United Arab Emirates University Scholarworks@UAEU

Theses Electronic Theses and Dissertations

4-2016 Sand Area Changes in the Emirate of Abu Dhabi, United Arab Emirates between 1992 and 2013 Using a Time Series of Satellite Imagery Rami W A Saeed

Follow this and additional works at: https://scholarworks.uaeu.ac.ae/all_theses Part of the Education Commons

Recommended Citation A Saeed, Rami W, "Sand Area Changes in the Emirate of Abu Dhabi, United Arab Emirates between 1992 and 2013 Using a Time Series of Satellite Imagery" (2016). Theses. 281. https://scholarworks.uaeu.ac.ae/all_theses/281

This Thesis is brought to you for free and open access by the Electronic Theses and Dissertations at Scholarworks@UAEU. It has been accepted for inclusion in Theses by an authorized administrator of Scholarworks@UAEU. For more information, please contact [email protected]. Q:u1.oJI a.H.J.5UI I...:'' 'J La� I Cisull? lmEU ited Arab Emirates University

nited Arab mirate nl ity

!lege r Humanitie and ocial ci nc

Depaltment of eograph & rban Planning

A H NG IN THE EM IRA T OF ABU DH BI C IT D RAB MIRATE BETWEE 1992 AND 2013 U ING A TIME ERI OF SATELLIT IMAGERY

Rami W aeed

ubmitted in partial fulfilment of the requir ment fo r the degree of \1a ter of cience in Remote en ing and G ographic lnfonnation J tem

nder the upervi ion of Dr. azml aleous

April 2016 ii

Declaration of Original Work

1. Rami \V aeed. the under igned. a graduate tu dent at the nit d rab

mirate Tni\ er. it) ( E ). and the author of thi the i entitl d --Scmel area chonges in the emirate of Abu Dhabi, 'niled Arah Emirate, between J 992 and

:;013 u\ing (/ rime series of satellite imageIJ,'·. h reby. olemnly declare that thi the i i m 0\\ n original research work that ba been done and prepared by me und r the upen i ion or Dr. az ml aleou . in the College of Humanitie and ocial clence at . Tlli work ha not pre iously been presented or pu bli h d. or fo rmed the ba i fo r the award of any academic degree. diploma or a imi lar title at thi or any other uni ersit . n)' materials borrowed from other ource (\vhether pubJi h d or unpublished) and relied upon or included in my the i haw been proper!. cited and acknowledged in accordance ith appropriate academic com ention . I fu rther declare that there i no potential conflict of interest \\ i th re pect to the re earch. data collection. au thorship. presentation and/or pu blication of thi the i .

_;"':�:'::' �';":' �= :...:..LL"=-,L? �,c. L..=�L,,- -=---- Date: .- __ '51- S- '20/0 tuden!' s ignature: " _ -=------'---=-- ;/ iii

opyright � 2016 Rami W A aeed All Rights Re erved iv

dvi ory Committee

1) d\i or: Dr. az mi Zeidan aleous

Title: A ociate Profe or

Department f eography & rban Planning

ollege of Humanities and cial ci nces

2) o-ad\'i or: Dr. alem Mohammed Ghaleb I a

Title: A ' ciate Prate or

Department of GeologJ

ollege of cience v

Approval of the Ma ter The i

Thi Master The i I appro ed by the fo llowing Examining Committee Members:

I) Ad isor (Committee Chair): Dr. azmi aleous

Title: Associate Professor

Department of Geography and Urban Planning

College of Humanities and Social ciences

. ..- ,.- Signature �'bl�b CV"\ 6 Date I ) , $ > I L

2) Member: Dr. hahrazad Abu Ghazleh

Title: Assistant Professor

Department of Geography and Urban Planning

College of Humanities and Social Sciences �r� -- Signature Date \ I ;. ) L l

3) Member: \l/A. Title:

Department of ...

College of ...

Signature Date

4) Member (External Examiner):

Title: Dr. Marouane Temimi

Department of Water and Environmental Engineering

Institution: Masdar Institute of Science and Technology

Signature c9f? Date If· ) . 16 vi Thi Master The is i accepted b):

Dean of the ollege of Humanities and ocial ciences: Professor aif alim AI-

Qa}di

�� ignature � Date � - /8, 5 . 6 �

Dean of the College of the Graduate tudies: Professor Nagi T. Wakim t-!1lf jgnature 4

CoPY l of lL vii

Ab tract

and encr achment i a major problem that affect arid countri s and ha evere con equence' n their infra tructure. It po e a thr at to the environment. roads. habitats. fa rm and plantation thu re quiring human intervention fo r the removal of the encroaching and . Tracking sand dune movement and sand changes in such reg n l and tud) ing their trajectori s in time IS \ ry important. It allow gO\ mment to plan better counter mea ure to prevent their occurrence and minimize their danger.

The Emirate of bu Dhabi under ent significant ocioeconomic changes during the

tudy period. \\ hich re ulted in an unprecedented boom in population growth. reflected b) increa ed demand fo r new infrastructure and urban development. Sand movements are accelerated b anthropogenic activities. as witnessed through their encroachment onto farm and interstate roadway . In thi study, "ve used remote sen ing data to map and movement and its effe ct on urban and agricultural areas.

ing six individual Landsat scenes to create a mosaic fo r each of the study dates of

1992. 2002. and 2013. we created land cover thematic maps using supervised ci a sification. The resulted map were checked and evaluated using higher resolution imagery. namely POT. IKO OS, and RapidEye. They were then imported into GI

\\ here change anal sis \ as run using the post classification procedure. Change analysis results indicated an increase in sand cover. between 1992 and 2013. by

1.26°�. The using of Landsat imagery to track changes in land cover fe atures over a large region and across time proved to be very useful fo r better understanding of changes. their trajectories. and their causes and impacts. viii

Kc) \\ ord : Remote en mg. I.La nd coyer cia ification. and ar ea ch ange. bu

Dh abi. IT ix

Title an d Ab tract (in Arabic) f'� 0::-: o�\ �.;aJ\ �\)..4)' \ '�.J:I1 o.)1.4l � �.J\ �\ji\ � �L...w

��\ .)W�\.)� lJ.4 �j � f'\�l:2013 f'�J 1992

�I

j-c> wla� 4..h...,1..J:l ;U�'il .J RapidEye .J Spot .J IKONOS ��I .J..,9'i1

}1.26��2013 .J 1992 x

",UuJ\ e--")l" � wl�1 � wL,lj �� jc�y.dJI fl�1 �;I 4...... :1.J�1 l')-;.'�i

J wl�\ o� Aj� \� � 4,ji 0 � �j o� r.J o� � > �)'11

,lAo.JLiIJ �4,.JJ \+i1.Jl...... o

�� JL.)\ A.Sy. ,�I�I wL.yh..J\ � '�jcJ,�.... ::, :i,u,'J'1 :�)I �\ �liA

,o�1 ��I wl.JL.),\ '�.,J:li ,�)I �I.J'J'I � Acknowledgement

) thank go to both my profe or 0 Dr. Nazmi aleou and Dr. alem I a

\\ ho c upport and teachings Jed me to better under tand the field of remote n mg

& GJ _ our di cu ion taught m a great deal.

p cia1 thank go to m) mother. fa ther. br ther _ sist rso and my significant other v-,ho helped me along the \\ ay.

To all my r lati\'e 'A ho waited eagerly fo r my dissertation. especially ncles

ultan & amer.

I am ure the) u pected it was endles 0 but here we are now.

Told you I'd get it done © xii

Dedication

To my belored parent· and fa l77i�v

Thank you for ) '0111' support xiii

Table of Contents

' " ...... I itle ...... i

Declaration or Original Work ...... ii OP) right ...... iii Ad\ i or)' 0111lnittee ...... iv ppro\ al of the aster The is ...... \' b tract ...... vii Title and bstract (in Arabic) ...... ix ckno\\ ledgenlent ...... xi Dedication ...... xii

Table of Content ...... xiii Li t ofTabJes ...... xv List of Figures ...... xvii

Li t of bbreyiation ...... xix

Chapter 1: Introduction ...... 1

1.1 Background ...... 1

1 .2 The tudy Area ...... 2

1.2.1 Phy ical etting ...... 6

1 .2.2 Clin1ate ...... 10

1 .2.3 Socio-Economic ...... 11

1.3 cope of the Research ...... 12

1.4 Aim & Objective ...... 12

1.5 Hypotheses and Assumptions ...... 13

1.6 Thesis tructure ...... 13

1.7 Sumnlary ...... 14

Chapter 2: and Dunes and Remote Sensing ...... 16

2.1 Background ...... 16

2.2 Benefitsand Limitations of Remote Sensing of Sand Dunes ...... 16

2.3 Multispectral Classification & Mapping of Sand Dunes ...... 20

2.4 Change Analysis Techniques ...... 24

...... 26 2.5 Sun1illary ......

...... · .. ··· .. ········ 27 Chapter 3: Research Methodology ......

...... 27 3.1 Introduction ......

...... 28 3.2 Spatial Datasets & Software......

.. ·· ...... ·.·.. ··········.... · .. · .. ······· 30 3.3 Data preparati on ...... xiv

3.4 la iflcation cheme ...... 32

..,. � Determining L L cIa es ...... 34

3.6 Image CIa ification & Po t Proces ing ...... 38 -'.7 Extracting and dune from sa tellite image ( 1992. ETM+ 2002. OLI 2013) ...... 39

3.8 ccurac) A e nlent ...... 39

3.9 Dune change anal)" i technique ...... 4 1

ha pter 4: Re ult & Di cus ion ...... 46

4. 1 Training ite CIa Identification ...... 4 6

4.2 Training ite election ...... 48

4.3 1992 LULC Map ...... 49

4. 4 2002 LULC Map ...... 51

4.5 20 13 LUL Map ...... 53

4.6 CCllIaC) A e Inent ...... 55

4.7 Change Analy is ...... 58

4.8 Di Cll ion ...... 70

Chapter 5: Conclu ion & Recommendations ...... 81

Bibliography ...... 84

ppendix ...... 89 xv

Li t of Table

Ta ble 1.1: Mineralog) o[th ...... 6

Ta ble". 1: ummar) r Da le and ene d ...... 90

Ta ble 3.2: mpari on bet\\een DR of Land at .f -7 and Land at 8 ...... 32

Ta ble" .3: Theme apr nder on 1976 ...... 33

Ta ble 3A : Re\ i ed nder on 1976 Theme a per Region of tudy rea ...... 33

Ta ble 3.5: Original Image ...... 91

Ta bl .,.6: In\'er e PC ...... 91

Table 3.7: List of accuracy asses ment sites ...... 95

rable 3. : Imagery u ed fo r accurac a essment...... 90

Ta ble 3.9: Union of three years ...... 43

Table 3.10: Final Product of from - to change across three ea rs ...... 43

Ta ble 3.1 1: Change traj ectorie betv;een two years ...... 44

Ta ble 3.12: Change traj ectories fo r sa nd and non-sand between two ea rs ...... 44

Table 3.13: Change traj ectories fo r and and non-sand across three years ...... 45

Table 4.1: tatistical CIa ifiable Clas es...... 48

Table 4.2: 199� fea tme clas areas ...... 49

Table .f.3: 2002 fea ture class area s ...... 51

Table .fA : 2013 fea ture class area ...... 53

Ta ble 4.5: Accuracy A essment of 1992 Classification ...... 56

Table 4.6: Accuracy Assessment of2002 Classification ...... 57

Table 4.7: Accmacy Assessment of2013 Classification ...... 58

Table 4.8: Changes from 1992 to 2002 statistics ...... 97

Table 4.9: Changes fo r sand features from 1992 to 2002 statistics ...... 97

Table 4.10: Changes from 2002 to 2013 statistics...... 99

Table 4.1 1: Changes fo r sa nd fea tures from 2002 to 2013 statistics ...... 99 xvi

'I able 4. 12: and "0:on and change anal} tati tic from 1992 -2002 -2013.. ... ] 02 xvii

Li t of Figure

Figure 1.1: ' E' land cape in true color Worl d iew-2 compo ite (5,'".2) ...... -+

Figure 1.2: E' e\ en mirate boundarie ...... 5

Figu ...... re l."': Dune Face and Wind Inter ...... actions ...... 7

Figure 1.4: Di tribution ...... of global de ert ...... 8

Figure 1. : Relati - n hip of Dune Pattern & Wind Directions ...... 89

mo ...... Figure 1.6: and ycment ...... with \\ ind ...... 1 0

figure 3.1: Di ertation ...... methodology process ...... 27

gure ...... Fi 3.2: Land at Path & Ro\ s Covering the UAE ...... 29

r Figure 3.3: The tud r a fthe Dissertation inside Red Boundary ...... 29

Figure"'.4: Representation of mosaic eamlines ...... 31

Figure 3.-: Pilot tudy Area ed to Identify Classes ...... 34

Figure 3.6: Forward PCA of Pilot tudy in First Three Principal Components

(PC ) ...... 35

Figure 3.7: Inver e PCA of the First Three PC in True Color RGB ...... 36

Figure 3.8: Entropy Texture of Pilot tudy Area ...... 37

Figure 3.9: I ODAT Results fo r Pilot Stud ...... 38

Figure 3.10: Distribution of accuracy assessment sites ...... 40

Figure 3.11: Multi Year change detection methodolog ...... 42

Figure 3.12: Change trajectories across three years ...... 45

Figure 4.1: Final Outcome afterMe rge of Classes ...... 47

Figure 4.2: Final Outcome afterMer ge of Classes ...... 47

Figure 4.3: Thematic Map of 1992 ...... 50

Figure 4.4: 1992 Thematic Map Showing and & on-Sand Classes ...... 5 0

Figure 4.5: Thematic Map 0[2002 ...... 5 2

Figure 4.6: 2002 Thematic Map Showing Sand & Non-Sand Classes ...... 5 2 xviii

Figure 4.7: Thematic lap of2013...... 54

Figure 4.8: 20 13 Thematic Map h \.\ ing and & on- and la e ...... 54

Figure 4.9: hange Trajector) Map from 1992 to 2002 ...... �9

Figure 4.1 0: bange Trajectory Map of and from 1992 to 2002 ...... 60

Figure 4.1 1: hange Trajector) Map of and from 1992 to 2002 ...... 61

Figure 4. 12: hange Trajectory lap from _002 to 2013...... 62

Figure 4. 13: Change Trajectory Map of and from 2002 to 2013...... 63

Figure 4.14: Change Trajector) Map of Sand from 1992 to 2002...... 64

Figure 4. 15: Change trajector) from 1992 to 2002 to 2013...... 65

Figure 4.16: Total 0[203 change trajectories from 1992 to 2002 to 2013...... 66

Figure 4. 17: Change in and trajectories from 1992 to 2002 to 2013...... 67

Figure 4.18: Total of 85 change trajectories fo r and from 1992 to 2002 to 2013..... 68

Figure 4.19: Change Trajectory Map of and / on Sand from 1992 to 2002 to

2013...... 69

Figure 4.20: Feature Clas ed as Non-Sand ...... 70

Figure 4.21: Feature s Classed as Sand ...... 71

Figure 4.22: Coastal areas in 1992 ...... 72

Figure 4.23: Coastal area in _013 since 1992 ...... 73

Fi gure 4.24: Sate of coa tal areas in 1992 ...... 74

Figure 4.25: Coastal change in 2013 since 1992 ...... 74

Figure 4.26: Population distribution in 1992 ...... 75

Figure 4.27: Population change in 2013 since 1992 ...... 76

Figure 4.28: Road network limited in 1992 ...... 77

Figure 4.29: etwork enhancement allowed growth in 2013 since 1992 ...... 77

Figure 4.30: Empt., unused spaces in 1992 ...... 78

Figure 4.31: Subsides allowed expansion in 2013 since 1992 ...... 79 xix

Li t of Abbreviation

E nited rab Emirate

GL Geographic Infomlation ystem

GDP Gro s Dome tic Product

DR limate Data Record

T 1 Thematic lapper

TM+ Enhanced Thematic Mapper Plus

Ll L Land U e / Land Cover

DVI omlalized Diffe rence Vegetation Index

J Crust Index

G I Grain ize Index

AVI oil dj u ted Vegetation Index

PC Principal Component Analysi

PC Principal omponent

1 Multi pectral canner System

LEDAP Landsat Ecosystem Disturbance Adaptive Processing Systems

�ODIS Moderate Resolution Imaging Spectroradiometer

DEM Digital Elevation Model

TOA Top of Atmosphere

I ODATA Iterative elf Organizing Data Analysis Technique 1

Chapter 1: Introduction

1.1 Background

The nited rab mirate ( E) i located in the ea tern region of the

rabian Penin ula, we t ia. It border the Gulf of Oman and the rabian Gulf.

° b t\\ een Oman and audi rabia, between coordinates of 22 45' and 25 45' and 0 ° 51 45' and 56 E. Its climate i e tremely arid and to semi-arid the east rn mountain of the country. It area cover a total of 83,600 km2 and i composed of ilat. ban-en coastal plain m rg ing into rolling sand dunes of va t desert making up

750 0 of the country (El- ayed, 1999), and mountains in the east (Figure 1.1).

The country is a union of even em irates; Abu Dhabi. Aj man. Dubai.

Fujairah, Ra al Khaimah. harjah. and Umm al Quwain. Six of the seven emirates first came together in 1971. with the seventh. Ra al Khaimah, joining in 1972

(Figure 1.2). bu Dhabi makes up 87% of the countr 's total surface area and ho t the capital city. bu Dhabi.

The UAE's land boundarie extend up to 1066 K111; 609 Km of which are

\\' ith Oman and 457 K111 are with Saudi Arabia. The country' s coastline extends to

1318 Km and is situated in a trategic location along the southernpa rt of the Strait of

Hornluz, which is a vital transit point fo r world crude oil transport.

Before the discovery of oil in the countr . the Emirate of Abu Dhabi was one of the poorest coastal states. where its economy was limited to and pri marily relied on oa is agriculture, fishing, pearling and trade with neighboring countries (AI

Ahbabi 2002) . However. after the discovery of the first commercial oil field (Bab 2 field) in 1958 � 1I0\\ ed b) the fir t crude oil hipm nt in 1962. Abu Dhabi quickl) became one of the riche t tate in the region, \\ hich allowed the fu nding of dc\ clopment project . not just acro s the emirate but also acro s the union just \\ ithin

10 years of entering the global oil market.

the fir t pre ident of the E. heikh Za 'ed bin ultan Al- ahayan immediately fo re a\\ the need to inve t in the de elopment of the emirates' infra tructure. Plans to dev elop nation quickly took place and many fa cilities were rooted uch a road netvvorks. educational institution , and ports. Oil re enues of the countr) were a major turning point. not just fo r the Emirate of Abu Dhabi. but the country a a whole.

By 20 14, the United ations estimated the UAE's population to be around

9.445.624. \v ith 800 0 of the total population made up of immigrants, all living in a country that is vastly dune . Coupled with the rapid urban expansion of the countr) .

and dune pose a risk to road networks, human habitats, and farn1 lands due to their migration by winds that reach speeds between 6 - 10 ml s.

This and movement bas fo rced the government to continuously and vigorously invest in either setting up barriers or relocating sand dunes via heavy machiner} . both being costly endeavors.

1.2 The Study Area

The Abu Dhabi emirate is located in the westernsout hern paIi of the country, and is divided into three administrations: Abu Dhabi, Al Ain, and the Western

Region Municipality. The emirate borders Oman to the East. Saudi Arabia to the

South and West. and the Arabian Gulf to the NOlih. It inc1udes over 200 islands, 3 including major one uch a Mubarraz. Zirk)u. Delma. ir Bani Yas, Marawah. bu

b} adb, and aadiat. The emirate could be di\ ided into five diffe rent landfoml : sa nd dune , interdune area . coastal abkhas. inland abkllas. and expo ed rock

(Glennie 1996).

The Abu Dhabi emirate wa elected fo r thi tudy because it makes up 87% of the total cOlmtr), in teml of land area , and the large majority of the mirate is cOvered \\- ith dune . that pose a threa t to fa mls and roads as they migrate.

Furthermore. dun s, in g neral, have not been tudied extensi ely in the emirate. thus pre enting the opportunity to fu rther explore ho\\ Landsat could support dune

tudie . 10reo\'er, the E, a a whole, sa w an unprecedented urban growth in the time pa n of 40 odd years, where the whole country changed from a nomad-like

of civilization. thu giving the oPPOltunity to see the life tv. le to a brimmin'-'g beacon dra tic change in land cover over a hort period of time. Legend

o VI l-.dA'IIFlEr'I"1If""

1:1 ,900,000 012525 iiO 15 100 Kl1omatort " �lb mi W � �.... oH1 ��t la-_19'i @I �!-_ _,--{_ � 7"loq101\ , f"��:�-i'

Figure 1.1: UAE's landscape in true color WorldView-2 composite (5.3,2 )

4 Emirates of the UAE

Legend

Eml,.., ••

Ab\I()ft,abO

.,...

Ouoo,

'�.n

Ratrtll(�..,...t'I . s....,." u-. .. AI.' DS'�

1:1 ,900,000 012525 so 15 100 _ _ Kilometer" r. .. . . ,-;-' �®, ' ,

Figure 1.2: UAE's seven Emirates boundaries

5 6

1.2. 1 Ph 'ira) etting

[he E' terrain i compo ed of a rich \ ariet) of landfom1 traver ing a

hort ra nge of di ta nce. tudie ha ve noted that the minerals that could be cnc untered are: Quartz. alcite. Dolomite. northite. Gyp um. Anhydrite. Halite.

ra gonitc. and Cele tit (I Iowari. 2007). Whil other studies that used La nd at and ba nd ra tio to perform la nd coyer clas ification ba sed on urface mineralogy. ba e re\ ca1cd that th fo ur main minerals that could be r motely sensed are Quartz.

Calcite and Dolomite (carbonate minerals). and Anorthite (feldspar mineral) (Pea se,

1999). uch. the minerloaogical compo ition of the E fa ll into five cla se :

arbonate. Mixed Carbonate. Quartz. Mixed Quartz. and Feldspar (Table 1.1)

Mineral Mineral Type

Quartz Silicate Mineral

Calcite Carbonate Mineral

Dolomite Carbonate Mineral

Anorthite Feldspar Mineral

G 'p um Sulfate Mineral

Anhydrite Sulfate Mineral

Halite Halide Mineral

Aragonite Carbonate Mineral

Cele tite Sulfate Mineral

Table 1.1: Mineralogy of the AE 7

Dune are frequent in de ert em ironment . and the) ar an accumulation of

sediment blo\\TI b) wind into m und or ridge . The) ha\ e two face . one i a gentle

\\ inoward lop that i \V ind fa cing and dominated by and depo ition and and

altation. \\ hile the other i a dov. nwind side that i commonly a steep avalanche

kno\,\.11 a the lip face. The lip rac . stand at an angle of repo e. which i the

maximum angle at which th material would lose stability. cau ing and dune

migration. and that angle lie b tween 30° and 34° fo r sand (Figllre 1.3); Mor over,

dune an have more than one lip face ( ummerfield. 1996).

� Crest �

� Windward Face

r-.... sand �

--- ......

� � --�---- � � r-.... �����,

---

Figure 1.3: Dune Faces and Wind Interactions

Desertification. sand encroachment. and dust storms are some of the geological

hazards related to the increasing climate change and anthropogenic ac tivity fo und in 8 arid land a ociatcd \.\ ith Aeolian } tern . eolian sa nd dune are one of the mo t

\\ idel} tudied land c ver fea ture in the \.\orl d. The} Co\ er almost 10°o of the land ° bc t\\een latitudes 30 and "'0 ( arnthein. 1978) and occupy about 75°'0 of the

nited ra b Emirate ) surface area (Figure 1. -/), thus ma king it one of the most important geom rphol gical fea tures in the country (El- ayed, 1999). and nearl} all 1...110\\ 11 type of and dune occur in it (Embabi. 1993 ).

Distribution of Global Deserts

Legend 1 :79,000,000 o 875 1 750 3 500 52S0 ... 000 - - K«me!eB

Figure 1.4: Distribution of global deserts

Aeolian sa nd dune are prominent in the UAE due to the country's climatological patterns. where the lack of ra infall and summer temperatures reaching

50°C. coupled with the scarcity of vegetation cover and strong wind pa tterns. all 9

ontribute to the uita ble c nditions that enhance and dune mOYement (Gl nnL

200 L 1- \\adi. 2004)

and dune are dynamic in nature, "" ith dune changing their location. length.

or height. depending on their t) pe (T oa r. 2001 ) and they are ma inl) influenced by

\\ ind regime. and uppl). \'egetation cm er. and grain size. Change in any of the aforementioned va riables \\ ould affe ct hov,: dune are fo rmed. fo r e 'ample, a linear dune i i'0l111ed under a bidirectional \:v ind regime, whereas a transverse dune is

fo ml ed under a unidirectiona l \v ind regime V\ ith an ample suppl of sa nd (Wasson &

Hyde. 1983)

ccording to T oa r (2004). sa nd dune could be cla sified into three di tinctive groups (Appendix A Figure 1.5):

a) Migrating Dune : where the whole dune body advances with little or no

change in the shape and dimension. Transver e and crescent (barchan) dunes

are exanlples of migrating dunes.

b) Elongating Dunes: where the length of the dune increases over time through

a very ditTerent proce than the ones that affect the migrating dunes. Linear

dunes are examples of elongating dunes.

c) Accumulating Dunes: where dunes exhibit little or no advance or

elongation. tar dunesar e an example of accumulating dunes.

In the stud area, the major types of dunes encountered are ba rchans.

Barchanoid. and trans erse fo r the migra ting group, linear fo r the elongating, and star

dunes fo r the accumulating (Embabi. 1991) . 10

s fo r their movem nt. and dune are knov, n to moye or jump abo\ e the

'urface \\ hen v..ind peed rea ch 5.5 - 6 m/sec. a pr ce known as altation that

ausc and to be accWllulated on the crest (Figure 1.6)( bu l-Khair. 1981)

WI D suspension

ueep and rol

-...... -+- • • • • • • • • • • 7 ...... ,.

b€ach Source: Londform�.eu

Figure 1.6: Sand movement with wind

1.2.2 Climate

. The Emirate 'v' eat her during the year ca n be summarized into two key

ea ons:

Summer - Starting in May and lasts until September, the countr fa ces extreme heat and hWllidity. During this period temperature could reach up to 48°C

\\'hile humidity ra nges beN/een 80 - 90 %, andstorms regularly strike between May to early July. while ra infall is ra re.

Winter - Starting in December to March, temperatures reach a high of 24°C during the day and a low of 10°C at night, and fo g becomes more frequent during this period. Rainfall occurs most often between ovember and February and. on average. reaches 12cm per year. 11

On a\ ra ge. the annual \\ ind speed in the Emirate of Abu Dhabi ra nges between 4.63 and 5.6 111/ , ,",.ith the a\'erage maximum reaching between 6 to IO n

The predominant direction are 11h We 1. orth orth-West, and est orth-

We·t 21.1% . 15.8° 0. and 14.6% of the time, respectively.

1.2.3 ocio-Economic

The E' economy ha prospered. and its nominal gross domestic product

(GDP) ha s grown 27 time ince 1975 (National Burea u of ta tistics). becoming the

ec nd largest econolll .. in the ra b world after audi Arabia. This unprecedented pa ce f gro\\1h over the past year has been largely supported by the revenues from

i 1 and ga exports. v" hich the country ha reinve ted in itself to prepare itself fo r a po t-oil era by diversifying its economy. These efforts ha ve driven an increa sed

p rformance 111 trade. financial erVlce , tourism, rea l estate, logistics, communications and manufacturing. The country has a uniquel flexible, globalized labor market. and the track record of economic growth has also ensured a very high growth ra te in population.

Thi population growth places challenges on the nation when it comes to trying to meet its demand fo r energy, water, and urban development. ince the country is located in the desert. where fresh wa ter is limited and temperatures are high during the summer, the UAE's carbon fo otprint is one of the highest in the

world (Globa l Footprint etwork. 201 0). There is a high demand fo r energy and

many goods cannot be produced by the country, therefore. goods need to be

imported. and freshwater needs to be desalinated. 12

1.3 cope of tbe Re earch

Various tudie '.\ere conducted \-vorld\\ide u mg remote sen lI1g and geographic infomlation y tem (Gf ) to map and understand the mo\ ement of and dune . Ihi stud) fo cu e on changes in sand over time in bu Dhabi Emirate using the e technologie and method logie .

1 A Aim & Objective

Thi tud) is of paramount importance as 75 % of the UAE is co ered by

and dune (El- ay ed. 1999). Their mo ement , aided b) wind. poses a threat to the em ironment. roads. habitats. fa rms and plantations (Yao. ::W07).thus requiring human interyention fo r the remo\ al of the encroaching sand. Therefore. by better under tanding the hifts in land co er fe atures. better planning could be done fo r the future.

The objectives of this study are:

1. To create mosaic of the newly released Landsat CDR imagery covering the

study area for 3 dates: 1992. �002. and 2013.

2. To identifythe classification scheme ofthe study area with the minimum

representative land cover classes of the study area.

3. Isolate the sand class and create sand maps fo r the study area from Landsat

imagery for the three dates.

4. Perform change analysis tlu'ough the selected time periods, and di cuss the

fm dings in regards to the movement of sand in the emirate. 13 l.� B y pothe e and A umption

We a ume that dune mo\ ment are a fa ct of nature. accelerated by anthropogenic acti \ itie . as v, itne ed thr ugh their encroachment onto fa rm and interstate road\... ay . Our h) p the i i that these movement can be measured and modeled u ing remote en ing technique and tatistical anal)' i .

1.6 The is tructurc

Thi di sertation is divided a follows:

• hapter 1 : Introduction

This chapter de cribe the study area. the Emirate of Abu Dhabi , and

provides a brief background into the geographicaL physical, and economic

etting of the UAE a a whole. It discusses the location, size, climate and

landscape of the tudy area, giving a ummar of the impact of dune

movement onto the various land covers and its impact on the socio-economic

field. Finally. it end with the research objectives, hypothesis, and thesis

structure.

• Chapter 2: Literature Review

In this chapter. we talk about the vanous remote sensmg and

geographic information system (01 ) techniques and research methods done

previously on dune movement across the gl obe. As well as how they are

closely related to the current study area in order to analyze the most suitable

methodology to utilize in studying the Abu Dhabi emirate' s sand dunes. 14

including the \anou thematic map development technique , the

cla ification pro e . and the change anal) i algorithms utilized.

• Chapter 3: Methodolog)

Chapter three di cu e the uti lization of what was learned from the

literatur reviev.. . data collection the tudy v. a perfom1ed on. de cription. and

pre-proce ing. and the \'arious methods and techniques that were performed

to analyz and explore it results to better understand the area of interest.

• Chapt r -+: Re ult nal) is & Discus ion

Thi chapter explains the result of the analysis of the methodology

that wa fo llo\\ed. the accuracy asse ment of the results. and the result

interpretation in scientific fo rm. while discu sing the driving fo rces and their

impact at the arne time.

• Chapter 5: Conclusion & R commendations

The final chapter concludes the research work done and the

significant issues that are covered to set up better fu ture research to obtain

better results.

1.7 Summary

The Abu Dhabi Emirate is made up of three municipalities, Abu Dhabi, the

Eastern Region, and the Western Region. The emirate has similar physical settings as the rest of the country: the same weather and landscape could be seen tlu'oughout the region. with climates being bi-seasonal; hot and humid in sun1mer, and temperate and sunny. with sparse rain during winter. 15

The exploration and production of il and ga in the emirate ha led to a drastic change in the land CO\ r of this regi n. Due to th continuou increa e in population, coupled \\ ith the limited re ources of freshwater and high temperature during the ummer. the need fo r con tant urbanization and infrastructure development i e 'entia!' The e pr �ect are lowl) changing the land fe ature over the )car . \\ here land i exploited and tranSf01111ed fo r human use (Gardner 2009).

Thi tud) aim at mapping the and fe atures. and tracking their trajectory across time to better under tand the process and call es of and movement. 16

Chapter 2: and Dune and Remote en ing

2. 1 Background

Field un e} or dun migration are difficult and expen i\'e; they cannot be repeated urficientl., oft n to permit ongoing monitoring. For that reason, the use of remotely , n ed data i a highly effecti\ e method that allows researchers to tudy and monit r \\ ide area of and dune .

There are man} different methods that can be used to mea ur the migration rate of and dune , uch a the change detection techniq ues offered by Kumar

(1993), neighborhood stati tic by Larue (2004) and Bishop (2008), and the vector extraction from ra�ter a done by Liao (2009).

2.2 Benefit and Limitations of Remote Sensing of Sand Dunes

and dune sy tem are large in areas. and when sand migrates, they cover long distance and wid area through transport pathways as indicated by previous field geochemical and remote sensing studies that will be explored in this chapter.

These studie also howed that the Landsat Thematic Mapper (TM) is capable of distinguishing between active and inactive sands ( cheidt, 2008).

The characteristics of the spectral. spatial, and temporal resolutions on both the Thematic Mapper on board the Landsat 5 and the Enhanced Thematic Mapper

Plus (ETM+) onboard the Landsat 7 are quite imilar (Masek, 200L Teillet, 2001).

However, the main improvements of the ETM+ over the TM are the improved thermal band. the addition of the panchromatic band. and the availability of two calibration modes from radiance into D (Levin, 2004). 17

[h reflectanc pcctral ignature of sand dun a detected by pace borne

'5cn or i highl) dependent on t\\O criteria. Fir t. the characteristics of the dune and their urface . \., hich include vegetati n cover. mineralogy. teo ture. and the pre ence or a biogenic oil cru t (White. 1997. Gerbermann and eh r. 1979. Karnieli and

T ·oar. 1995). ec nd. i the geometry among t the un. the urface. and the en or. and the atmospheric attenuation (Howari, _007). It hould be noted that under the

[1 en or. pectra of and would be registered as sand a long as the cell under investigati n does not contain more than 30 to 40 percent vegetation (Hutchison.

1982. Tueller. 1987. mith, 1990. Janke, 2002).

ince part of this tud ' fo cuses on and dune classification and extraction.

eyeral aspects that affect hO\ . the ensor records the dunes are investigated:

a) Mineralogy: Diffe rent minerals reflect diffe rently. and in the spectral

resolution of Landsat Quartz and Carbonates show a significant diffe rence in

the Mid-Infrared (band 5 and 7). Whereas Quartz hows high reflectance.

Carbonate shows half that of Quartz, therefore, using band ratio 5 / 7 helps

di tinguish Quartz from Carbonate (Pease. 1999), which in turn generates a

gradient from back to white indicating the decreasing Carbonate and the

increasing Quartz. Another band ratio s used in previous studies was the

Them1al Infrared / ear-Infrared (6 / 4) ratio. This ratio plays on the

temperature variation that minerals have from one another, where mafic

minerals are distinguishable from Quartz and Carbonate (Pease. 1999).

because of the higher absorption of the mafic minerals. However. due to

subtle variations, it is coupled with the infrared band 4 to highlight them 18

better. Thi method proved u eful becau e ma fi c minerals a have \\ eaker

rc pon in ba nd 4 than in band 6 (Hunt 1982. incient. 1997. Pease. 1999). b) Texture: Te. ture i relat d to the grain ize that covers the top of the and

dune. and using remote en ing. the ob en d pectral diffe rences between

the active and inacti\ e and dune i a result of grain size and composition

diffe rence. ctiYe and i con istently brighter and i generally expre ed b ..

a uni-modal di tributi n of brightne s va lues a a re sult of the fine sa nd

particl s (le s than 62 ).1m) while the inactive and has a lower albedo due to

the more coa r e pm1icles (gr ater than 250 ).1m) (Lam. 201 1). c) Geometry: Topographic shading and cast shadows can introduce enor into

the accurate analy is of remote sensing ince the mask significant spectral

features. and ometimes are trea ted as noise that needs to be remo ed.

However. shading also stores vital infom1ation about the topographic

characteristics and in thi sense. the noi e becomes an impo11ant signal that

could be extracted and anal� zed (Levin. 2004). d) Atmospheric Attenuation: Ground fea tures measured by remote sensmg

platforms always include en ironmental phenomenon that is measured as a

result of the attenuation of electromagnetic ra diation through the atmo phere

till it reaches the sensor. This data can be trea ted as either noise or a signal

depending on the study subject. In other words. the absorption of ra diation

due to the gases in the atmosphere ca n be seen either as a noise masking the

ground reflectance or as a signal that shows the distribution of the

atmospheric ga ses (Gao. 1993).

Due to the nature of remote sensing data from the Landsat series sensors, it requires that solar radiation passes through the atmosphere before it is measured by 19 it. \-\ hich in tum lead to en ed data to include information about the atmo phere a

\\ ell a the urface.

This tudy u e image pro\ ided from the Land at eries of atellites ince the) pro\ ide the longe t continuou record of pace bome imagery that began in carl) 1972 \\ ith the launch of Land at I. Hovv ever. the inherent issue that is passed along due to the use of multi-temporal images in a study is the radiometric con i tenC) among ground fe atures. It is difficult to maintain this consistency due to the change 111 en or characteri tic . atmospheric conditions, solar angles. and sen or Yle\\ angles. Therefore, radiometric correction is often perfom1ed to reduce al1) or all fthe influence abO\ (Chen. 2005).

There are two types of radiometric corrections that al'e utilized to correct

atellite image : ab olute and relative correction. Absolute correction aims at extracting the ab olute reflectance of the target at the Earth's surface. It requires

inputting imultaneous atmospheric characteristics and ensor calibration, \ hich is difficultto acquire in many cases, especially in historical datasets, but also diffi cult to obtain when plannedfo r. As fo r relative cOlTection. it aims towards perfonning the

ame objectiYes met through absolute correction, by adjusting the radiometric properties of the target images to match a predetem1ined base image.

Radiometric correction is also a vital step when it comes to mosaic scene classification. Jay states that unless radiometry of the mosaicked images are of acceptable levels. then digital analysis is not suitable (Jay 2009). 20

2.3 �1ulti pectral Cia ification & Mapping of and Dune

ari u method are implemented to better enhance the captured image and then be able to extract the and field of interest. In one tudy by Howari CW07). the image proce ing technique u ed \\ r image enhancement. band ratioing. and

pectral cla ification. Pea e (1999) tated, that u ing Land at TM band ratio 6/.f and 5 7 re\ ealed the mo t infom1ation fo r mineralog) of the and dune : Pease used band ratioing because it helps in reducing the detected diffe rences in the ame fe ature due to topograph) overall variation in reflectance. and brightne diffe rences related t grain ize. and all the vvhile. enhancing the diffe rences due to the shape of the

pe tral refl ctance ( ultan. 1986, abin . 1997. Pease. 1999). Thereafter, both un upen i ed and upervi ed classifications were performed to create the land cover thematic map based on surface mineralogy. Unsupervised classification does not require ground reflectancedata becau it is solely ba ed on the natural reflectance of each pix 1. while the supervised classification is used to assign an unkno��n pixel to on of the classes (Lillesand and Kiefer. 1994. Sabins. 1997. Richards. 2000).

Another method fo llowed to create a thematic map ofthe Land Use / Land Cover

(LULC) \\"as done by Hadeel (2010). where upervised classificationwas used. This produced more de irable results over unsuper ised cla sification when trying to delineate LULC based on band indices to retrieve class boundary rather than the ra band data set. The three indices used are: ormalized Diffe rence Vegetation Index

DVI). Crust Index (CI). and Topsoil Grain ize Index (GSI).

• NDVI is the most common fo rm of vegetation index. and it is the diffe rence

between the red and near-infrared band combination divided by the sum of the

red and near infra-red. 21

NfR - R NDVI =--­ NfR + R

I [O\\C\ CL !nce thi tud) is conducted in an arid area v. ith little continuou

\ cgetati n c \ er, we will uti!iz e it counterpart the oil dj u ted Vegetation Index

( I) \\ hich use a canop) ba ckground adjustment fa ctor that make up fo r the fa ct that there ar pa r e vegetation and vi ible oil.

NIR - RED 1 + L) N I R + RED + L * (

(Where Li the fa tor that highlight the diffe rence between vegetation and

oi\)

• CI i u ed in the tudy of the fine sa nd content in the topsoil: this is fo r

monitoring the change of surface oil using remote sensing. Appl ing the index

to a and oil environment ha shown the ability to detect and ma p the diffe rent

lithological/morphologic units such as active cru ted and areas.

R-B Cf = 1 - -­ R+B

• GSI was developed ba sed on fieldsur ey of soil surface spectral reflectance and

lab analysis of soil grain composition perfomled b Xiao (Xiao. 2006). High

va lues in the G I correspond to high content of fine sa nd. 0 fo r vegetated area .

and negative value fo r wa ter bodies.

R-B CSI = R + B + C

Other authors used simple classification methods to extract the sa nd dunes that varied betw een supervi ed and unsupervised processes. Ghadiry (2010) extracted the sa nd dunes and then transfonned them into vector data to calculate the dune movement. On the other ha nd, AI-Haj ri (2009) perfonned supervised 22 cia ifi ation ba ed n fi eld ample . and the re ultant image led up to on- creen digitilation \\ here their mo\ ement \"a ca lculated.

ne of the 1110 t \\ idel] u ed r mot en ing technique to cla sif)' and image that c ntain 'everal ba nd of information is the multi pectral upen i ed cia sifi alion. Thi method pro\ ed to be reliable and r bu t over the past decades of

it implementation. The upe[\. i ed multi pectral classitlcation process relies on the pr ba bilitie of a igning ea ch pixel in the image to a defined class (Otazu, 2000).

Another method wa the use of pa tial tllter to delinea te the boundarie of the and dune . thu enabling them better extraction either with the use of supervised cia itlca tion with an image add-back (Janke, 2002) or with the use of on-screen digitization (Al-Dabi. 1997). After that, the movement wa s calculated ba sed on a

ubtraction change detection algorithm that extracted the change between the two

images.

nlike the previously mentioned methods. one author. Chang (2006), used

texture anal)' is. Texture analysis is an approach used to identify the characteristics of a fea ture from the relationship among the proximate pixels ra ther than the

reflection va lues on a per pixel ba sis. Simple naked eye inspection of texture does

not allo,,, distinguishing the sa nd dunes from the other land use, except the linea r

pattern of the long axis of sa nd dune area . For that, the use of filters assists in

delineating these fea tures from one another. The most common filter is the high pa ss

filter. which is designed to ignore trivial variance by multiplying higher va lues to a

matrix. the distortion problem of the high pass filter wa s pointed out by Goodchild

(1993), fo r that. the use of other filters is more appropriate. The area co ered by sa nd

proved to have one homogeneity va lue; therefore, the homogeneity filter is good fo r 23

\.:xtra tion [ and dune . On the other hand. the di imilarity value showed zero fo r the and dune . \\- hich mean di imilarity value i also a good fi lter fo r extraction of certain object which have continuoLl \ ·alue . Moreo\er. the Kru kal- Vv allis

tati tic' of texture anal) how a ignificant diffe rence in the po l} gons of sand dune .

Then there i the Principal Component naly is (PCA). where the analysis of satellite imagery ha repeated I) demon trated it capability in identifying the interrelation bet\veen the fe atures based on their lithology. structure. and landforms

(Pu and Gong. ::WOO.Kri ImamUl1hy and riniva . 1996). Using the PCA method, six

Principal Component (PC) are generated from the Landsat ETM+ sensor. fo llowing a maximum likelihood cia sification algorithm. and an 82% accuracy map was de\ eJoped (Raj e h. 2008). Fung and LeDrew (1987) used PCA to detect LCLU changes u ing the multi-temporal M and TM images. and Gong (1 993) used band pair image diffe rencing fo r each pectral band. and then used PCA fo r the multi pectral difference image, which proved to be better than the simple image differencing.

In multispectral remote sensmg. the large number spectral bands being collected would sometimes lead to redundancy in the data being collected as some ground objects hardly vary over certain wavelength ranges. Since not all multispectral bands contribute the same amount of infonnation, it is necessary to reduce these redundancies to improve data storage and image processing perfo1111 ance. (Richards and Jai, 1999)

Data redundancy is usually visualized by plotting a scatterplot of two bands. if the distribution of pixels within the plot fo rm along a line then it is said that the 24 t\\ O band' arc highly correlated, and if the cattering of the pixel i \\ ide, it i said that the information \'v ithin the e t\\O band are uncorrelated ( I11dev Ren. 2005).

nothcr method II ttowed i the cal ulation of the PC . PCA is u ed to pr duce un orrelated output band v. hich egregate noi e components and reduce data et dimen ionality b) fi nding n \'v of Olihogonal axe that have their origin

at the data mean and that are rotated 0 that data variance is maximized (Ifarraguerri and hang, 2000), and it ha the fo llO\ving advantages ( abins, 1987) :

1. Mo t of the \ aTiance in a multi pectral dataset i compres ed into one or two

PC image .

') oise ma) be relegated to the less correlated PC images.

3. pectral diffe rence between materials may be more apparent in PC images

than in indi\'idual band .

PC were derived from the solar reflective spectral bands of the selected imagery, and, as it is expected. the first three components provide almost the whole ofthe explained yariance. therefore, we consider PCs up to thethi rd component.

2 .... Change Analysis Techniques

Change anal sis is the process of identifying the differences in the state of an object or a phenomenon by observing it at different times (Singh, 1989). To better understand the relationships and interactions between the human and the natural processes and to better manage and utilize the resources available, timely and accurate change detection of the Earth's surface fe atures is required (Lu. Mausel,

Brondizio, & Moran, 2004). The use of remotely sensed data to map LULC changes 25 slaI1ed a earl) a land remote en ing imager) were made available in the 1970

(l luang. 2009) and ince then. many L LC change det tion technique ha\ 'e been dc\ cloped (Jen en. J 996):

• Po l Cia ification

• Write Function Memory

• 1ulti Dale ompo ite Image

• Image Igebra

• Binary 1a k

• mg ncillary Data

• lanual On creen Digitization

• pectral Change ector

• KnO\\ ledge Based

In thi study. \\'e will be fo llowing the post clas ification method to meet the et objectiYes. Before implementing the change detection analysis, several conditions must be met. according to Al Kuwari (201 1):

• Precise registration of multi-temporal images

• Precise radiometric and atmospheric calibration

• Similar phonological state between multi-temporal images

• Selection of the same spatial and spectral resolution images to produce high­

quality change detection results 26

2.5 ummary

Overall. up n examining various techniques and method previou ly u ed to identif} Ll'L of d rt em ironment using remote en ing. we can identify �hich method i the ea iest and mo t uitable to recreate oyer three ) ears. Therefore, we will b abJe to pr duce con tant re ults with minimum interference and en'ors. and to anal:ze the hange and me t the object set by this study.

\\'e \\ ill be u ing maximum likelihood to classify the image fo llowed by the po t clas ification change detection algorithm. With this, \ve should be able to track the mo\ ement of sand and tudy its change acro time. 27

Chapter 3: Re earch Methodology

3. 1 Introduction

fter re\ iewing the literature and previou re earch perfonned on dune and th II E. and to meet the requirement of the tud) . it was decided that the fo lJo",;ing

teps ", ould be taken (Figure 3. 1):

Figure 3.1: Dissertation methodology proce s

In short. we would create a mosaic from 6 scenes of the study years. have its vegetation extracted using SA VI, and then perfom1 the supervised classification using the scheme derived from the pilot study. Using higher resolution imagery. we would check fo r the accuracy of the classification before the post classification change analysis. 28

3.2 ' patial Data et & oftw are

hi chapter con ider a et of method fo r and dune cla ification, and

their re ult 0\ er the pilot tudy area. The be t method that meet the tudy objecti\ c i elected fo r impl mentation and is u ed to process the data fo r the entire

tudy area.

S1l1g a ene of space borne images from the Landsat series. thi section \\'ill look at de\ eloping a L LC thematic map that \ ill be us d in change detection calculations to d termine the direction of movement, whether the dunes are expanding or receding. and the change in tenns of area coverage over time. thus allowing u to predict what would occur in the future based on earlier ears.

The tudy require the use of ix scenes of imagery from the Landsat series using diffe rent ensor fo r extracting and dune from other land cover fe atures, to create a complete image fo r the three time frames that are 1992, 2002, and 2013. An

Abu Dhabi Emirate boundary shapefilewas used to generate an area of interest that

pans over both land and water and was implemented in Exelis Visual Information

y tem E VI 5.3 to ub et the study area fo r all three datasets.

sing a six scene mosaic of path 160 rows 43 and 44, path 161 rows 43 and

44. and path 162 rows 43 and 44 (Figure 3.2) fo r the years 1992, 2002, and 2013

(Appendix A Table 3. 1), the study area is covered fo r the region that lies from Al Ain

(eastern UAE-Oman border) to Bayah al Sila (western UAE- audi border) (Figure

3.3). A remote sensing image dataset covering the study area was produced fo r all future LULC classification. 29

[o r the oft\\are. thi tud) u E VI � .3. which wa u ed fo r all digital

Image pro e mg need . fo llov" ing by Rf s rc GI 10.3 fo r map production needs.

.,.,.. 3 ptS""3

......

Figure 3.2: Landsat Path & Rows Co ering the UAE

Study Area legend o

1:1,750,000

Figure 3.3: The Study Area of the Dissertation inside Red Boundary 30

3.3 Data preparation

s part of this di ertation. the data that \vill be u ed 10 the tudy \',as

ubjected to the fo lloY\,ing data preparation task :

.. tep 1: 'pload and tack the band of each cene into one u able file.

tep 2: Perf01111 a georeferenced 1110 aic proce that is upported through ENVr 5.3

to get a full image fo r ea h of the 1992. 2002. and 2013 datasets. The 1110 aic

proce s wa d ne u ing the " eamle Mosaic" tool in stages. since paths 161

and 162 would perfectl) mosaic with each other. as the have very imilar

land co, er features. However. path 160 has the inclusion of th mountain

regions in the Ea t. If all six cenes were to be performed at one shot. then

color balancing is ues would arise. That is why. paths 161 and 162 are

mosaicked firt. and their outcome is again mosaicked with path 160. teach

tage data ignore value are set and that is to remove the image boundary

from the picture. with color correction of histogram matching from overlap

were it would. and cuts were done along the seamlines that were

automatically generated by the system (Figure 3.4).

tep 3: sing the study area AOI that covers both the waters and the land of the

emirate. a subset , as created to fo cus the analysis on the areas of interest.

Furthermore. it should be mentioned that no radiometric correction was performed since the images used are already Surface Reflectance products from the

Landsat Ecosystem Disturbance Adaptive Processing ystem (LEDAP ) Climate

Data Record (CDR) repository that is provided by the USGS fo r Landsats 4. 7. and 8. 31

Figure 3.4: Representation of mosaic Seamlines

The oftware applies moderate resolution imaging spectroradiometer

(MODI ) atmospheric correction routines to level 1 Landsat TM and ETM+. Water vapor. ozone. geopotential height, aero 01 optical thickness, and digital elevation model (DEM) are inputted into the 6S (Second imulation of a Satellite Signal in the

olar pectrum) radiati e transfer models to generate the surface reflectance; this applies to the series from 4 to 7. As fo r Landsat 8, surface reflectance is generated using a specialized software called L8SR, that uses an intemal algorithm rather than the 6S, and its inputs come from either on-board calculations, such as geopotential height, or from MODIS CMA. such as water vapor and ozone (Table 3.2). 32

Parameter Land at 4 -5, 7 Land at 8 OLI (L8 R) (LEDAP )

Radiati\e tran fe r 6S Internal algorithm model

Thermal cOlTection TO only TOA on I) le\ el

Thermal band units K I in Kelvin

Pre ure NCEP Grid Surface pre sure i calculated internally ba ed on the ele ation

Water \apor NCEP Grid MODIS CMA

ir temperature NCEP Grid MODIS CMA

OEM Global Climate Model Global Climate Model OEM DEM

Ozone OMI/TOMS MODIS CMG Coarse re olution ozone

AOT Correlation between MODIS CMA ch lorophy II absorption and bound \\ ater absorption of cene

Sun angle Scene center from Scene center from input metadata input metadata

Vie\\ zen ith angle From input metadata Hard-coded to 0

Brightness Yes (Band 6 Yes (Bands 10 & 11 TlRS) temperature calcula ted TMlETM+)

Table 3.2: Comparison between CDR of Landsat 4 -7 and Landsat 8

3A ClassificationScheme

A classification scheme is a list of all potential land co er types that are pre ent in a study area that could be identified from a satellite image_ and this list

hould be comprehensive as to encompass all the cover. Therefore, a sound classification scheme should make sure that one cover should not fit into another 33 cia . Table ".3 i the mo t popular cherne dev i ed b) nder on (1976). and it i used b) the ..Geological un ey ( G)Land e / oyer classification.

Le\el I

Urban or built-up land

Agricultural land

Rangeland

Fore t land

Water

Wetland

Barren land

Tundra

Perennial no\-\ or Ice

Table 3.3: Themes as per Anderson 1976

However. due to some classes being none existent in the study area, uch as the perennial snow or ice, the list has been revised as fo llows (Table 3. -1)

Level l

Urban or built-up land

Agricultural land

Water

Wetland

Barren land

Table 3.4: Revised Anderson 1976 Themes as per Region of Study Area 34

Furthennore, to reduce the ize of the data to be cla ified and to improve the cIa . di 'crimination, a vegetation ma k VI \\lith a 0.5 factor was implemented to identil)' vegetation co\'cred area and exclude them from the cIas ification proces .

3.5 Dctcrminino L L cIa c

Before commi sioning the upervi ed maximum likelihood, it is imp rative to dctennine the cia ifiable L L of the Abu Dhabi Emirate. For that, unsupervi ed

Iterati\e If-Organizing Data naly i Technique (l ODA TA) classification was performed on three different data et : multispectraL Inverse PCA and Texture \ ith

fulti pectraJ.

Thi pilot study, to determine the cIa sifiable fe atures, was done on a single

cene. The representati e was from path 160 row 43 since it was the most diversified image CO\' ring water. urban, wet lands, desert, vegetation, and the mountains

(Figure 3 -).

Figure 3.5: Pilot Study Area Used to IdentifyCla sses 35

rhe fir ( data et te ted \\ a the tandard si multi p ctral band (Figure 3. -I).

Tht: second data et was the im er e PC , v. here thi data et \\ a deriv d b) performing the fO f\\ ard principle components (PC rotation on the multi pectral band t produce uncolTeJaled output band that allowed the segregation of the noise comp nent in th original image tlm r ducing the dimensionality of the dataset.

fter identif). ing the mo t informati\' 3 Pc, inver e PC rotation was performed using the tali tic from tJl fo nvard PC rotation. therefore, giving u an image \\ ith reduced noi e (Figures 3. 6 & 3. � Clnd Appendix A Ta bles 3.5 & 3.6).

Forward peA

Legend FOIWattI PCA RG8

1:600,000 0510 £ XI

, Oe-.lt!d., �"A.� �� 'hGSl., I./TIoII ZOtW'JN 'OM,,"" D_�_l"'"

a.tf' l7.m.,mlS ® ______. " L-__ � �

Figure 3.6: Forward PCA of Pilot Study in First Three Principal Components (PCs) 36

Inverse peA

Legend lnvern peA oca - - - ..... _ ...... ,

1:600,000

Figure 3.7: Inver e PC of the First Three PCs in True Color ROB

The third data et that got tested was texture (Figure 3. ) and multispectral

(Figure 3.5) datasets combined into a single set. The texture was derived from the entropy of the pilot image that showed the areas of heterogeneity (areas that are texturally diffe rent) thus allowing to further enhance the multispectral image by adding the texture of the features as an add-in into the image. 37

(

Entropy

1:600,000 () S 10 .. """ � .... '*.a..... � � s,u.... 1-'& Z-'lJlj ' 0..- DWQ1" � , ,.U I �------�----�

Figure 3.8: Entropy Texture of Pilot tudy Area

The parameter that \vere u ed are as fo llows:

• The study area has no less than 30 and no more than 300 diffe rent classe

• The minimum area fo r each cla to be considered as such is one km2 (i.e.

11. 111 pixels)

• The clustering will nm to a maximum of 100 iterations

Upon that, it was noted that there is a significant diffe rence between the number of spectral classes between the multi spectral (31 Classes), texture + multi pectral (41 Classes), and the PCA (295 Classes) (F;gure 3. 9). Then to reduce' the number of fe atures classes. class merging was performed using high-resolution imager) from IKO OS 2006 and RapidEye 2012 to confim1 in ariant sites. 38

ISODATA Results for Pilot Study

1 :1,324,467

Figure 3.9: I ODATA Results fo r Pilot Study

3.6 Image Cia sification & Po t Processing

pon determining the possible stati tical classes, the implementation of the

uperVI ed cla si ficationwas undertaken, llsing 142 training site . Class merging wa performed and each training site was merged into one of the 4 classes below:

• and

• Wet oil

• Intertidal Zone

• Exposed Bedrock

After which, class Sle e was performed to sol e the problem of ha ing

isolated pixels that occurred after the image classification process. Sieving removes

isolated pixels using blob grouping of 4 neighbors to determine if a pixel is grouped 39

\\ ith pixel of the ame cIa s. if not. that pixel will be removed and cia died a

"uncIa ifi d .. ·

Then. to till the ne\\ ly created "uncla ified" cla . clump wa perfom1ed to clump adjacent imilar cIa iiied area tog ther using morphological operators. The

la e were clumped together. fi rst, by performing a dilate operation. fo llowed by, an erode operation on th cla i fied image using a kernel size of 3 by 3.

3.7 Extracting and dUDe from atellite image (MSS 1992, ETM+ 2002, OLI

2013)

The econd main objecti\e of this tudy is to extract the and dunes from the three image data et to analyze the dune movement and change across time. using per pixel maximum likelihood supervi ed classification algorithm to extract the dune from the re t of the land cover features. Thereafter, sand fe atures are extracted and tran fo rmed into binary map . resulting in sand and non-sand maps.

3.8 Accuracy A se ment

Accuracy as essrnent is the next step taken to determine the thematic map's cia itlcation accuracy when compared to accuracy assessment sites. which are sites collected separately from the training sites and used fo r the classificationprocess. In this study, the accuracy a e sment sites will be stationary aero s all three years and will be selected using higher resolution imagery than that of Landsat (Figure 3. 10 &

Appendix A Table 3. l.

To do so. the study uses a RapidEye fo r the year 2012, and IKO OS fo r the year 2006. and Spot fo r the year 1986. each with 5. L and lO-meter resolution. respectively (Appendix A Ta ble 3.8). These dates will be used to identify the 40 im arianl aCCllra a e nlenl sl'te that \- \\ I' ll be llsed t 0 mea ure owh we 11 t h e cIa ification wa p rformed.

Distribution of Groud Truth Sites

Legend

. . . -... . • .., o • . .

• . . . . • 0 . .

1:1 ,250,000

0;",c,,"ii::52:,;.5 _,;.:50=::i75___ '� lers (...... :1 .. ...,.. .A .... �� WGl t "' lI�l... ,'J� � D.IIiIII'GS_lJII..t OM. 1�'(w'-20.15

Figure 3.10: Di tribution of accuracy assessment sites

There are four t pes of accuracy that are measured in the confusion matrix cal culati on:

U er' Accuracy : is the error of commission. It measures the probability that a pixel is clas "k' given that the classifierhas labeled the pi el into clas "A". It is calculated by the division of the total number of conectly classified pixels of a certain class by the total number of pixels that were classified in that same class.

Producer's Accuracy: is also referred to as the error of omISSion. It measures the probability that the classifierla beled an image into class "A" given that the accuracy assessment site is class "A". It is calculated by the division of the total 41 number f c rreetl) ela ified pixel f a certain cia es b) the number of total pixel that actuall) belong t that am cia

Overall Accur'acy : i calculated by umming the total number of pix Is that arc la ilied properl), and divid d b) the total number of pixel

Kappa Coefficient: m a ure the agreement between classification and

'' urac} ' acc a e ment pixel . It produce an inde . ranging from 0 to L where ' 1 repre ent pcrD ct agreement. while "0" represents no agreement.

Where:

i = cla number

= the total number of classified pixels that are being compared to the accuracy a se ment data

ml.l = the number of pixel that belong to the accuracy assessment class in "i ' that are

al 0 cla ified'wit h class "i"

C = the total number of classifiedpixels that belong to "i"

G1 = the total number of accuracy assessment pixels that belong to "i"

3.9 Dune change analysis techniq ues

The post classification change analysis was performed using the fl owchart in

Figure 3.11. After performing the supervised classification algoritlml and verifying its accuracy, the third step was to vectorize the raster image into polygons that held 42 the cia ified fe atur information. ing the union overla) tool from ArcGI 10.3. we \\ cre able to get the fo llo\\ ing from-to change (Table 3.9) that congregated the three stage 0 r the tud) into on dataset. Then. in order to map the from-to changes acro the three year . an additional field was added and named "From_To" \\ith the

1'011 \\ ing calculation [ las _ ame_ 1 99:?] &" -"& [Cia _ ame_200:?'] & 11 -"&

[Ia ame_20 13] in order to anange the cIa s names in order of 1992 to 200:? to

2013. fo llo\\ ed by the dissolve in order to remove trajectory duplications. The final re ult in Table 3.10 could be een and mapped.

Multi Year Change Detection Flowchart

Figure 3.1 1: Multi Year change detection methodology 43

flO DIssolve 1992 Class Hame 1992 AD Dissolve 2002 Class_Name 2002 flO Dissolve lOll Class N1IITle 2013 1 VeqellltJon , VegelatJon 1 VegetatlOo 2 Open ..•.. ster , Vegetahon 1 VegetatIOn 3 Sand , VegetatIOn , VegetaJon I I • ExposedRod 1 VegetalJon 1 VegetatDn I 5 I'1lertJdal 1 Veget8tDn , VegetatJDn I 6 111.'e1Sol 1 Veget8tJon , VegetatJDn 1 VegelBtJon 1 VegetatJon 2 I Open IVater 2 Open \Vater 1 VegetatJon 2 I Open Water 4 ExposedRod 1 Vegetalton 2 Open W.'er 5 nlertJdal 1 Vegetabon 2 Open Water 6 I ...et Sol , VegetatJon 2 OpenWater 1 VegetatDn 1 VegetatIOn 3 Sand 2 Open Water , Vegelalton 3 Sand 3 Sand 1 VegetatIon 3 Sand 4 ExposedRod 1 Vegetabon 3 Sand � ntertJdal 1 VegetatIOn 3 Sand 6 Wet Sol 1 VegetatIon 3 Sand 1 Vegetabon 1 VegetatIOn 4 ExposedRock 2 Open Vlater I VegetatIOn 4 Exposed Rock 3 Sand I Vegetation 4 ExposedRock 4 Exposed Rock I VegetatIOn 4 Expo�ed Ro cl. S ntertld81 1 Vegetahon � ExposedRod 6 Wel Soi I VegetatJon 4 Exposed Rod I 1 Vegetabon I Vegetabon S IntertJdal I '"' __f" ... '... '_6 ...... "...... C I.... � ...... I: _I

Table 3.9: Union of three years

From To Exposed Rock - Expose

Exposed Rock - Exposed Rock - Open Water

Exposed Rock - Exposed Rock - Sand Exposed Rock: - Exposed Rock - Vegetation

Exposed Rock - Exposed Rock - 'Net Soil

Exposed Rock -Intertidal-Exposed Rock

Exposed Rock - Intertidal - Intertidal Exposed Rock - Intertidal - Open \\fater

Exposed Rock - Intertidal - Sand Exposed Rock - Intertidal-Vegetation

Exposed Rock - Intertidal - \j'Vet Soil Exposed Rock - Open Water - Exposed Ro ck

Exposed Rock - 0 pen Water - Intertidal Exposed Rock - Open V.raler- Open INater

Exposed Rock - Open Water - Sand Exposed Rock - Open Water - Vegetation

Exposed Rock - Open Water - Wet Soil

Exposed Rock - Sand - Exposed Rock

Exposed Rock - Sand - Intertldal

Exposed Rock - Sand - Sand Exposed Rock - Sand - Vegetation

Exposed Rock - Sand - \Vet Soil

Exposed Rock - Vegetation - Expo sed Rock 1 c __ ".. ",,",.. ,,4 n ...t, ..... \I.... -...... �... 1 ..... _..,&.-�,., ...

Table 3.10: Final Product of from - to change across three years 44

Thi tud;. produc d a total of 31 trajectories of possible from-to change bet\\ een t\1" O dates, and they are identified a could be een below (Table 3. 1l).

From Year To Year 2 1 ') Clas 1 Cia Cia 3 Clas -+ Class 5 Cia 6

CIa I No Change

Cia 2 No Change

.., Cia .:> No Change

Cia s-+ No Change

" CIa No Change

Cia 6 No Change

Table 3.1 1: Change trajectorie between two years

When the classes are merged to represent the feature of interest sand to non- sand. we obtain the fo llowing three trajectories of possible from-to change across two dates. and they are identified as seen in the table below (Table 3. 12).

From Year 1 To Year 2

Sand None Sand

Sand No Change

None Sand No Change

Table 3.12: Change trajectories fo r sand and non-sand between two years 45

Ho\\c\ cr. overall. there ar 212 po ible from-to traje torie aero all three year of the tud)- (Fif;ure 3. 1]).

Year 1 ' Y,ear3

Figure 3.12: Change trajectories across three years

When looking at and and non- and changes. we are leftwith seven across the entire three year of the tudy (Table 3. 13).

From Year 1 to Year 2 To Year 3

Sand None Sand

Sand - Sand No Change

Sand - None Sand

None Sand - Sand

None Sand - None Sand No Change

Table 3.13: Change trajectories fo r sand and non-sand across three years 46

Chapter 4: Re ult & Di cu ion

.t. 1 Training Site Cia Identification

een in the pre\ i u chapter, ection 3.6. we \\ ere attempting to identify the tati tical!) cIa ifiable land covers through the u e of I ODAT un upervi ed cIa ification on fo ur diffe rent data ets (I ODA T of entropy. Multi pectral \-\ith high pa filter. multispectral, and PCA) (Figure 3. 9).

I Io\\,ever. after the po t cia sification proce s of merging the cIa se into their parent cia a per the grouping ite of Figure -1. 1. the outcome was the same a in

Figure -1. 1. It became apparent that I ODA T A performed on the fo ur diffe rent data t. though produced diffe rent number of classes in terms of output, still recognized the same features. Statistically speaking, I ODAT A is unable to diffe rentiat between Wetlands. Urban. and the Expo ed Bedrock fe atur , which indicate that confusion will occur when attempting the supervised clas ification. 47

legend

Grouping Area c;-.N.....

0 ..... o 1998 Poiol Study .GO - _ r- ...., - _ .... .

1 ;100,000 DO 5 t 2' � .. " �...... A� ----, � 1-" V"1roII1 _ _ ...

Figure 4.1: Final Outcome after Merge of Classes

-

- - ­ . .-

- -� :... ------

Final Output of Same Feature Merging of ISODATA Resutls

1;600,000

Figure 4.2: Final Outcome afterMerge of Classes 48

4.2 raining , ite ' elect ion

Ba oed on the finding of the I OD T coupl d with A I mask created from xperimenting \\ ith \ ari 11 thre hold . it \\ a dete1111ined that -1 to -0.2 produced the be t re lilts in xtracting \ egetation acros all three years . Thu . Ta ble

-I. J bov, the cIa e that will b II ed in thi tlld) .

Cia se

Vegetat i n

Surface Water

Sand

Exp ed Bedrock

Intertidal

Wet Soil

Table 4.l: tati tical Classifiable Classes

1-1-2 invariant training sites that have not changed, were elected across the three year (1992. 2002. and 2013), so a to better represent each year when compared to the other years. ite that will be used in the supervised classification algorithm were te ted fo r their lass separability. to ensure that the selected sites do not get confu ed w-ith one another statistically.

The leffreys-Matusita Distance results fo r the 1992 dataset concluded that the first conflict bel:\Veen two classes was at 1.938, between class Sand 38 and Wet Soil

. As fo r the 2002 dataset. the first conflict between two classes was at 1.897, between Sand 38 and Wet Soil 10. As fo r the final dataset. 2013. it was concluded that the first conflict between two classes appears at l.980, between Sand J 07 and

Expor;edBed rock 11. 49

are ult. bet\\een all 142 training ite ( and. Wet oil. Intertidal. Expo ed

Bedrock). the 1M te t proved that there vvas no con iderable clas confu ion. and up n fu rther examination it v. a detem1ined that the cla ses were homogeneou ; therefore the u e of the Maximum Likelihood algorithm v, ould be uitable.

·t3 1992 LULC Map

ppl) ing the maximum likelihood along \ ith A VI mask. we produced 2 re ult that ho\\ ed the area in km fo r the initial tage of this study (Table -1. 2) and two thematic maps: on being all cIasse as per cherne (Figure -1. 3) and the other being and \ non- and fe ature (Figure -I.-/).

elas Name Class Area (Km-)

Vegetation 146.53

Surface Water 321 72.23

Sand 521 86.27

Expo ed Bedrock 1874.55

Inte11idal 3069.62

Wet Soil 5168.40

Table 4.2: 1992 Feature Class Areas 50

ClassIfication of 1992 landsat TM Image

legend

1992� •• lfoed Image Feature

-

1:1 ,600,000

Figure 4.3: Thematic Map of 1992

Merged Classification of 1992 landsat TM Image

legend 1992 Merged Features Future ....

1:1,600,000 .tOlD 40 ID 10 --

0-00 .. �",..... 1_"""", :-'" o_� � ,'... :zan

Figure 4.4: 1992 Thematic Map Showing and & Non-Sand Classes 51

-tA2002 LULC Map

. ing the ame methodology a \\ ith the pre\ iou year. the 1992 data et. and the tud) ho\\ change around the coa tal area v, here and ha expanded. either due to reclamation r due to the change in en or that it re-cla sed covers in land. Table

4.3 sho\\ a significant in rea e in \ egetation, a decrease in wet oils. and ome change in and and intertidal. Tvv o thematic maps are produced as well; one being all cia ses a per cheme (Figure -1. 5) and the other bei ng sand vs. non-sand fe atures

(Figure -1. 6).

Class Name Cia Area (Km-)

Vegetation 518.39

Surfa ce Water 32873 .45

Sand 5273 1.71

Expo ed Bedrock 1863.74

Intertidal 2788.27

Wet Soil 4341 .34

Table 4.3: 2002 Feature Class Areas 52

Classificallon of 2002 Landsat ETM+ Image Legend 2002 Classified Image

-\.

1:1 ,600,000

Figur 4.5 - Thematic Map 0[2002

Merged Classification of 2002 Landsat ETM+ Image

Legend 2002 Merged Features F.�lIIu", Soood _ .....

1:1 ,600,000

Figure 4.6: 2002 Thematic Map Showing Sand & on-Sand Classes 53

....52 013 L L Map

In 20 13. fu rther land reclamation could be seen towards the coa t. as well

urban expan i 11. and a decrea e in egelation cover. Table 4.4 shows a significant increase in expo ed bedrock. a decrease in wet oil . and some changes in sand and int rtidal. Two thematic maps are produced; one being all cla ses a per scheme

.J.�) (Figure and the other being sand v . non-sand fe ature (Figure .J. 8).

CIa s Name Clas Area (Km")

Veg talion 439.78

Surface Water 32678.95

Sand 52848.49

Exposed Bedrock 3821.13

Intertidal 3174.64

Wet Soil 2650.98

Table 4.4: 2013 Feature Class Areas S4

Classific.atlon of 20 13 Landsat OLi lmage Legend

2013Cl assified Image

.... .

"

1 :1.600,000

Figure 4.7: Thematic Map of2013

Merged Classification of 2013 Landsat OLi lmage

Legend 2013 Merged Features F ••tu,. Sand _Sand

1:1 ,600,000 01020 40 10 II) -...... OWW .. "'A.� "--" s.- WQl I'tto''''''''' w.. �'" - 0- lH"m.S

Figure 4.8: 2013 Thematic Map Showing Sand & on-Sand Classes 55

-t.6 Accurac) A e ment

Follo\, ing \.\ hat len en C :W05) tated regarding accuracy a es ment. all)

Kappa \ alue more than 80 % \\ ill be con idered o[ good cla ification. value

0 bet \\ een 40 and 0 0 m derat . and anything les than 40 0 0 is of poor performance.

fh rcfi r . when the confusion matrix was performed on all thre year . it wa conclud d that fo r the )ear 1992. a moderate performance at 76.990'0 was achieved

(Table -1.5). \-\ here the main confu ion occurred between exposed bedrock and sand or \\ et ,oil. While [o r the year 2002. moderate performance \, as again achieved at

78.290 0 (Table -1.6) \\ here confusi ns fo r exposed bedrock with sand or wet soil increa ed. fu rthermore. in thi year wet soil was misclas ified with sand, inte11idaL and expo ed bedrock. A fo r 2013, good performance was achieved at 81.14%

(Table -I. ") where the sanle eITor betvveen exposed bedrock and sand or wet soil remained.

0, erall ACCillaC) = 87.81 %

Kappa Coefficient = 76.99 % Ground Truth (Percent) Cia Sand Wet Intertidal Exposed Surface Vegetation Total

Soil Bedrock Water

Sand 90.98 0 0 23.02 0 0 64.23

Wet Soil 2.88 100 0 25.68 0 0 17.56

Intertidal 0 0 100 0 0 0 1.76

Exposed 6. 14 0 0 51.30 0 0 10.56 56

Bedrock

urface 0 0 0 0 100 0 4.59

v.. ater

Vegetation 0 0 0 0 0 100 1.29

Cia Commis ion Omi Ion Producers Accl.lrac) User Accuracy

Sand 4.49 9.02 90.98 95.5 1

Wet Soil '29.37 0 100 70.63

I ntert idal 0 0 100 100

Expo ed Bedrock 39.21 48.70 51.30 60.79

Surface Water 0 0 100 100

Vegetation 0 0 100 100

Table 4.5: Accuracy ssessment of 1992 Classification

Overall Accuracy = 89.73 %

Kappa Coefficient = 78.29 % Ground Truth (Percent) Cia s Sand Wet Intertidal Expo ed Surface Vegetation Total

Soi l Bedrock Water

Sand 99.06 9.76 0 44.60 0 0 73.89

Wet Soil 0.94 80.59 0 16.95 0 0 12.68 57

I Intert idal 0 9.43 100 0 0 0 2.94

I.:\.p sed 0 0. 18 0 38.44 0 0 4.52 Bedroch.

Surface 0 0 0 0 100 0 4.65

v.. ater

Vegetation 0 0 0 0 0 100 1.3 I

Cia Commis ion Omi Ion Producer Accuracy User Accuracy

Sand 8.72 0.94 99.06 91.28

Wet Soil 20.69 19.4 1 80.59 79.3 1

Intertidal 39.95 0 100 60.05

b.posed B drock 0.5 1 61.56 38.44 99.49

Surface Water 0 0 100 100

Vegetation 0 0 100 100

Table 4.6: Accuracy Assessment of 2002 Classification

Overall Accuracy = 91.06 %

Kappa Coefficient = 81.14 % Ground TlUth (Percent) Clas Sand Wet Intertidal Exposed Surface Vegetation Total

oil Bedrock Water

Sand 99.75 0 0 56.15 0 0 74.38

Wet Soil 0.06 85.67 0.85 0. 12 0 0 10.63 58

Intertidal 0 9.97 99. 15 0 0 0 2.99

E:--po ed 0.19 4.37 0 43.73 0 0 6.10

Bedrock.

Surface 0 0 0 0 100 0 4.60

v.. ater

Vegetation 0 0 0 0 0 100 1.30

Clas COll1mi Ion Omi sion Producers Accuracy User Accuracy

Sand 9.38 0.25 99.75 90.62

Wet Soil 0.64 14.33 85.67 99.36

Intertidal 41.01 0.85 99. 15 58.99

Exposed Bedrock. 10.94 56.27 43.73 89.06

Surface Water 0 0 100 100

Vegetation 0 0 100 100

Table 4.7: ccuracy Assessment of 20 13 Classification

·t 7 Change Analysis

As discussed in the previous chapter. change analysis was perfonned across the three datasets 1992. 2002. and 2013 using post classification change anal sis. Map & stati tical results of change between J 992 - 2002 could be seen in Appendix A Table

-1.8 & -1. 9 & Figure -1.9. -1. 10, and -1. 11. As fo r the results. between 2002 - 2013. both

-1. 12, map and statistics could be seen in Appendix A Ta ble -1. 10& -1. 11 & Figure

-1. 13. and -1. 1-1. Abu Dhabi Change Trajectory Between 1992 and 2002 Legend

From .. To v.g.UIIO'I f.powd �oOI, NoCh.� V.ltt!OI"I , ln'''''�1 E.pOMdRoc::t. .. Inlflftd., W9t1t.von ..()pefl WiI''" o L.poMa Rod• • ap." WlI.' vt1J ....lJOO ..Sind E.� Roc� ..S."d _ _ \l�I.!oon - � SOII E -I'O'M<' Rock.- V�t'IGn �I 5011.. £ lpoMdRoo f.powdRock .. VVtt Sa.I Willi Sod � Imef1oda1

Inle"",.1 E.� Roca _ w.1Soe1.Op.fI...... IM h'lle(\)d.1 � W.t., W!:I Sotl .. S.nd Inl.fUelal $af1d WitISoel .....w hon _ 'n'.""'" "." 1101\ ", _ Inlf'�t WeISoil

()peoWlt., ..l�poMCI RoO:

_ Optnwa,ltl' . lnlM"'ill

OPen"NI lff " S'too � Wit., · Ve9IJtiJUOO '\ _ ep.n 'Water - 't'Vtl Soli

�I'IG ['OOHd('('d " PrOjeCI.on SV\I('m wGS 1984 UTM Zont: 9)N " .6)/\ O"um O_WGS, 1984 O,te 11 09 lOIS \(C3l ' Figure 4,9: Change Trajectory Map fro m 1992 to 2002

59 Abu Dhabi Sand Change Trajectory Between 1992 and 2002

Legend

From - To

NoC",,"9'I

_ blJOM(J Roc" S.nd _ 'nltrtlde' Sood . OI)4W'lINaJII!( gMd :--..,-.. '. --_ ... SMId ElIpoMdR �� ...... ""'" .,)-""", _ s.no Intf!ft!d.til -:� ��� ' ...... ,.,! , Sand QoenVt.ltf !' , :!:" Sand VtoetlloOn �h:� �4Ot . � SwIG Wtt SOt! /� , Vegtt.llOt't 5.tnd ...... l� � :, .� .-�- -.� _ Wtt Sool 5""" 1:1 ,600,000 o '0 20 40 60 80 • • Kllomelen; , (fflle:d By Raml WA Sart"d Pfoje

60 Abu Dhabi Sand to Non-Sand Trajecyory Between 1992 - 2002

Legend

No Change

Non Sind Sand

Sanc:J Non Sand 1:1 ,600,000 o 15 30 60 90 120 _ _ Kilometers Crfiu�d By RamI W /it S,}t"� PrOltcUon �Y1lcm WGS 1984 U1M Zoot" 91N �$I Datum 0_ WGS. P�84 Date 1709 10IS Figure 4.1 1: Change Trajectory Map of Sand from 1992 to 2002

61 Abu Dhabi ChangeTrajectory Between 2002 and 201 3 Legend

From · To VtOflJltoan bp(JMd Roc:lrI

NoCNliI"lQ* V�I.loon · lnl"rt� _ hpoMd R� II. ''''!1tfl,cs.I VI9I"",gn • Oprn WIlle' f.PQIfk.I R, Optn'NAt., Vf9f'.I'./M • Sand o b,1OMIj Rock s.oct . VeQreLlfJOn ""-'SoiI e.QOWd Rot... IJwge'I.toOn • � SOII r.poMCI Rort. t . r (poMt1 Roclo, 'NIl' &oil WIH SOlI Intltf1JOtf Inl�f1IG.1 e ..PQHd Roc... • Wtl Sol apc.n wat,,! lnlttrtld.1 Op4n w., _ _ � SoI � S.nd ., lnl",ntClal SttncJ _ Wltt Solt"eqtllIIloon _ 11'I�"'cUl �.hon Inlttr1tCll!I ...... , $011

Op.tn..... ,I!" E llDOt*d R()(:�

Open WIIIf1' InI�nldftl

�nWltet 'i.tnd

()pen\Valet VegM.tt.,.,

_ ep."'9f. vv,, We, Sotl , _ S.nd ExpandR()('A ,�' _ S.nd Inl./'lIdI1 �, Sand 0pItnWalt" -.--: � Sind.. VtOtl8 111Of'1 ,,-..... - - 5ancI Wflt SOlI ...... ;..... �, : �...... , 1:1,600,000 o 10 20 40 60 80 KIlometers (tfllt>d 8"1 �a/TII W A Sa�d " PrOI«lK>f'l Syu('m WGS 1984 UTM Zone!' 93N �®I O.lUm to0 WGS 1984 0". 1-2;0911m , Figure 4.12: Change Trajectory Map from 2002 to 20 13

62 Abu Dhabi Sand Change Trajectory Between 2002 and 201 3

Legend I' � From - To �

NoCh�

_ t,po� Rod" 4So\nd In'.f1ldal SallO -\��;,�, Opon WIo'.. . SoncI , r)' \ SanCl ElDOMdRock .' ;;� �. Sand Inl�l " -.- ""! .. .;:,. . � _ S.ncs ()penWaitt' \ ...... f . _ Sand VeQellloOn ....'<} �...... Sonmeters Cr.aled By Rami W A Saeed " , P,ojt

63 Abu Dhabi Sand to Non-Sand Trajecyory Between 2002 - 2013

Legend

No Change

Non Sand Sand

Sand Non Sand 1:1 ,600,000 o 15 30 60 90 120 _ _ KIlometers (ruled By R.aml W A ��d , pcoJe 1 27..()9-201S Figure 4.14: Change Trajectory Map of Sand [ro m 1992 to 2002

64 When considering the overall change across the three years. Figure '; /5 and Figure -1. /6 best depict these changes:

Abu Dhabi Change Trajectory Between 1992 - 2002 - 201 3

,

�.... .- .. .:"" "-:,.' •.. . -'� ..

....r , ·1· " 'Y;,:. ... . -: ..�� S___ 1:1 ,600,000 '· :;, o .0 20 40 60 110 .. , 'ornet Ct."f�d BV . R..Jml W iii \,}if"td " r'otNtl()n �Y"M'm,t-- WG"J lfJlJ,4 U1M lU'f'<' 11JN " 1 + O"t",,,,, 0 WGS $ lq84 t· . O"I� 11·01)·201') Figure 4.15: Change trajectory [rom 1992 to 2002 to 20 13 65 Legend

Ch.n� 'tn.1

... � ...... -

.. t.,..... • • • .. ..--- ...... i_._ ...., ......

-.�--

� �-'Il..ao._ " '�l___ - - - � _f__ - ...... ,.. , ...... -­ __.l ._'" I...... - - .. - Figure 4. 16: Total of 203 change trajectories from 1992 to 2002 to 20 13

66 Since this study fo cuses on the movemcnt and changes in sand, the resultants were extracted fo r the fe ature of interest and the changes can he seen in Figure -I. /7 & Figure -1. 18. Figllre -1. 19 better defines a clearcr understanding of sand / non-sand interactions and statistical area change fo r sand throughout the three years or the study could be seen i.nAppendix A Tahle -I. J 2.

Abu Dhabi Sand Change Trajectory Between 1992 - 2002 - 2013

1:1,600,000 0;.:,'0.. :.;."",,"' '<::0 =_OO ..;eo'KIkJme�f'ft ... Cr• .urC a., FUm, W 11 �r!"d PfOlM:II()I't \Y'It"m WC.!. lQU UIM 1M" �nli\l L5l/\ O'\�f'I'I D. WG�� 19U \\ \({Sl ' 17:0-,·101 J 011.. , Figure 4.17: Changes in sand trajectories from 1992 to 2002 to 2013

67 Legend

From_To Open Waler - Sand - Wet Soli Sand - Vegetation - Wet Soli No Change Open Water - Vegetation - Sand Sand - Wet SOil - Exposed Rock Exposed Rock - Exposed Rock - Sand Open Water - Wet SOil - Sand Sand - Wet SOil · Intertidal Exposed Rock - Intertidal - Sand Sand - Exposed Rock - Exposed Rock Sand - Wet SOil - Open Water Exposed Rock - Open Water - Sand Sand - Exposed Rock - Intertidal Sand - Wet SOil - Sand Exposed Rock - Sand - Exposed Rock Sand - Exposed Rock · Open Water Sand - Wet Soil - Vegetalion Exposed Rock - Sand - Intertidal Sand - Exposed Rock - Sand Sand - Wet SOil - Wet SOil Exposed Rock - Sand - Sand Sand - Exposed Rock - Vegetation Vegetation - Exposed Rock - Sand Exposed Rock - Sand - Vegetation Sand - Exposed Rock - Wet Soli Veg etation - Intertidal - Sand Exposed Rock - Sand - Wet SOil Sand - Intertidal - Exposed Rock Vegetation - Open Water - Sand Exposed Rock - Vegetation - Sand Sand - Intertidal - Intertidal Vegetation - Sand - Exposed Rock Exposed Rock - Wet Soil - Sand Sand - Intertidal - Open Water Vegetation - Sand - Intertidal Intertidal - Exposed Rock - Sand Sand - Intertidal - Sand Vegetation - Sand - Sand Intertidal - Intertidal - Sand Sand - Intertidal - Vegetation Vegetation - Sand - Vegetation Intertidal - Open Water - Sand Sand - Intertidal - Wet SOil Vegetation - Sand - Wet SOil Intertidal - Sand - Exposed Rock Sand - Open Water - Exposed Rock Vegetation - Vegetation - Sand Intertidal - Sand - Intertidal Sand - Open Water - Intertidal Vegetation - Wet SOil - Sand Intertidal - Sand - Open Water Sand - Open Water - Open Water Wet SOIl - Exposed Rock - Sand

Intertidal - Sand - Sand Sand - Open Water - Sand Wet Soli - Intertidal - Sand

Intertidal - Sand - Vegetation Sand - Open Water - Wet SOil Wet Soli - Open Water - Sand - Sand - Wet Soli Intertidal Sand - Sand - Exposed Rock Wet SOil - Sand - Exposed Rock Intertidal - Vegetation - Sand Sand - Sand - Intertidal Wet Soli - Sand - Intertidal

Intertidal - Wet SOil - Sand Sand - Sand - Open Water Wet Soil - Sand - Open Water

Open Water - Exposed Rock - Sand Sand - Sand - Vegetation Wet SOil - Sand - Sand

Open Water - Intertidal - Sand Sand - Sand - Wet Soil Wet SOil - Sand - Vegetation Open Water - Sand - Exposed Rock Sand - Vegetation - Exposed Rock Wet SOil - Sand - Wet Soli

Open Water - Sand - Intertidal Sand - Vegetation - Intertidal Wet Soil - Vegetation - Sand

Open Water - Sand - Sand Sand - Vegetalion - Sand Wet SOil - Wet Soil - Sand

Open Water - Sand - Vegetation Sand - Vegetation - Vegetation Figure 4. 18: To tal o[ 8S change trajectories [o r sand from 1992 to 2002 to 20 13

68 Abu Dhabi Sand to Non-Sand Trajecyory Between 1992 - 2002 - 2013

Legend

� CPlA/'IQ't

/'Iton S..-wlI ""on S.fId 5,1'1(1

t.-;w'II'ef"d $..a N"r\S","4 \

� SiII"od Send $end

_ F-tM NorIS,rod NOf>SiIIIl'Id J _�

SIno Non IW'CISand

s..nd, s.nc. I'IIf)fI S,1'!d

1:1,600,000 o 1� �n 60 90 120 _ _ Kilometers Crf>'ted BV Rilmi WA Sclt"t'd , ProJ�'lon 'Synt"m WGS 1984 UTM Zone>9]N Collum 0_ WGS 1984 "($' D.le I 11...()9 lOH Figure 4. ] 9: Change Trajectory Map of Sand / Non Sand from 1992 to 2002 to 20 13

69 70

4.8 Oi cu ion

rhere are apparent and dra tic change betwe n the three ) ear u ed in thi

tud) . due to t\'.o fa ctors: the di rference between the platforms and the OC IO- economic gro\\1h of the emirate.

hange atlribut d to the diffe rence in plat� rm could be best seen in Figure

-1. 20 and Figure -1. 2 J \\ here area cia sified a exposed bedrock in the year 1992 become cla ified a and in 2002, although the fe ature did not change temporally

\\ hen checked against higher resolution imagery used in this study.

Classific:atJon of 1992 undsat MSS Image Legend

1992 Clou,fiedImage Fe.tu...

�"

. -. ' • "

1:400,000 , , @

Figure 4.20: Features Classed as on-Sand 71

Classification of 2002 Landsat ETM+ Image Legend 2002 CI.. alfied I."."" Feature

------_ 0.-"" _ 0.- __

1;�0.000

Figure -+.21: Feature Classed as and

The ocioeconomic growth of the Abu Dhabi Emirate is a re ult of a never ending continuou proce of urban and social development that the nation has vowed to undertake to place them as fo rerunners in the global theater. For the past

-to odd year . the emirate transformed itself from being confined to specific geographical areas. to urban developments across the wide spread desert. all part of a plan to exploit unused land to enhance the emirate' s portfolio further. There are

many drivers that caused such changes to occur 0 er time. these drivers could be categorized as fo 11o\ \." :

• Geographic

The areas that witnessed the mo t change are those surrounding the

coast. Waterfront properties have always been a selling point, coupled with

the fact that the emirate plans to play a more predominant role in orId

tourism as a holiday and recreation destination regions surrounding the coast 72

ha\ e een continuou land reclamation. �eanwhile. a a re sult of inland

dredging to create artificial canal , other area \\ ere being transfonned to wet

land du to th change in the \-\- ater table (Figure -+. 22 & -1. 23 ).

Geographic Growth 1992

Legend

1992ClasSIfied Imag

P.

. . .. ' 1.1 30,000

Figure 4.22: Coastal area in 1992 73

Geographic Growth 2013

Legend 2013C l aSSlr.. d Image FUN,.

...- _ E..-.. _ - "'*'- - "-",-

1.130,000

. , - \ .® \',

Figure 4.23: Coastal areas in 2013 since 1992

• OClOe OnOmiC

ince the discover of oiL the emirate has undergone large-scale

modernization that re ulted in the increase of infrastructures and facilities to

cater to the ever increasing desire fo r better service and social outlets. By

that. the increase of vegetation could be seen not just in farm lands the

government ubsidized to the natives. but also in parks, recreational spaces.

international golf cour es and green paces around government, public. and

private structures. \vhich contributed to the overall transformation of the

de ert into a lush green garden. Since all these new developments required

empty spaces. much of the desert was transformed into an interconnected city

(Figures -1.2-1& -1.25). 74

Sand Coastal Expansion 1992

Legend 1992CI��lfied "mge Future ....,

- - --- _ Ooon '_ 00Me ___

J �.

1:150,000

Figure 4.24: ate of coastal areas in 1992

Legend 2013Class,fied Image Futu-re $0... -.as -- - ...--- _ 0-- _ 0... -

1:1 50,000

I , • . , - , ...... A ,... -- �1"1iTW:-' , -- o_wca_... .@ - p*,ou

Figure 4.25: Coastal change in 2013 since1 992 75

• Population

The emirate a"" a bo m in it population in recent 'ears. a there ha

been an inc rea In life expectanc and a decrease in child mortalit) a a

re ult of impro\'cd ocioeconomic and li\'ing tandard . coupled \\ ith an

inllux of expatriate looking fo r better li\ ing tandards. This growth spurt

re ulted in fu rther need of urban dev lopment to hou e. educate. ho pitalize.

and govern the population. This could be seen by the changes 0 er the years.

a the de rt urrounding the citie has been urbanized to better suit the

growing population (Figure -1. _6 & -I. 2�).

Population Growth 1992 .. _... c . ;; �::::. .... 1_ _ --- Legend

1992Clau,fied Image

- --

1:300,000

-

Figure 4.26: Population Distribution in 1992 76

Population Growth 2013

legend 2013 Classlf"oed 1",.9'1

- - _ £..."...... _ 0.- _ _ 0.-_

1:300,000 . . --=-==- .-.. .. -- -

Figure 4.27: Population Change in 2013 since 1992

• Tran portation

The expansion of the emirate's road network and the establislunent of

the motorways permitted the ettlement of new frontier deep in the des rt

that might haye been harder to reach if such infrastructures did not exist.

uch success tories include the prospering of the city of Liwa. situated in the

outhem region of the emirate sUlTounded by sand dunes, the cOlmecting

roads bet\veen Al Ain and Abu Dhabi, and the roads connecting Dubai and

Abu Dhabi. where a near horizontal expansion could be seen happening

across the years of the study. These networks allowed the population to

spread and not congregate at certain points in the emirate (Figures -1.28 &

-1.29). 77

Tnlsnportation Growth 1992

Legend

1992 Classified Image Futur...... , - - . . _-- . """" ­ . 0...__

1:350,000

Figure 4.28: Road etwork Limited in 1992

. Tnlsnportation Growth 20 13 •

Legend

2013 Classlfie

....., - - . �""'" . """" ­ . 0... _

1:350,000

Figure 4.29: etwork Enhancement Allowed Growth in 2013 slI1ce 1992 78

• Policie

The emirate' lead r hip tri\ed to en ure that their populace are

\\ell cared [o r b) en uring that the policies enacted ensure the benefit and

v. cllbeing of the people. In thi tudy, the mo t predominant polic that has

dra ticall) changed the emirate's LULC is the land grant . ative who meet

the criteria could be awarded a grant in one of fo ur categories; government

hou ing. re idential land. commercial land, or fa rm land. This grant allowed

further expan ion of the built-up area acros the emirate as more and more

eligible persons \\ ere able to develop the desert even further (F;gures -1. 30 &

-1. 31).

"

.F

Vegetation Expansion 1992 .l

legend .... 1992 CIus.fiod Image

_.... ------"- - _ eon.- "

1;170,000

,..-" _ ...... ",-",_ "-- 1" iN""" 0.0.- c._wc.s..�" a.- l1*XU

Figure 4.30: Empty Unused Spaces in 1992 79

Vegetation Expansion 2013

Legend

Futu,..

_.... -­ _ ....- .... - "- - - ""- -

1'170,000

Figure 4.31: ub ides llowed Expansion in 2013 since 1992

Overall. it is apparent that land development is occurnng 111 several sites causing the total area of dune to decrease around them. These ite are Abu Dhabi

City and the highway leading to Dubai Cit and Al in City, and Al Ain City and the high\ ay leading to Dubai City, thus capsuling the northern part of the Abu Dhabi

Emirate. Moreover, land cover change over time could be seen in the region uITounding Liwa City and the highway leading from and to it which is seen in the outhern part of emirate. Lastly, the area with the lowest development, compared to the pre\ ious 1\\'0, is Rmvais City, which is situated in the western part of the emirate the slow development in Ruwai City is due to the fact that it is an industrial city and the development there is mainl related to the emirate's oil & gas development strategic plan.

Although sand area had seen a recession in some parts of the study areas, it was able to gain grounds in other parts. These gains could be attributed to channel dredging that allowed the water table to drop enough causing the drying of the wet 80

oi I . or to land reclamation project that are part of th Emirate' urban de\ e]opment ma ter plan. Both are fa ctor in the sand gro\\1h in the area surrounding the coastline.

Furthermore. there i the fa mlland abandonment phenomenon that could be seen between the ) ear 2002 & 20 I" in the regions outh of 1 Ain City near the Omani border . and other ite uch a in Liwa City and along the highway connecting Abu

Dhabi it} and I Ain City. Where farmlands where abandoned thus allowing the natural reclamation of the e ites by and that previously the fa nn owner would pre\ ent fr m occurri ng. 81

Chapter 5: Conclu ion & Recommendation

Ba ed n the tud) perf ormed. the potential of remote sen lllg IS learly dem n trated v, hen it come to tracking dune movement acro time and pace. e peciall:) \\ hen implemented on a regi nal level. The tud) howed that \\ ith time. dune in the bu Dhabi mirate are expanding in the sen e of it replacement of other land coyer fe ature to pa\ e the \\ ay fo r fu ture urban development. Thi phenomenon \\ ill continu a the long as the need fo r coastal / waterfront prope11ies remain engrained in the population' need and want . onetheles . it hould be mentioned that such changes do not count fo r deseliification. since they are part of a bigger plan that involve preparing the land fo r urban development. Furthermore. these reclaimed site clas ified a and do not beha e in the same manner a dunes.

ince there are no apparent accumulation or migration of the e fe atures. Moreover. it

\\ a apparent that even though the increase of egetation saw a sharp ri e between

1992 and 2002. a clear and obviou recession truck between 2002 and 2013. wh n many fannlands were abandoned. but nonetheless. the overall vegetation land cover

howed a continuous increase.

To quantify the change that has occurred to the emirate between ] 992 and

2013. this study relied on the u e of Landsat imagery to create land use & land cover thematic maps at intennodal stages. The Landsat series is considered to be a wealth of information on global change re earch since it has been collecting images of Earth

since mid-1972. Thi information permitted the monitoring of dune changes across

the past two decades. in this study. to better understand the change in sand land cover area in the emirate of Abu Dhabi and its trends. and to better predict the fu ture

model and what it would look like in the future. 82

fhe la ification proc undertaken in thi stud) i not free from error . as challenge were confronted including the continuou tatistical confusion ben\ een land c ver feature . The u e of the un upervi ed classification. namely the Iterative

elf-Organi7ing Data nalysi Technique (l OD TA) algorithm, made it apparent

\\ hat ort of land cover fe ature are expected to be encountered in this study. and

\\ hat confu i n between diffe rent fe atures are fo und. such as betvveen urban and the e\:po ed rock. and urban and wetlands. which led to the grouping of these fe atures as n n- and ince the main objective is and mapping and movement analysis through tim . Furthermore. another error that becan1e apparent was that of the sensor being anal)zed. \\ here v" ith time. the diffe rent s n or mapped land cover fe atures more accurately.

Analyzing the change that occurred tlu'oughout this study. it is revealed that the sand cover increased by a total area of 545.44 km2 benveen 1992 to 2002. and by

116.78 km2 between 2002 to 2013. Furthern1ore. the mo t notable change is that of the vegetation cover: bet\veen 1992 and 2002. we could see a total growth of 371.84 ktn2• however. between 2002 and 2013. the total vegetation coverage dropped. and

2 78.56 kIn were 10 t.

Bet\veen 1992 and 2013, the land cover features have drastically changed. and though it might appear that dunes are increasing in area. it is mainly attributed to the pre-urban development projects that the government underiakes. The land first has be to be initiated to be better suitable fo r the construction projects they are being developed fo r. therefore. reclaiming coastal lands from the wetlands and surface water and intertidal areas are turned into sand fields. 83

In conclusion. the methodolog) fo lloVv ed by thi tudy could be adopted and expanded to tud) and examine \ ariou time points to keep track of dune movement and change in land cover. Ilov,ever. it is Ie s suitable fo r doing so fo r other land

CO\ er fe ature uch a urban. wetlands. exposed rock ince confusion is dominant. tl1U not allo\\ ing a better examination. uch cOver fe atures \vould need different

en or at higher re olution . The Land at series offe r a vast opportunity to better under tand the change that are occurring in the Abu Dhabi Emirate as a whole. and to better map and model it fu ture and po ible change . 84

Bibliography

L Ahbubi, 20 1"'.L L change analy i of bu Dhabi region between 1986 and 2010: extra ted from a mo aic of land at imagerie of bu Dhabi emirate bet\\ een ] 973 and 2010. \fa.ster ·· Thesis, L'n ited Arab Emirate Unirer it)'

bu-leid. M. 1.. Baghdad) . .R.. l-Etr. H.A.. 2001. Te. tural attribute , mineralogy und provenance or and dune fields in the greater AI Ain area, United rab Emirate . Jo urnalo/ Arid Em'ironment , 475-499

Ku\\ uri. .Y.. Kai er. M.F .. 20 1 1. Impact of orth Ga Field development on landu e/landcover changes at Al Khore, orth Qatar. using remote ensl11g and GI . Apphed Geography, 1]44 - 1153

AI-A\\udi. lM.A.. 2004. and Flo1\', pecialized Book serie , First Edition. Kuwait In tilute fo r cientific dvancement. Directorate of cientific Culture, state of Kll\\ait. P. 216

l-Dabi, H .. Koch, 1.. 1- ara\\ i. M .. EI-Baz, F .. 1997, Evolution of sand dune pattern in pace and time in north-western Kuwait using Landsat image . Jo urnalo/ Arid Em'ironments. 1 5-24

AI-Hajri. D., Abahu ain, A. ., Abdeh. .S.A.. adiq, A.A.. 2009. A Study on the mov ement of and dunes and its effect on Umm Sa'id Industrial City. Qatar, ing Remote ensing and GIS, F0!1h ational GI Symposium in audi Arabia. Dammam. Kingdom of Saudi Arabia

Amdev Ren. H .. Dem. R. .. Kanik. A .. Kesk. .S .. 2005. Use of principal component cores in multiple linear regression models fo r prediction of Chlorophyll-a in reservoirs. Ecological Afode1ing, 581-589

Bishop. 1.A.. 2008. COMPARA TNE POI T PATTTERN A AL YSIS OF HYPERBORE E DAE. MARS. A D THE RUB' AL KHALI SA D EA. EARTH. Plane/my Dune Wo rkshop: A Record 0/ Climate Change, Alanlogordo. ew Mexico

Chang. E.M.. Park, K., 2006, Feature Extraction in an Aerial Photography of Gimnyeong and Dune Area by Texture Filtering, Jo urnal 0/ the Korean Geographical Societ)·. 139-149

Chen. X .. Vierling, L., Deering. D .. 2005, A simple and effe ctive radiometric correction method to improve landscape change detection across sensors and across time. Remote Sensing a/ Environment, 63-79

EI-Sayed. MJ., 1999. Sedimentological characteristics and morphology of the aeolian sand dunes in the eastern part of the UAE: a case study from Ar Rub' Al Khali, Sedimenlmy Geology, 219-238 85

I:mbabi .. . . EI harhan. .. Yeiha. bdel Kader. 0.. ha\.\ ki. E .. El Badry. K .. Kariem. F .. 1993. alional tla of the nited rab Emirate . Geoproject'i LimUed. 198 plate

Embabi...... 1991. Dune t) pe and pattern in the niled Arab Emirates using Land at TM-data. Proceedings of the _..fth Intemotional Y1l1posium of RC!II/ote , C!n ing& Em'ironlJlC!nt. Rio de Janeiro. BraziL 27-3 1.

Fung. T .. LeDrew. E .. 1987. Application of principal components analysis change detection. Photograml71etric Eng;'7eedng and Remote en ing. 1649 - 1658

Gao. B. .. Heidebrecht. K, B .. Goetz. . F. II.. 1993. Derivat ion of caled surface reflectance from VIRI data. Remote Sensing of Em-ironment. 145-163

Gao. L 2009. Digital Analy i oj Remotely Sensed Image/)'. McGraw-Hill ompanie . ew York. 689 pp

Gardner A . Hov. arth B (2009) rbanisation in the United Arab Emirates: the challenges fo r ecological mitigation in a rapidly developing country. In: Krupp F. Mus elman LJ. Kotb MMA. Weidig I (Eds) EnvirolID1ent. Biodiver ity and Con ervation in the Middle East. Proceedings of the First Middle Ea tern Biodiver ity Congress, Aqaba, Jordan, 20-23 October 2008. BioRi k 3: 27-38.

Gerbern1aru1. A. H., & eher. D. D .. 1979. Reflectance of varying mixture of a clay soil and and, Photogral11117etric Engineering and Remote Sen ing, 1 1 45- ] 151

Ghadiry. M .. Koch. B., 2010. Developing a Monitoring ystem fo r and Dune Migration in Dakhla Oasis. Western Desert, Egypt Remote Sen ing fo r cience, Education, and Ta/ural and Cultural Heritage. 313-320

Glennie. K.W.. 2001. Evolution of the emirates' land surface: an introduction in AJ­ Abed. 1.. Hellyr, P .. United Arab Emirates: A eH' Per peclive, London. Trident Press. 9-27

Gong. P .. 1993. Change detection using principal component analysis and fuzzy set theory. Canadian Journaloj Remote Sensing, 22 - 29

Goodchild M. F .. Parks B. O. and Steyaert L. T., 1993, Environmental Modeling w'ith GIS. OxJord Unirersity Pres . ew York. 488 p.

Hadee!. ... Jabbar. M.T .. Chen, X .. 2010, Application of remote sensing and GJ in the study of environmental sensitivity to desertification : a case study in Basrah Province, southern part of Iraq, App/ Geomal. 101-112

Howari, F.M.. Baghdady, A.. GoodelL P.e., 2007, Mineralogical and gemorphological characterization of sand dunes in the eastern part of United Arab Emirates using orbital remote sensing integrated with field 86

im e tigation . Geomorphology, 67-8 I

Huang. C, Kim. .. ong. K., Town hend. J.R.G, Davi , P.. Itstatt, A.. Roda , 0 .. Yano kyo .. Clay.R .. Tucker. C.J., Mu in kyo J .. 2009. essment of Paraguay' fo re t co\ er change u ing Land at observations. Global and PlanetaTY '/wnge. J - 12

1 runt. G.R.. 1982. pc tro copic propertie of rocks and minerals. In Carmichael. R ... Handhook of Ph), iccd Properties of Rock . vol. 1. CRC Pres . Boca Raton. FLo 416 P

Hutchi on. .F.. 1982. Remote en ing of arid and semi-arid rangeland, International Geo cience and Remote Sensing . ),mposiul11. Munich, Federal R public of Germany. 5.1 -5.6

Ifarraguerri. .. Chang. C.r.. 2000. Unsuper ised Hyperspectral image analysis with projection pur uit. IEEE Transactions on Geo cience and Remote Sen ing. 2529-2538

Janke. J.R.. anal) i of the current stability of the Dune Field at Great and Dunes Jational M nument u ing temporal TM imagery (1984-1998), Remole en ing of Environment. 488-497

Jen en. J. R. (2005): Introductory digital image proces ing: a remote sensing perspective. John R. Jensen. 3rd ed., Pearson Education Inc., Upper Saddle Ri\er.

Karnieli. A.. Tsoar. H .. 1995, pectral Reflectance of biogenic cru t developed on desert dune sand along Israel-Egypt border. InternationalJo urnal of Remote en. ing, 369-374

KrislmamUl1hy, 1.. rinivas. G.. 1996. Demarcation of the geological and geomorphological fe atw'es of parts of Dharwar Craton. Kamataka, using IRS LIS -II data, fnternalional Journal of Remote Sensing, 3271-3288

Kumar. M ..Goossens. E., Goossens, R.. 1993. Assessment of sand dune change detection in Rajasthan (Thar) Desert India, International Jo urnal of Remote en ing. 1689- 1 703

Lam. D.K .. RenU11el. T.K.. Drezner, T.D .. 201 1, Tracking Desertification in California Using Remote Sen ing: A Sand Dune Encroachment Approach. Remote Sen ing, 1-13

Larue. E.M.. 2004. Using GIS to establish a public library consumer health collection. Biomedical Digital Libraries. pp 3

Levin. .. Ben-Dor. E .. Karnieli. A.. 2004. Topographic information of sand dunes as extracted from shading effects using Landsat images. Remote Sensing of Environment. 190-209 87

Liao, ., Bai y, 2009. ne\\ grid-c lI-based method fo r rror eyaluation of \'ector- to-ra ter comoer ion, 'omputational G osciences

LiJIc and, T.M .. Kiefer. R. \ .. 1994. Remote L el1sing and image interpretation. 3rd Edition. John Wile) & n,In c .. 230 pp

Lu. 0 .. Mau eL P .. Brondizio, E .. Moran, E .. 2004. hange detection technique . infernationalJo urnal o.lRemote Sensing, 2365 - 2407

la ek. J. G .. Honzak. M., Goward, . ., Lill, P . . & Pak. E .. 2001. Land at-7 ETM+ a an ob er\,atory fo r land cover: Initial radiometric and geometric compari ons with Landsat-5 Thematic Mapper. Remote Sensing of EI7l'ironment, 118- 130

.d. Geocaching. Web. 27 Jan. 2016.

.

Otazu. X., rbioL R .. 2000, Land u e map production by fu sion of multi pectral cla ification of landsat images and texture analysis of high resolution image . international archieved fo r photogrammetry and remote sensing. Vol. XXXIII. Supplement B7.

Pea e. P.P.. Bierly. G.D., Tchakerian, V.P .. Tindale, . W., 1999. Mineralogical characterization and transport pathways of dune sand using Landsat TM data. Wahiba and Sea, Sultanate of Oman Geomorphology. 235-249.

Pu. R.L., Gong. P .. 2000, H) perspectral Remote en ing and Its Application, High Education Press, Beijing. p. 70

Rajesh, H.M.. 2008. Mapping Proterozoic unconfom1ity-related uranium deposits in the Rockhole area, Olihern Territory. Australia using Landsat ETM+, Ore Geology Re1'ie1l" 382 - 396

Richards. 1.A.. 2000. Remote Sensing Digital image A nal) 's is , 3rd ed. Springer­ Verlag. Berlin, p 190

Richards, 1.A., lia, x., 1999, Remote Sensing Digital Image Anal)' i: An introduction. 3rd ed. Springer-Verlag, Berlin, 363 p

abins. F .. 1987, Remote Sen ing: Principles and Interpretation, 2nd ed. Freeman, ev. York

abins, F.F .. 1997. Remote Sensing: Principles and interpretation, W. H. Freeman and Company, ew York. 549 pp

arnthein, Mo, 1978, Sand deserts during glacial maximum and climatic optimum, Na ture, 43-46 88

' heidt. .. Ram y. 1.. Lanca ter. ., 2008. Radiometri normalization and image mo aic g nerati n of TER th rmal infrared data: An application to c�ten 'i\"e and heet and dune field , Remote Sensing of Environment, 9_0- 933

ingh. .. 1989. Digital change detection techniques using remotely en ed data, InternationalJo urnal o/ Rell1ole Sensing. 989 - 1003

mith. 1. ., stin, .L.. dam , J .B., Gillespie, .R .. 1990, Vegetation in deselis: 1. regional mea ure of abundance from multispectral image . Remote , C!nsinf( 0/ Enrironl71ent. 1-16

ultan. L rvid on. R.E.. Duncan, I.J.. tern. R., El Kaliouby. B .. 1986, Extension of th aj d Fault ystem from audi rabia to the central Eastern De ert of Eg) pt ba ed on int grat d field and Landsat observations. Te ctonics. 1291- 1306

ummerfield M .. 1996. Glohal GeoIJ101pholog),. Longman Group, Essex. .K

Teillet. P. M .. Barker. J. L., Markham. B. L.. Iri h. R. R., Fedo ej evs. G., & Store , J. C .. 1001. Radiometric cros -calibration of the Landsat-7 ETM+ and Land at-5 TM ensor ba ed on tandem data sets. Remote Sensing 0/ Em'ironment, 39- 54

Tsoar. H.. 2001. Types of Aeolian sand dunes and their fo rmation. In Balmforth. J .. & Provenzale. A.. Lecture ote in Physics. 403- 429

T oar. H .. Blumberg. D.G .. Stoler. Y .. 2004. Elongation and migration of sand dune . GeoI1101pholog),. 293-302.

Tuller. P.T.. 1987. Remote Sensing science application in arid environment. Remote Sen ing of Enrironmenf. 143- 154

Vincent. R.K.. 1997. Fundamentals of Geological and Environmental Remote Sensing. Prentice Hall. Upper addle River, 1. 270 pp

Wasson. R. J.. & Hyde. R., 1983, Factors detern1ining desert dune type. 'ature. 337- 339

White. K ..Walden. J.. Drake. .. Eckardt. F .. Settle. 1., 1997. Mapping the iron oxide content of dune sands. amib sand sea. an1ibia, using Landsat Thematic Mapper data. Remote Sensing of Environment. 30- 39

Yao. Z.Y .. Wang. T.. Han. Z.W .. Zhang, W.M .. Zhao. A.G .. 2007. Migration of sand dunes on the northern Alxa Plateau. Inner Mongolia. China. Jo urnal of Arid Environments, 80-93 Barchan Dunes are crescent Linear Dunes are long shaped and they form in areas straight dunes that form in where there is a hard ground areas with limited sand surface a moderate supply of supply and converging wind sand constant wind direction directions

Parabolic are "U" shaped Transverse Dunes are large dunes with an open end fields that resemble sand ripples facing upwind. They usually on a large scale; they form in stabilized by vegetation, and areas where there is abundant occur where there is > supply of sand and constant "0 abundant wind direction, a "0 wind direction � constant wind direction, == and an abundant sand Q. ;..( supply. They are common In coastal areas

W.nd \. \. Barchan Dunes can merge into Star dunes are dunes with \.. transvers dunes if the supply of [) several arms and variable sand increases slip face directions that \ \ form in areas where there is \ abundant sand and variable wind direction

Figure 1.5: Relationship of Dune Pattern & Wind Directions 89 Row Path Resolution Parameter Sensor Platform

162 161 160

43 04-06- 1 992 05-06- 1992 14-06- 1 992 30 meters COR Surface [ andsat 4 Reflcctance 03-07-2002 10-07-2002 12-07-2002

01-08-20 13 25-07-20 13 18-07-20 13 Landsat 7

44 04-06-1 992 05-06- 1992 14-06- 1 992

03-07-2002 10-07-2002 12-07-2002 Landsat 8

01-08-20 13 25-07-20 13 18-07-20 13

Datum WGS 1984 WGS 1984 WGS 1984 Projection UTM Zone 39 N UTM Zone 39 N UTM Zone 40 N

Table 3.1: Summary of Dates and Scenes Used

Data Type Date Spatial Resolution Purpose

Rapid Eye 20 13 5m Accuracy assessment

lKONOS 2003 1m Accuracy Assessment

SPOT Panchromatic 1986 10 m Accuracy Assessment

Tahle 3.8: Imagery used fo r accuracy assessment

90 91

Ba ic Stat \t1in Ma:-. Mean Stde\

Band 1 -9999 16000 -6252.606775 5504.63962

Band 2 -9999 6170 -5980.524268 5904. I 84034

Band ., -9999 6906 -5732.788246 6269.470532

Band 4 -9999 7800 -5505.377137 6605.09676 1

Band 5 -9999 16000 -5 I 96.9497 7083. I 01278

Band 6 -9999 16000 -5375.882562 6805.800292

Table 3.5: Original Image

Ba ic Stat Min Max Mean Stde\

Band I -9998.404297 I 1334.2871 1 -6252.606446 5504.383371

Band 2 -9999. 1 54297 10335.90527 -5980.52446 1 5904.081714

Band 3 -9999.5 1 0742 8558.262695 -5732.787964 6269.3 13087

Band 4 -9999.255859 8426. 101563 -5505.3 76756 6604.653347

Band 5 -9998.989258 16014.30566 -5 1 96.94967 1 7083. 101044

Band 6 -9998.638672 13575. 10645 -5375.882695 6805.6553 11

Table 3.6: Inverse PCA

Number CIa Name Area (m- ) Longitude X Latitude Y

1 Expo ed Bedrock 239400 55.85790 24. 1 7674

2 Exposed Bedrock 272700 55.80807 24. 1 2984

.., J Exposed Bedrock 329400 55.92488 24. 1 0874

4 Exposed Bedrock 219600 55.87740 24.1 1057

5 Exposed Bedrock 213300 55.98582 24.07560

6 Exposed Bedrock 78300 55.05485 23.4585 1 92

7 E�po ed Bedrod. 54000 55. 14645 23.5501 1

8 [.'.po· d Bedrock 90000 55.25873 23.38995

9 ["po ed Bedr ck 57600 55.06358 23.2604 1

10 ["po ed Bedrock 40500 55.08874 23 .11769

II ["po ed Bedro k 18900 55.07350 23 .09072

12 E"posed Bedrock 56700 55.1 2020 23.340 16

13 E�po ed Bedrock 102600 55. J 3896 23. 1 4545

14 E.'.posed Bedrock 39600 55.05860 22.9904 1

15 E.'.po ed Bedrock 36900 55.02905 22.93987

16 E.'.posed Bedrock 59400 55. 1 7806 23.07842

17 Exposed Bedrock 69300 55.14979 23.00472

18 Ex.posed Bedrock 20700 55.1 0206 22.92965

19 Expo ed Bedrock 45000 55.19510 22.92255

20 E.'.posed Bedrock 73800 55. 1 0592 22.66125

21 Intertidal 212400 54.42082 24.3 1528

')') -- Intertidal 78300 54.30486 24.29348

23 Intertidal 126000 54. 1 3290 24. 1 7732

24 Sand 39600 54.9091 1 24.56342

25 Sand 24300 55.20990 24.49442

26 Sand 230400 54.93839 24.2 1582

27 Sand 496800 55.4 1940 24.45562

28 Sand 80100 55. 1 7505 24.41627

29 Sand 277200 54.20540 23.96498 93

30 Sand 162000 54.29597 24.02682

31 Sand 55800 54.95474 24. 1 4206

., .., .) � Sand 54900 54.87611 24. 1 0853

33 Sand 78 1 200 55.3007 1 24.02830

34 Sand 121500 51.83614 23.89597

35 Sand 27000 52.02787 23.80 108

36 Sand 900 52.04082 23.76266

37 Sand 98 100 52.03862 23.7599 1

38 Sand 54000 52.1 0286 23.74464

39 Sand 37800 52.1 0627 23.69375

40 Sand 900 52. 10412 23.69340

41 Sand 234000 52.69980 23.84309

42 Sand 77400 52.6 1 254 23.78852

43 Sand 103500 52.6 1 076 23.77605

44 Sand 233 100 52.88266 23.86424

45 Sand 208800 52.75354 23.84472

46 Sand 232200 52.80203 23.80996

47 Sand 90900 52.97576 23.68664

48 Sand 343800 53.28420 23.90807

49 Sand 225900 53.09927 23.9 1057

50 Sand 359100 53. 16407 23.87686

51 Sand 432000 53.5083 1 23.92628

52 Sand 285300 53.83 196 23.92 158 94

53 Sand 369000 53.868 12 23.88984

54 Sand 140400 53.87799 23.803 17

55 Sand 171000 53.93 159 23.935 18

56 Sand 181800 54. 1 6804 23.92578

57 Sand 206 100 54.02015 23.92853

58 Sand 107100 53.96309 23.90842

59 Sand 309600 53.96782 23.79683

60 Sand 178200 54. 1 5475 23.68377

61 Sand 161 100 54.09939 23.67099

62 Sand 665 100 54.50263 23.86638

63 Sand 393300 54.36611 23.75372

64 Sand 188100 54.25477 23.70131

65 Sand 203400 55.01 988 23.78133

66 Sand 277200 54.85270 23.71927

67 Sand 200700 54.94555 23.70860

68 Sand 219600 55.3 1098 23.85337

69 Sand 160200 55.15812 23.7848 1

70 Sand 70200 52. 1 3726 23.53918

71 Sand 181800 52.89853 23.67225

72 Sand 153900 53.83042 23.623 15

73 Sand 172800 53.68902 23.61 130

74 Sand 179100 54.01 496 23.64829

75 Sand 164700 54.020 13 23.624 19 95

76 Sand -+96800 54.26765 23 .5046-+

77 Sand 748800 54.47902- 23.47654

78 Sand 230400 54.8 1 004 23.61255

79 Sand 63000 55.34 124 23.-+7955

80 Sand 68400 55.3 1925 23.44978

81 Sand 190800 53.32713 22.97557

82 Sand 199800 53.54490 22.94772

83 Sand 162000 53.52320 22.922 14

84 Sand 128700 54.2 1428 23.04 178

- 8 Sand 162900 54.23936 23.02597

86 Sand 243000 53.5960 1 23.94774

87 Wet Soil 360000 54.69626 24.72054

88 Wet Soil 167400 51.88 1 18 23.973 13

89 Wet Soil 593 100 52.56356 24.08089

90 Wet Soil 22 1 400 52.93 105 24.11449

91 Wet Soil 547200 53.30494 24.07995

92 Wet Soil 360900 53.72570 24.01713

93 Wet Soil 541 800 54. 13168 24. 12557

94 Wet Soil 148500 54.34816 24. 1 7384

Table 3.7: List of accuracy assessment sites 96

From - To Area (Km- )

No Change 53775.08

E:\posed Ro k - I ntel1idal 5.34

E:\po ed R d. - Surface Water 1.72

E:\po d Rock - Sand 617.22

E:\po ed Rock - Vegetation 72.20

E:\po ed Rock. - Wet Soil 212.83

Intert idal - b.po ed Rock 13.23

lntertidal - Surface Water 271 .24

Intertidal - Sand 33.42

Inteltidal - Vegetation 48.89

Intertidal - Wet Soil 122.54

Surface Water- Exposed Rock 0.34

Surface Water- Intertidal 73.84

Surface Water- Sand 8.06

Surface Water- Vegetation 0.03

Surface Water- Wet Soil 10.54

Sand - Exposed Rock 684.87

Sand - Intertidal 10.29

Sand - Surface Water 0.12

Sand - Vegetation 3 I 0.57

Sand - Wet Soil 1208.96

Vegetation - Exposed Rock 54.78

Vegetation - Intertidal 6.21

Vegetation - Surface Water 1.94

Vegetation - Sand 8.46 97

Vegetation - Wet Soi I 1.95

\\ et Soi I - Exp ed Rod. 159.92

Wet Soil - Intertidal 112.57

Wet Soil - Surface Water 1.84

W t Soil - Sand 2095.20

Wet Soil - Vegetation 13.48

Table 4.8: Change from 1992 to 2002 statistics

From - To Area (Km-)

No Change 49967.90

Expo ed Rock - Sand 617.22

Intertidal - Sand 33.42

Surface Water- Sand 8.06

Sand - Exposed Rock 684.87

Sand - Intertidal 10.29

Sand - Surface Water 0. 12

Sand - Vegetation 310.57

Sand - Wet Soil 1208.96

Vegetation - Sand 8.46

Wet Soil - Sand 2095.20

Table 4.9: Changes fo r sand fe atures from 1992 to 2002 statistics

From - To Area (Km- )

No Change 53689.72

Exposed Rock - Intertidal 7.43

Exposed Rock - Surface Water 0.30 98

E"po ed Roc I-.. - Sand 42 1 .40

E.'po ed R cl-.. -Vegetation 132.50

E:\.p ed Rocl-.. - Wet Soil 43.88

Intertidal - E:-..po ed Rocl-.. 71.34

Intettidal - Surfac Water 20.55

Intertidal - Sand 179.47

Inteliidal - Vegetation 46.35

Intertidal - Wet Soil 79.54

Surface Water- E:\.posed Rocl-.. 41.23

Surface Water- Intel1idal 648.32

Surface Water- Sand 8.00

Surface Water- Vegetation 0.32

Surface Water- Wet Soil 5.09

Sand - Exposed Rock 1408.30

Sand - Intert idal 8.23

Sand - Surface Water 7.30

Sand - Vegetation 51.14

Sand - Wet Soil 765.0 1

Vegetation - Ex.posed Rock 219.40

Vegetation - Intertidal 25.89

Vegetation - Surfa ce Water 0.09

Vegetation - Sand 82.14

Vegetation - Wet Soil 1.63

Wet Soil - Exposed Rock 797.72

WetSoil - Intertidal 94.61

Wet Soil - Surface Water 6.04

Wet Soil - Sand 1665.73 99

egetation 19.90

Table 4.10: hange fr om 2002 to 201" statistics

From - To Area (Km- )

NI Chang 50488.08

E:-.po ed Rock - Sand 42 1 .40

Intet1idal - Sand 179.4 7

Surface Water- Sand 8.00

Sand - EApo ed Rock 1408.30

Sand - Intertidal 8.23

Sand - Surface Water 7.30

Sand - Vegetation 51.14

Sand - Wet Soil 765.0 1

Vegetation - Sand 82. 14

Wet Soil - Sand 1665.73

Table 4.1 1: Changes fo r sand fe atures from 2002 to 2013 statistics

From 1992 to 2002 to 2013 Km-

No Change 48467.49

EAposed Rock - Exposed Rock - Sand 134 .1106

Exposed Rock - Intertidal - Sand 0.287151

Exposed Rock - Surface Water- Sand 0.0027

Exposed Rock - Sand - Exposed Rock 116.5672

Exposed Rock - Sand - Intertidal 0.036308

Exposed Rock - Sand - Sand 491 .2252

Exposed Rock - Sand - Vegetation 4.983226 100

bpo ed Rock - Sand - Wet Soil 4.343693

E:\.po ed Rock - Vegetation - Sand 4..+97708

E:\po ed Rock - Wet Soil - Sand 122.1677

Intertidal - E:\po ed Rock - Sand 0.73805"

Intert idal - Intertidal - Sand 15..L 1987

Intertidal - Surface Water- Sand 2.71 1898

Intertida I - Sand - E:\po ed Rock 5.080 186

Intertidal -Sand - Intertidal 0.861521

Intertidal - Sand - Surface Water 0.008743

Intertidal - Sand - Sand 20.34425

Intertidal - Sand - Vegetation 0.412198

Intertidal - Sand - Wet Soil 6.685668

Intertidal - Vegetation - Sand 5.144544

Intertidal - Wet Soil - Sand 37.92 144

Surface Water- Exposed Rock - Sand 0.0067 12

Surface Water- Intertidal - Sand 0.754 12

Surface Water- Sand - Exposed Rock 0.232169

Surface Water- Sand - Inteltidal 0.01 1 572

Surface Water- Sand - Sand 0.338892

Surface Water- Sand - Vegetation 0.046 158

Surface Water- Sand - Wet Soil 0. 1 59404

Surface Water- Vegetation - Sand 0.0018

Surface Water- Wet Soil - Sand 1.131525

Sand - Exposed Rock - Exposed Rock 371.2412

Sand - Exposed Rock - Intertidal 0. 1 11265

Sand - Exposed Rock - Surface Water 0.050273

Sand - Exposed Rock - Sand 252.3309 101

Sand - b.po ed Rock.- Vegetation 51.22 143

Sand - E:\.po ed R ck - Wet Soil 9.705026

Sand - Intertidal - Expo ed Rock 1.412122

Sand - Intert idal - Intertidal 0.76 174

Sand - Intertidal - Surface Water 0.007277

Sand - Intel1idal - Sand 6.568865

Sand - Intertidal - Vegetation 0. 1 18052

Sand - Intert idal - Wet Soil 1.41369 1

Sand - Surface Water- Expo ed Rock 0.02 1575

Sand - Surface Water- Intertidal 0.0304 19

Sand - Surface Water- Open Water 0.029469

Sand - Surface Water- Sand 0.01 89

Sand - Surface Water- Wet Soil 0.0 1 566

Sand - Sand - Exposed Rock 1058.593

Sand - Sand - Intertidal 1,470492

Sand - Sand - Surface Water 0.00882

Sand - Sand - Vegetation 40.24828

Sand - Sand - Wet Soil 397.9692

Sand - Vegetation - Expo ed Rock 158.3852

Sand - Vegetation - J ntertidal 0.1 25357

Sand - Vegetation - Sand 66. 16132

Sand - Vegetation - Vegetation 85.52665

Sand - Vegetation - Wet Soil 0.21 0803

Sand - Wet Soil - Exposed Rock 293.82 13

Sand - Wet Soil - intertidal 2.563589

Sand - Wet Soil - Surface Water 0. 1 34048

Sand - Wet Soil - Sand 642.4969 102

Sand - v., et Soil - Vegetation 5.43866

Sand - Wet Soil - Wet Soil 264. 1563

Vegetation - E �po ed Rock - Sand 12.7771 1

Vegetation - Intertidal - Sand 0.300238

Vegetation - Surface Water- Sand 2.57E-05

Vegetation - Sand - Expo ed Rock 3.165749

Vegetation - Sand - Intel1idal 0.0144

Vegetation - Sand - Sand 4. 1 37737

Vegetation - Sand - Vegetation 1.105305

Vegetation - Sand - Wet Soil 0.035177

Vegetat ion - Vegetation - Sand 5.9 1 6207

Vegetation - Wet Soil - Sand 0. 1 98693

Wet Soil - Exposed Rock - Sand 21.38466

Wet Soil - Intel1idal - Sand 17.3575

Wet Soil - Surface Water- Sand 0.00 1 003

Wet Soil - Sand - Expo ed Rock 224.6085

Wet Soil - Sand - Intertidal 5.829352

Wet Soil - Sand - Surface Water 0.0 1 1211

Wet Soil - Sand - Sand 1504.255

Wet Soil - Sand - Vegetation 4.33453

Wet Soil - Sand - Wet Soil 355.794 1

Wet Soil - Vegetation - Sand 0.407027

Wet Soil - Wet Soil - Sand 861 .73 74

Table 4. 12: Sand / on Sand change analysis statistic from 1992 -2002 -2013