FIGURE 4.26: OBJECT BASED CLASSIFICATION RESULTS OF ASTER 2010 DATA 124

FIGURE 4.27: PIXEL BASED CLASSIFICATION RESULTS OF ASTER 2010 DATA 125

List of Tables

TABLE 1.1: FLOOD DAMAGES DURING LAST FIVE DECADES FEDERAL FLOOD COMMISSION PAKISTAN.

NATURAL DISASTERS DAMAGES IN PAKISTAN 18

TABLE1.2: NASA LANDSAT SENSOR SPECIFICATION 28

TABLE 1.3: ASTER IMAGES BANDS DETAIL WAVELENGTH, GROUND SAMPLE DISTANCE ETC 30

TABLE 3.1: PRECIPITATION AND TEMPERATURE DATA (1931 – 2000) 64

TABLE 4.1: ERROR MATRIX OF OBJECT BASED CLASSIFICATION BY REGION GROWING METHODS 129

TABLE 4.2: ERROR MATRIX OF OBJECT BASED CLASSIFICATION OF REMOTELY SENSED DATA 133

(10)

Acknowledgment

From the depth of my heart I would like to express my gratitude and praise to almighty Allah the most merciful creator of universe who has enable to complete this study project for the fulfilment of the degree of Master’s of Science in Geoinformatics.

During the period of my study, I received all kind of study and moral support from almost every staff member of Geoinformatics team to whom I am grateful. I would like to express my heartfelt thanks to Associate Professor Åke Sivertun at Department of Computer and Information Sciences, LiU, for the aspiration, inwardness, sincere guidance and tremendous cooperation in all aspects. My thanks will also go to Mr Nasir Zaka and all my sincere classmates for their true friendship and cooperation.

Special Thanks goes to Bob Kimmel officer USGS and all United State Geological survey team members for their guidance and great cooperation. Their help enabled me to complete this research work successfully. I would say thanks from the depth of my heart to program counsellor Mr Par Svenson for his guidance and encouragement during the period of my study.

Last but not least, my thanks go to my family because without their support and encouragement the completion of this research would not have been possible.

Waqar Ul Hassan Linkoping University Linkoping Sweden

(11)

Abbreviation

and

Acronyms

ASTER Advanve Space Borne Thermal Emission and Reflection

Radiometer

DEM Digital Elevation Model

DN Digital Number

ETM Enhanced Thematic Mapper

GIS Geographical Information system

GPS Global Positioning System

GCP Ground Control Points

GLFC Global Land Cover Facility

Grib Grided Binary File Format

INPE National Institute of Space Research Brazil

LIDAR Light Detecting and Ranging

MODIS Moderate Resolution Imagging Spectro-radiometer

MSS Multi-Spectral Scanner

NASA National Aeronatical and Space Administration

Pan Panchromatic Imagery

SPG Spring Grided File Format

TM Thematic Mapper

TM Thematic Mapper

UAV Unmanned Aerial Vechile

(12)

Pixel Based and Object Oriented Multi-Sensors Remotely

Sensed Data Analysis for Flood Risk Assessment and

Vulnerability Mapping.

Abstract:-

Geographical information system with remotely sensed data can be instrumental in many ways for disaster management and post disaster rehabilitation. During last few decades the usage of remotely sensed data has extensively increased, although image interpretation tools are not highly accurate but still considered as fast, reliable and useful way to get information from imagery.

Disaster assessment, management and rehabilitation are always creates challenge for experts. Population growth, expansion in settlements either in the rural or in the urban areas bring more problems not only for the humans but it also affect the global environment Such global changes on the massive scale disturbs the ecological processes.

GIS along with Remote sensing data can change the whole scenario in very short period of time. All the departments concerning to strategic disaster planning process can share their information by using the single platform, so for this purpose spatial database can be helpful by providing the spatial data in digital format to the department concerned. Spatial phenomena can be observed by using different image analysis techniques and the resultant thematic map display the spatial variations and changes that describe the particular phenomenon whether it was any disaster or change in soil type or vegetation type.

Remotely sensed data like aerial, satellite and radar images are very useful for disaster management strategy formulation process. Integration of GIS and remote sensing proved itself the best especially for land-use, land-cover mapping. For this purpose pixel based, sub-pixel based, pre-field and object oriented classification approach are being in use around the world. But thematic maps created from image analyzed by using object oriented classifiers contain more accuracy than any other technique.

(13)

CHAPTER ONE

Introduction

1.1. Background of Study

Geo-informatics as a subject deals with the data collection, analysis and representation of spatial data with diverse means. Today space images are widely used for information acquisition and rapid sharing of reliable information (Sahin H et al, 2006)1

In 21st century growing population in the urban areas considered as a major threat to environment. Although human settlement only takes a very small part of the earth’s surface but population growth, expansion in settlements either in the rural or in the urban areas bring more problems - not only for humans - but it also affect the global environment. Such global changes on the massive scale disturb the ecological processes on local and global scale that causes flood (Zhang Hao et al, 2008)2.

As widely known various typological, meteorological, climatological, biological and hydrological factors are associated with flood but researchers have shown that land use and land cover changes initiated by the human beings actually disturbs the hydrological processes on this earth and such disturbance gradually increases the flood risks (Zhang Hao et al, 2008). These natural disasters are some time devastating to property and human lives so therefore detection or forecasting indeed play a vital role especially to minimize the material and human loss but in case of any sudden calamity, the disaster damage detection for the rescue has great importance.

For disaster detection purpose time is one of the most important issues along with spatial data because map timeliness acts as important precision parameter (Trianni G, Acqua F.D, Gamba P, 2004)3 Geographical information systems along with remotely sensed imagery can be instrumental in many ways for disaster management and post disaster rehabilitation. During last few decades’ usage of remotely sensed data has

1 Sahin. H, Tophan. H, Karakis. S, Marangoz A.M (2006), Comparision of object oriented image analysis

and manual digitizing for feature extraction; ZKU, Engineering Faculty, 67100 Zonguldak Turkey

2 Zhang Hao, Ma Wei-Chun, Wang Xiang-Rong (2008), Rapid urbanization and implication for flood

risk management in hinterland of the pearl river delta china: The Foshan Study : Sensors Publisher ISSN 1424-8220

3Trianni G, Acqua F.D, Gamba P, (2004) Damage detection at different scales from SAR and QuickBird

(14)

extensively increased, although image interpretation tools are not highly accurate the method is still considered as a fast, reliable and useful way to get information. Disaster assessment, management and rehabilitation are always a challenge for experts.

Several factors of uncertainty are associated with accuracy of human flood prediction capabilities but generally rainfall is considered as the basic factor for this type of disasters (Kavetski et al, 2006, Krystofowic, 1999, 2001, Harris Amanada et al, 2007)4. Syed et al (2004) further added that about 70% - 80 % of the variation in terrestrial hydrological cycle are directly linked with rainfall and this source of disasters considered as most frequent global problem that affect people more than any other type of disasters (World Disaster report, 2003; Harris Amanada et al, 2007).

Observation of different earth’s objects without having any physical contact with them is known as remote sensing. In the 21st century remote sensing has been made accessible for human beings to understand, forecast, manage, and respond on any worst climatic condition or on any natural hazard. Advance earth observation sensor technology helps us to make decisions and manage the planet’s resources to overcome several crises situations.

Remote sensing technology provides us cost effective, accurate and quick data collection opportunity by covering wide areas for study. This is especially through if we compare Remote Sensing with traditional field survey methods (Kokalj Ziga, Kristof Ositer, 2007).

Data obtained from different earth observation sensors are often used for various purposes or simply we can say that it has tremendous applications that include terrestrial, atmospheric, geological, as well as oceanographic applications.

Remotely sensed data is being widely used for change detection, land use, land cover mapping, environment modelling and earth’s resources management. Remote sensing techniques and methods proved efficient for all these above mentioned purposes (Downie et al, 1999, El-Raey, 1996, Kushwaha, S.P.S, 1996 et al).

Spatial phenomena can be observed and analyzed by using different image processing and GIS techniques and the resultant thematic map display the spatial variations and changes that describe the particular phenomenon whether it was any disaster or change in soil type or vegetation type. In hydrological perceptive different studies shows that human activity or any natural event profound great impact on the global hydrological

4 Harris Amanda, Rahman. S, Houssain. F, Yarborough. L, Bagtoglou A.C, Easson G (2007), Satellite

(15)

cycle either by increase in the volume or rate of surface runoff, which mount flood risk, soil erosion and degradation of water quality (Dewan Ashraf M, Yamaguchi Y, 2006). By keeping this all in view the main objective of this research is to quantify the risks related to River Indus by adopting the most modern and sophisticated techniques. River Indus flows from Northern to Southern part of Pakistan and covers a large area, so for the researchers of developing country Pakistan it’s difficult to monitor and access the whole land-use area. Therefore such lack of accessibility make remote sensing the only practical option for monitoring any sudden change or inundation that occur due to any atmospheric phenomena at the river basin scale (Hess et al, 1995, Barbosa Claudio et al, 2004). Accuracy and reliability of the thematic maps always depends on the analysis of remotely sensed data, it also includes the type or technique adopted by the analyst for the analysis of satellite images or remotely sensed data.

Image analysis is indeed an interesting task but selection of appropriate technique for analysis always challenging for any researcher because it depends on the various factors that include nature of the problem, quality of remotely sensed data and availability of the resources. Various image analysis techniques are being in use these days for this purpose, it includes the traditional analysis techniques as well as the most modern. For remote sensed data a traditional and more common technique that are being in use these days known as pixel based image classification technique. Pixel based image classification based on the labelling of pixels of any image to obtain the meaningful information from it. But the most modern object based image analysis technique produced better results because unlike pixel based approach it not only analyzes pixels spectral properties but also spatial properties of them Therefore many researchers clearly indicated the object based approach better than pixel based analysis.

Indus Basin and Natural Hazard

Natural disasters and hazards occur suddenly and can be observed around the world. Even the technologically most advance countries are facing such natural calamities with certain interval of time. There are lots of factors that imitate such disasters or we can simply say that there are various things that act as catalyst and being helpful for such events.

(16)

Researchers normally consider physical environment as vulnerable factor for such disasters like typhoons, landslides, volcanic eruption, storms surges, droughts, floods, earthquakes, tornados, tsunami, Rita, Katrina storm etc (Otieno Adhaimbo J, 2004)5 Floods, typhoons and earthquakes are considered as most frequent and most fatal events on this planet earth because the frequency and magnitude of these events directly impact on society and economy of any country (Otieno Adhaimbo J, 2004)

One of the major set back that observed after natural disaster are their impact on earth surface because natural calamities like flood, earthquake and typhoons seriously disrupt the agriculture based economy(Otieno Adhaimbo J, 2004, Brown et al, 1991).

To get understanding about these natural disasters we should first study about the nature of these calamities first.

Flooding as Natural Hazard

Floods are considered as the most disastrous activity on the earth and major contributor of human property and lives losses. Normally any natural disaster is a sudden phenomenon but with the advance telemetry system now it’s possible to receive an early warning but in many developing countries still such systems are not functional so there natural hazard like flood strikes without any early warning. In developing countries economic constraints and population growth compel the poor people to build their houses near rivers and flooding diminishes everything. During 21st century researcher observed increase in the flooding activity on many parts of the world and to counter this problem there is need of effective mapping, data preparation and analysis the threatened zones.

Definition of Natural Hazard

Normally hazard is taken as the occurrence of the potential threat whether natural or man made to any specific area within specific time period (Alexender, 1993, Otieno Adhiambo.J, 2004). Explicitly we categorize hazards on the basis of their nature

5 Otieno Adhaimbo J( 2004), Scenario study for flood hazard assessment in the lower bicol floodplain

Philippine using A 2D flood model; Based on the 1988 flood event caused by typhoon yonning; International Institute of Geo-Information Sciences and Earth Observation Enschede, The Netherlands.

(17)

because every hazard can be based either on geophysical, hydro-meteorological or technological mishaps, human alteration to natural system are the most phenomena (Otieno Adhiambo.J, 2004)6.

Type of Floods in Lower Indus Basin

As one of Asia´s largest river basin the river Indus comprises of about 1 million Km2 area from which 56% lies in Pakistan and constitute about 70% of the country area (Rehman H, Kamal A, 2010)7 . There are many causes of flood observed in the river Indus normally Regional rainfall (Monsoon) and northern snowmelt results into rise in the water level of Indus River. Some major tributaries of the Indus River are Ravi, Chenab, Jehlam and Kabil rivers greatly influences the water level therefore the magnitude of the flood always high towards the downstream or the lower basin area from the headwater especially during the monsoon rainfall and snow melting seasons8. The recent 2010 flooding episode was the result of persistent seasonal (monsoon) rainfall (Flood Report 2010). Monsoonal season in Pakistan normally starts in the month of June and ends in the September and during this period persistent rain observed in the northern areas of Pakistan including Indian held Kashmir (Flood Report 2010) that leads to the flooding.

History of Worst Natural Disaster in Pakistan

River Indus has a long history of flooding since 1929 (Flood Report 2010) after independence during the last six decades tremendous life and material loss particularly to infrastructure, agriculture, lives, rural and urban settlements has been observed by the Pakistani nation (Tvedt . T, Jakobsson E, 2006). From independence 1947 Pakistan has been faced worst natural disasters particularly floods these disasters not only shattered the economy of the country but also cause damage to existing infra-structure (Kamal A, Rehman H, 2005).

6 Otieno Adhiambo J (2004), Scenario study for flood Hazard Assessment in the lower Bicol Floodplain

Philippine using A 2D Flood Model, Based on 1998 flood event caused by typhoon Yonning; International Institute for Geo-Information Science and Earth Observation Netherlands.

7 Rehman H, Kamal A (2005), Indus Basin River System – Flooding and Flood Mitigation; Cardno

Lawson treloar Pty Ltd Austrila

(18)

Disaster Type Date Death Toll Total Number of Affected Material Loss (USD Millions) Earthquake May 31,1935 60,000 - - Earthquake / Tsunami November 27, 1945 4,000 - - Flood 1950 2,910 - 9.08 (Billion Rs) Flood 1955 679 7.04 (Billion Rupees/Rp) Flood 1956 160 5.92 (Billion Rs)

Wind Storm December 15, 1965 10,000 - - Flood August, 1973 474 4,800,000 662 Earthquake December 28, 1974 5,300 - - Flood 1975 126 12.72 (Billion Rs) Flood August 2, 1976 425 5,566,000 505 Flood June, 1977 10,354 1,022,000 - Flood July, 1978 393 2,246,000 - Flood 1988 508 15.96 (Billion Rs) Flood August 9, 1992 - 12,324,024 1,000 Flood September 1992 1,334 6,186,418 - Flood July 22, 1995 - 1,255,000 - Flood August 24, 1996 - 1,186,131 - Flood March 3, 1998 1,000 - - Drought Late 1999-March 2000 - 2,200,000 247 Flood 2001 - - 246 Earthquake October 8, 2005 78,000 - 5,200 Flood 2005 - - 327 Wind Storm 2007 - - 1,620 Flood 2008 - - 103

Flood July, August

2010 1,645 17,600,000 15,000 – 43,000

Table 1.1: Flood Damages during last five decades Federal Flood Commission Pakistan9. Natural Disasters Damages in Pakistan10

9 Tvedt . T, Jakobsson E (2006)A History of Water Vol 1; Water Control and River Biographies Palgrave

Macmillan Publishers

(19)

The worst flood in the river Indus observed in September 1992 due to intense monsoonal rainfall, medium flooding observed in 2005, 2007 and 2007 as well.

Natural Hazard Assessment

Identification of potential threat is considered as the most crucial stage of hazard management because rehabilitation often based on the characteristics of the hazard that require real time data and sufficient information. And the mitigation process actually comprises of the strategy formulation process on the basis of the available resources for the reduction of the existing and futuristic vulnerability of any natural disaster to the human settlement.

Mitigation measures can either be structural or non structural in nature but for both type of measures, proper planning, capitalization and deployment of resources are required. Mitigation process in fact ensures the safety of the community and material resources on short term as well as on the long term basis. Quick decision making process from the government side always have positive correlation with effective mitigation planning scenario because integration of different entities on public and private level for assurance of safe and rapid development.

Indus Flooding Mechanism

Monsoonal season often brought floods in South Asian countries particularly in Pakistan (Rehman H, Kamal A, 2005)11. In the river Indus flooding observed normally in during late summer due to raise in the water level of its tributaries those are also the major rivers of the Pakistan like Chenab, Jhelam, Satluag (Kamal A, Rehman H, 2006). Seasonal climate change particularly affects of scrotching summer heat Monsoonal Intense rainfall snows melt somehow turns into the worst flooding in the rivers. (Kamal A, Rehman H, 2006). These are the two main reasons of flooding observed by the researchers 1: the monsoonal depression 2: glaciers snowmelt (Werner Micha, Dijk Van Marc, 2005). Flood initiated from upstream therefore all the flood warning issues on the basis of data collected from upstream flood warning system, Although a well sophisticated telemetric system also installed on the downstream as well (Werner

11 Rehman H, Kamal A (2005), Indus Basin River System – Flooding and Flood Mitigation ; Cardno

(20)

Micha, Dijk Van Marc, 2005). Most of the precipitation, water flow and water level data collected from two radar systems, many gauges and barrages (Guddu Barrage, Sukkur Barrage, Kotri Barrage ) build by the government on the river Indus both at upstream and downstream stream (Werner Micha, Dijk Van Marc, 2005).

1.2. Problem Definition

Environment is greatly affected by global land use land cover change, Rapid urbanization, pollution, expansion of human settlements contribute to these changes that initiate natural disaster like tsunamis, cyclones, floods, volcanic eruption and tornados. Natural disasters especially flood are often sudden and greatly affects the infrastructure of any place. For natural disaster management remote sensing has great application and analysis of such data provides us real time information. Classical image analysis approach based on labelling of single unit known as “pixel” such technique denoted as pixel based image analysis technique. Logic behind pixel based image analysis actually revolves around the utilization of spectral information of each pixel and to classify it on the basis of homogeneity of the pixels. Whereas the modern applied analysis methods considered more accurate and reliable particularly for the development of error free thematic maps. GIS and remote sensing have great applications in developing country like Pakistan where it’s difficult to gather the data due to inediquate resources.

1.3. Motivation

Many image analysis approaches are being in use these days. In each image analysis approach image classification play an important role or simply we can denote that the accuracy of the analysis always depends on the classification technique used for analysis of remotely sensed data especially for the monitoring any natural hazard or natural hazard risks analysis purpose.

The basic criterion followed during classification is the fact that every earth object contains specific spectral information which may vary from other objects.

On the basis of this unique concept of the properties of earth’s objects images analysis process proceed. In the classification phase on the basis of spectral information theory of earth’s objects followed this ended with pixel by pixel classification.

(21)

Pixel based image analysis approach based on spectral information of the objects that any satellite image contains and this approach has been successfully in practice from last few decades. Different software has been developed for satellite images analysis by adopting this approach. But with the passage of time many objections raised by analysts, like do spectral information of the pixels are capable to provide enough information about any image? Is it possible that in pixel based image classification technique the objects have similar spectral information can be separated from each other?

But later studies showed that pixel based image analysis approach have limited capabilities especially if we compare it with objects based classification. Segment or object based classification lead to improved classification results over pixel based approach (Dorren et al. 2003, Geneletti and Gorte 2003, Gitas et al. 2004, Gao.Y, 2007) The shift towards object orientation is driven by the limitation of the pixels in addressing the issue of location, scale, neighbourhood and distance (Gamanya Ruvimbo et al, 2007 Strobl & Blacked, 2001).

Various algorithm and analysis techniques are being in use for image analysis purpose. Image classifications are the most important phase of analysis process because during this phase in fact researcher tries to find the fact about the different earth objects having different spatial and spectral information. Classification either traditional or advanced it normally addresses the various characteristics of the different objects within any image. Traditional or classical image classification approach have unique algorithm that classify the image on the basis of spectral properties of single pixel of the image and gives us the final classification result pixel by pixel. Such unique approach is widely used around the globe and known as the pixel based image analysis classification. There are lots of limitations and drawbacks by the practical implementation of this classical image classification technique as we know that every earth’s object have specific characteristics or physical properties and each object contain specific spectral information that differs from each other as described earlier But in some cases some objects shows similar spectral characteristics that creates problems during classification phase. In general pixel based image analysis technique often fails to isolate those objects from each other who have same spectral properties during classification phase. Therefore researchers often put the question mark on the accuracy of this traditional classification technique particularly when the different objects with similar spectral

(22)

information assigned same class during classification process and after such classification a thematic map that we obtain sometime contain many errors in it.

Such limited capability of pixel based image analysis compels the GIS and Remote sensing experts to think in new dimension and to work on new analysis algorithm or techniques.

The new and state of art technique that developed by the researchers is unique in many ways because this image classification approach address both spatial and spectral information that object’s of any image contain. Therefore classification of the image entirely based on both the spectral and spatial properties of earth’s objects instead of only spectral as we observe in the Pixel based approach. This newly developed and bit sophisticated technique is known as Object based Image classification approach.

Object oriented image classification technique has many advantages over the traditional pixel based technique the most prominent advantage that it have on traditional technique is the analysis algorithm the “fuzzy theory”. Therefore as we observe in the pixel based analysis technique that each pixel assigned specific value it may be either 1 or 0, unlike pixel based technique in object oriented approach each object assigned the specific value that based on the fuzzy logic. Such advanced technique helps us to understand the various things including objects properties, class hierarchy and object class relationship.

1.4. Aim of Study

The aim of this research is to analyze the remotely sensed data for vulnerability of flood hazards in the lower basin of River Indus during last four decades. Although Pakistan is a developing country and has been facing tremendous economic crises but still each year Pakistani government bear the loss of millions dollars property and lives from natural calamities like flood, earthquakes, wind storm and forest fire. It is the matter of fact that developed countries are well equipped with all kind of resources and detailed land use, land cover spatial data but still under developed and developing countries like Pakistan lacks such sophisticated technologies. Space borne remotely sensed data provided by the USGS and University of Maryland’s online land-cover \ land-use web

(23)

portal are useful for both either developed or developing countries to counter any natural disaster (Dewan Ashraf M, Yamaguchi Y, 2006)12.

Accurate and updated spatial data is essential for planning, mitigation, prepardness and to understand the impact of different phenomenon on terrestrial ecosystem (Dewan Ashraf M, Yamaguchi Y, 2006). For the analysis purpose a unique approach is adopted by using the most sophisticated and modern tools. Objects oriented data modelling or object oriented image analysis approach has many advantages over the traditional pixel based approach like accuracy level of object oriented analysis is better as compared to pixel based analysis.

The main study objectives that I have planed to achieve through this technique are: 1: To explore the information from different data sets including Digital maps, statistical data and satellite images.

2: To analyze the results of the by combining different bands to extract spectral and spatial information.

3: To find out the classification accuracy by using most modern image analysis approach.

4: To explore the importance and effectiveness of the segmentation process in the object oriented image analysis.

5: To study context and textural information of different image objects by using different classifiers.

6: To explore the analysis results of different types of image classifications 7: Compare the results of object based and pixel based image analysis

8: To develop awareness with modern thematic mapping techniques by using remotely sensed data.

9: To learn and practically implement these analysis techniques for disaster management

1.5. Conceptual and Methodological Framework

Methodological and conceptual framework is entirely based on the systematic and scientific approach of investigation or analysis of the data obtained by different sources. As we know that scientific investigations are purely based on some facts and figures

12 Dewan Ashraf M, Yamaguchi Y (2006), Remote Sensing and GIS for Mapping and Monitoring the

Effect of Land use\Cover Change on Flooding in Greater Dhaka of Bangladesh. Department of earth and Environmental Science Nagoya University, Japan

(24)

obtained from experimentation. Coherent methodological approach along with systematic study tracks us for accurate scientific investigation of the given problem.

1.6. Data Required for this Research

Most important thing required for any research is the data and for geographical research real time data that depict the whole situation is the most significant thing. Results of any research always depend on the useful data products and how we can use the spatial data accurately and a point to be noted that study result directly or indirectly related with accuracy of the data more specifically the source from where data obtained. Vulnerability analysis of natural hazard was done with the help of imagery obtained from different satellite sensors. For the purpose of data collection it was preferred to contact and request to USGS and NASA.

As classification of satellite imagery produced best results shown in the different studies conducted by the researchers in the past and it proved to be the best alternative to the field observations of the bigger areas because it enable the researchers to classify the data up to ten or more land, cover land use classes in short period of time for spatial analysis purpose (Kokalj Ziga, Ostir Kristof, 2007).

Different types of maps and statistical information collected for the research purpose from different published journals either online journal or from printed material. Broad and through discussion about the spatial data and other related information are mentioned in the different sections of this research report.

1.7. General Information about Remotely Sensed Data

For the study purpose different type satellite imagery was obtained directly from USGS (United State Geological Survey) Natural disaster’s data portal and Land-cover \Land-use data portal of university of Maryland’s. For this purpose written permission were obtained from USGS country manager for download and usage of the imagery. During the data acquisition special emphasis were given to obtain the latest imagery. For the study purpose cooperation of USGS officials were really admirable.

Remote sensing data have been widely used and received immense attention around the world due to its importance in the global change detection analysis because as we know

(25)

that land cover changes either human induced or natural greatly influence the global environment (Dewan Ashraf M, Yamaguchi Y, 2006).

Sensor either installed on the airborne or on space borne platform it provides us the data and the information related to earth resources by the process of emission, reflection and reception of electromagnetic energy that data later on used for analysis, inventorying and mapping purpose (Lillesand, 2001, Yan Gao, 2003)13

Electromagnetic Energy and Electromagnetic Spectrum

Remote sensing are totally depend on the energy either from sun or other active devices those emit energy in form of electromagnetic waves.

For the remote sensing electromagnetic energy serve as the backbone because in remote sensing sensor acquire information about any object that emit or reflect the electromagnetic energy from the surface of the earth, such information or data later on analyzed.

Electromagnetic energy exist in different forms like X-rays, radio waves, ultraviolet, infrared, visible light etc. Emission of the electromagnetic energy follows the basic wave theory Principal those are

C= v × λ

For the description of the characteristics of the electromagnetic energy we take help from electromagnetic spectrum in which different forms of the energy with their wavelength and location explicitly explained.

Figure: 1.1 Electromagnetic Spectrum specifications 14

13 Yan Gao (2003), Pixel based and object oriented image analysis for coal fire research; ITC Netherlands 14 Remote Sensing a Tool of Discovery; Application of Remote Sensing to Biological and Environmental

(26)

Sensors used for the remote sensing purpose records the information about earth objects containing distinctive features spectral characteristics of emission and reflectance of electromagnetic energy and these features can be study separately on the basis of their spectral response pattern in the electromagnetic spectrum.

Digital Image Characteristics

Due to tremendous applications of remote sensing technology the satellite data especially climatic data product got recognition globally and these real time satellite data products now fulfil the needs of researcher around the globe (Ebert et al, 2007, Harris Amanda, 2007).

Spectral Information in Digital Image

As discussed previously in this project project that digital image obtained through any remotely sensed media, whether satellite scanners, airborne Lidar, Radar or UAV born cameras, are composed of pixels. The basic properties of a pixel may vary from other pixels of the same image in many ways on the basis of their brightness, radiances that depict the image objects information. So each pixel has specific digital number (DN) that correspond these properties of that pixel which may numerically range from 0 to 255, 0 to 511 or higher depending upon the pixel information of the given data, these numerical ranges actually represent the set of the integers which provide the ease to the researcher to analyze the image by using computer (Yan Gao, 2003, Lillesand, 2001). Digital image contain the information stored after recording the reflected or emitted energy from the object present on the earth and normally it is 2-dimensional array (Yan Gao, 2003, Janssen, 2001).

Philosophy of 2-Dimensional and 3-Dimensional Feature Space of Digital Imagery

In this study analysis done by using the multispectral images either by using two bands or three bands combination to get the better analysis results. In the remote sensing value

(27)

of two bands are often categorize as the two dimensional vector the feature vector and researchers denotes the graphical representation of either two or three bands feature vector as feature space (Janssen, 2001; Gao Yan, 2004).

Distance Calculation in 2-Dimensional and 3-Dimensional Feature Space of Digital Imagery

Feature space distance in the digital imagery is denoted as “Euclidian Distance” and DN is the measuring unit for it, Interesting thing to note that for the distance calculation in two, three or more dimensional feature space normally Pythagorean theorem are being in use and for all calculation done in same way (Janssen, 2001, Gao Yan, 2004).

Landsat MSS Data

Landsat MSS data is considered as one of the most sophisticated, detailed and the oldest domain of remotely sensed data products but with the passage of time new scientific sensing devices that are placed either in space or in mounted on any airborne system, these electronic systems received lot of attention that includes passive multi-spectral sensors, thermal sensors, active radars and laser altimeter systems, these devices are unique in many ways and provides us the required remotely sensed data, Among all these data capturing devices and data still available landsat MSS imagery truly provide us the best results (Kvamme, 2005, Kokalj Ziga, Ostir Kristof, 2007)15

Remotely sensed data particularly landsat MSS imagery is useful for the analysis of various hydrological phenomenon’s particularly estimation of water recharge potential by determination of lineaments, water drainage frequency density, litho-logic character and land cover land use analysis (Shaban et al, 2006, Kokalj Zigo, Ostir Kristof, 2007). For the analysis purpose different region growing image classifier has been used. Detail related to the different steps of image classification is provided in the chapters of image analysis and results. Traditional pixel based image analysis approach is based on the conventional statistical algorithm and has been used for the decades (Mather, Tso, 2001).

15 Kokalja Ziga, Ostir Kirstof (2007), Land cover mapping using landsat satellite image classification in

the classical Karst – Kras Region ; Institute of Anthroplogical and spatial Studies ZRC SAZU, Novi trg 2, Si- 1000 Ljubljana, Slovenia, Acta Carsologica Publishers.

(28)

In the Object oriented image analysis for the classification purpose of spatial data like maps, images and other type of remotely sensed data classical combination of the spectral and spatial information are being adopted.

Landsat TM and ETM Imagery

As discussed earlier that both oldest and Latest Landsat TM and ETM imagery obtained from web based United States Geological Survey (USGS) Disaster Management Imagery Portal. The data acquisition was done on August 2010 of the Indus flood episodes. All kind of Pre and Post disasters Landsat TM satellite images of the flood affected area obtained with written permission from USGS and university of Maryland landcover/landuse data portal.

Let’s take a look of the various prominent properties of both ETM and TM data. One of the unique features that exist in the landsat imagery is the high correlation of adjacent spectral bands (Li Guiying, Weng Qihao, 2005)16.

Table1.2: NASA landsat Sensor Specification17

16Li Guiying, Weng Qihao (2005)Using Landsat ETM+ Imagery to Measure Population Density in

Indianapolis , Indiana , USA Photogrammetric Engineering and Remote Sensing Vol 71 No. 8 August 2005 pp. 947-958

17 NASA Landsat Sensor Specification

(29)

MODIS Imagery

Moderate Resolution Imaging Spectroradiometer (MODIS) is a unique satellite system by NASA’s onboard Terra and Aqua spacecraft, primarily used for vegetation analysis or monitoring of agriculture systems. Because of the effectiveness of MODIS system’s seven spectral bands are proved by the researchers particularly in land use mapping of bigger areas (Lobell and Asner, 2004, Hong. G et al, 2006)18 But with the advantages of the MODIS data there are few disadvantages. These disadvantages or problems that faced by the researchers during the analysis of MODIS data are mainly mixed pixel issue and number of samples issue therefore accuracy of crop identification analysis particularly for land use \ landcover mapping are greatly affected and lead to false judgment (Hong G et al, 2006)

Aster Imagery

Aster (Advance Space borne Thermal Emission and Reflection Radiometer) launched by NASA in 1999 flown in the Terra satellite acquire simultaneously data of three different resolution 15m 30 m and 90m respectively (Yale Center EO, 2010).

Figure 1.2: Wavelength coverage relationship of ASTER and Landsat Imagery created by Dr Andreas Kaeaeb Department of Geography University

of Zurich Switzerland19

18 Fusion of MODIS and RADARSAT data for corp type classification – An initial study ; ISPRS

workshop on updating Geo-spatial databases with imagery and the 5th ISPRS workshop on DMGISs

19 Selby Richard (2006), Creating Digital Elevation Models and Ortho-images from ASTER Imagery,

(30)

Table 1.3: Aster Images bands detail wavelength, ground sample distance etc20

20 Kilby W.E, Kilby C.E (2006)Examining ASTER Imagery with MapPlace Image Analysis Toobox ; A

tutorial Manual ; Cal Data Ltd, Ministry of energy mines nd petroleum; Geosciences BC Report 2006-3 British Columbia

(31)

CHAPTER TWO

2.0. Introduction

2.1.1. GIS and Disaster Management

Natural disasters are considered as the burden on economic development as well as in the lives of victims of any country (Yu chaoqing et al, 2009, World Bank, 2006)21 . GIS with remote sensing provides us the best solution to deal with any emergency situation. To understand the application of remote sensing with GIS for disaster management, let’s take a look of steps and tools required for disaster management.

2.1.2. Definition and Types of Disasters

Earthquakes, hurricanes, storms, landslides, floods are all categorized as the natural disaster. The definition of the natural disaster given by the ESRI is; “Natural disasters are all those unplanned events, activities or phenomena’s that occurs due to any change in the natural system or these occurs as the result of any disturbance the natural processes” (Johnson Russ ESRI, 2000).

21 Yu Chaoqing, MacEachren A.M, Peuquet J Donna, Yarnal Brent (2009), Integrating scientific

modelling and supporting dynamic hazard management with a GeoAgent -based representation of human-environment interactions: A drought example in Central Pennsylvania USA; Elsevier Ltd

(32)

Figure 2.1: Types of Disasters

And after any disaster the process of disaster management comprises of different activities that engage different departments to cope the existing problem at different levels either federal, state, county and local level.

2.1.3. Data Required For Disaster Management

For effective management experts require quick and easy access to the data about the disaster affected area, so at this stage GIS along with remote sensing tools enables the experts to formulate any strategy with the help of available spatial and non spatial data sets either satellite images, digital maps or UAV aerial Photos.

(33)

Researchers divide data sets that are used for disaster management into three basic categories those are as under (Belaji et al, 2002, Stoimenov L et al, 2005).

1: Magnitude, Frequency, location and other relevant data about disaster specific Emergency Phenomena.

2: Topography, geology, hydrology, vegetation, landuse, soil and other relevant data or environmental data.

3: Infrastructure, settlement, socio-economic, census and other relevant data of the affected area.

2.1.4. Remote Sensing and GIS for Disaster Management

Along with remote sending geographical information system provides us a platform to manage the information and different tools for analysis of spatial data for effective management. The process of disaster management involves the contribution of various physical entities like emergency technical services, health care services, engineering services etc, all these entities unite together to minimize the effects of any worst situation (Stoimenov L et al, 2005).

Most important task for the researchers is to take initiative for the disaster management. Remote sensing with geographical information system is greatly helpful to cope with any natural, technological or man made calamity. First let’s take a look of different disaster stages:

Figure 2.2: Stages of Disaster 22

Maps developed by GIS experts have edge over the traditional maps in many ways particularly for disaster management. The GIS hazards maps can be used as the most effective tool for analysis, risk assessment, strategy formulation and visualization of

22 Shaluf Ibrahim Mohammd (2007), An overview on the Technological Disasters; Emerald Group

(34)

disaster vulnerability (Tran Phong Et al, 2008) Hazard mapping considered as the first step towards the development of vulnerability inventory (Tran Phong et al, 2008;Noson, 2002; wisner et al, 2004)

Researchers takes maps as an effective instrument that gives motivation for the process of disaster management because as a visualization tool it provide a thought provoking platform that can never be attained by using the ordinary tools (Tran Phong, 2008; Pardan, 2004)

Modelling, mapping and data analysis contribute to the disaster management in many ways lets take a look of general management strategies and their relationships with GIS.

2.1.5. GIS General Management Strategies

GIS with remote sensing play an important role to identify the disaster vulnerable area. As a general management strategies GIS and remotely sensed data can be used for exact identification of the disaster affected area and distribute or divide the area into different zones on the basis of obtained information related to the pre and post disaster phenomena. Such analysis provides the opportunity to relief worker to prioritize their tasks and start the rehabilitation process. On the basis of remotely sensed GIS data about any natural hazard generally affected areas could be divided into three categories; Less Hazard Zone

Moderate and High Hazard Zone Very High Hazard Zone

GIS strategies revolve around the resources management during the rescue, recovery and rehabilitation in these affected area marked on the basis of damage intensity.

2.1.6. Disaster Management Process Flow

Geospatial data are helpful particularly for effective decision making process for the management of any disaster because planning process often entirely based on the analysis of spatial data (Stoimenov L et al, 2005)23.

23 Stoimenov Leonid, Predic Bratislav, Mihajlovic Valden, Stankovic Miomir (2005)GIS Interoperability

Platform for Emergency Management in Local Community Environment; Poppen .F Painho. M (Eds), Proceedings of AGILE 2005.

(35)

As an efficient tool GIS provide opportunity to the researchers for spatial analysis by displaying various risk factors related to any disaster, calamity or epidemic. GIS can help the experts to integrate the different technologies to obtain best analysis results that can help during planning phase (Stoimenov L et al, 2005). Disaster Management process can be divided into different sub-phases or sub process those are as under (ESRI white paper, 1999; Zlatanova & Holweg 2004; stoimenov L et al, 2005):

Planning Mitigation Preparedness Response Recovery

Figure 2.3: Disaster Management Process Flow

Planning

Planning are actually those activities that performed by the managers to counter any disaster by analyzing and documenting the consequences and possibilities of any worst scenario (ESRI- 2000)24. In planning phase potential impacts of any disaster on human

life, material things and ecosystem are studies carefully and formulates the strategies to cope with such loss effectively. The process of planning starts from identification of any potential threat either to human life or to material things. The responsibilities of GIS experts at this stage are to evaluate the intensity or predict the consequences of that disaster (ESRI-2000). As discussed there are many types of disasters any disaster can be either natural, human induced, or of hybrid type. But during planning phase when GIS experts combines the disaster data with map data containing all the attributes or details like street, residential areas, power lines, hospitals etc then it provides an ease to

24 Johnson Russ(2000), GIS Technology for Disaster and Emergency Management; An ESRI white Paper

(36)

managers to formulate any strategy in the available resources. All other phases of disaster management directly or indirectly linked with the planning phase and planning provides the management experts to draw road map for mitigation, preparedness, and response possibilities before or after any calamity.

Mitigation

Mitigation is always a core issue after every disaster and requirement of mitigation determined by the policy makers. Tasks for the mitigation prioritized by the people and GIS experts those involve for disaster management process on the basis of intensity of disaster.

After any disaster GIS experts locate the area and divide it into different zones on the basis of disaster impact. For example in case of flood experts mark the area or zones on the basis of flood intensity and damage done by the flood into less hazardous zone, medium hazardous and high hazardous zone. Same in case of any earthquake in which GIS experts mark the area on the basis of magnitude. With the help of existing database GIS experts links different geographic features like topographic, climatic, vegetation, land use and land cover together to identify the potential hazard zones. During flood disaster the role of GIS is to identify the vulnerable areas of the flood to protect the man and material things. GIS experts also try to find the possible evacuation ways by locate the bridges, under passes or overpasses. Quick identification of the resources for protective actions is the basic mitigation tasks for GIS experts in the mitigation phase.

Preparedness

Like other steps of disaster management mitigation play an important role to cope with the existing problem. This stage comprises of various activities that linked with the preparation process during crises situation. So in this phase GIS provide the solution of different questions that includes the location of the actual disaster vulnerable zone in shortest possible time along with the quick response schedule. GIS also provide the answers relating to the preparation about emergency traffic handling, evacuation routes, and shelters preparations. Quantity of the resources including man and material that can be used for the emergency situation determined in this phase.

As we know that GIS deals with the real-time monitoring of the early warning systems especially during any crises situation, so quick weather forecasts, temperature, humidity reporting are basic services provided by the GIS expert to the disaster management

(37)

units. Such information equips them to tackle those crises situation because during a fire or volcanic eruption, information about wind speed and humidity are so important. So in the preparedness phase relief worker equip themselves with the necessary information required about the location, weather, possible routes, and emergency plans.

Response

Quick response against any emergency situation always worth a lot and GIS as a subject enables the people to respond quickly during any crises for this purpose computer aided system serves as a primary component. On the basis of available data disaster management department can fix the location and plan the route as a response for evacuation and disaster management.

Computer aided system handled by the GIS experts as a primary tool of analysis provide great ease to the disaster response units working in the affected area. GIS and such sophisticated systems are also helpful to fix the location and plan the shortest possible rout for quick response. So in this way GIS can engage the response units to their closet disaster affected areas by assigning the task to the quick response units and managing the whole relief or response activities.

Recovery

Recovery is the last step of the disaster management process it normally starts when the after disaster or emergency situation immediate threat to any human, environment or material thing are over.

Recovery process can be divided into two important phases those are as under (ESRI, 2000):

Short Term Recovery Long Term Recovery

In short term recovery process GIS experts formulate strategy to recover vital things after any disaster. These vital things includes the necessary services and the potential components of the system like recovery of temporary exit ways, water reservoirs, temporary food, electricity services, shelter homes etc (ESRI, 2000).

After any disaster quick relief may provide the ease to the victims as well as restoration of the services provide better opportunities for work in the disaster affected area to the people involve for disaster management.

(38)

After any disaster the assessment of damage done by the disaster are often the most problematic task for the planners and for the accomplishment of this task There’s always GIS experts required for this purpose. GIS experts work for the short term recovery stage with the help of most sophisticated and modern equipments like GPS, laptops and spatial digital data. The works of GIS experts are to identify the intensity of the damage, locate the most, medium or less affected location and facilities. After doing this they prioritize the tasks for recovery and rehabilitation. All these phases of disaster management are interlinked with each other and for disaster management process these phase are instrumental in many ways.

2.2. Pre Disaster Risk Mapping and Management Model

(39)

Figure 2.4: GIS and Remote Sensing Pre-Disaster Risk Mapping and Management Model

2.3. Role of Remote Sensing and GIS during Planning Phase

As discussed above that different kind of physical either material or human resources are required for the disaster management. These material resources include the spatial data obtained from the various sources that can be used for the designing phase of any disaster management program. Because size, scope and effectiveness of any formulated disaster management program based on the collection, organization and presentation of the available spatial data for each strategic planning phases of the program. Many factors like time, authenticity and logical presentation of data for different management stages are involved for disaster management process along with the human and material resources. Therefore numbers of government departments and personals often celebrate with each other by sharing different kind of information, expertise and efforts. GIS and remote sensing play a key role at every stage of the disaster management process right from planning to its implementation phase. The people involve physically in the disaster management and rehabilitation process often require the detail information about the different things related to the disaster affected area like sewerage system network detail, electrical, communicational infrastructure detail, building layouts, evacuation ways detail etc.

So GIS along with Remote sensing data can change the understanding of the whole scenario in a very short period of time. All the departments concerning to strategic disaster planning process can share their information by using the single platform, so for this purpose spatial database can be helpful by providing the spatial data in digital format to the department concerned. Remotely sensed data, satellite imagery, aerial images and radar photos are also useful for disaster management strategy formulation process. Integration of GIS and remote sensing proved itself the best especially for land-use, land-cover mapping and for this purpose pixel based, sub-pixel based , pre-field and object oriented classification approach are being in use around the world (Araya Y.H and Hergarten, 2008)25

Without such GIS based information system, for emergency or disaster relief workers it is indeed a great task to obtain information from different department in shortest

25 Araya Y.H, Hergarten C (2008), A comparison of pixel and object-based land cover classification : A

case study of the Asmara region, Eritrea ; WIT Transactions on the Built Environment, Vol 100, WIT Press

References

Related documents

Measuring the learning outcomes of the artefact and evaluat- ing whether the involved participants understand the concepts of OOP is a decisive aspect of this research. Although

Furthermore an automatic method for deciding biomarker threshold values is proposed, based around finding the knee point of the biomarker histogram. The threshold values found by

In detail, this implies the extraction of raw data and computation of features inside Google Earth Engine and the creation, assessment and selection of classifiers in a

Medan Reichenberg lyfter fram vikten av att väcka elevernas läslust, vilket är viktigt att tänka på när du som lärare väljer ut texter och böcker Reichenberg (2014, s. 15)

Similar to the previously discussed examples, the behaviour for component observation needs to superimposed on the component object. The solution in the context of the layered

In contrast, to provide connection function- ality without relying on specific language design aspects (beyond the standard notion of functions), nodes along with branches are

Powerful equation- based object-oriented (EOO) languages such as Modelica are successfully used for modeling and virtual prototyping increasingly complex physical systems

[r]