Spatiotemporal flood hazard and flood risk assessment using remote sensing techniques. Case study: Khartoum State, Sudan
Abstract
The state of Khartoum being the most populated state in Sudan, faces the consequences of floods recurrence almost annually during rainy season. Policy makers and on ground NGOs need to tackle the hazard of floods in an effective and efficient manner. Recent research demonstrated the capabilities and potentials of remote sensing in flood hazard and risk mapping. This study aims to map flood hazard and assess the risk of floods in state of Khartoum, Sudan. In order to identify the flood hazard in state counties, an inundation indicator is used, namely the relative frequency of inundation (RFI). Flood events that occurred from 1988 to 2018 were mapped using Landsat satellite images, and maximum flood extent was then delineated. RFI was obtained using maximum flood extent maps and served as the flood hazard map. We developed a Land Cover Land Use (LCLU) map using Landsat 8 to identify affected urban and croplands areas in the state of Khartoum. RFI values was used along with LCLU map to assess state counties, and to assess the vulnerability of public facilities (health and educational facilities) using zonal statistics. It was demonstrated that, in terms of average RFI values for LCLU classes per county, croplands had the highest flood hazard, and Urban areas carried a relatively moderate flood hazard. The results of this study indicate that croplands on the riverbanks are the most inundated areas in the state of Khartoum, and the most urbanized counties have the highest flood hazard.
Full text
i
SPATIOTEMPORAL FLOOD HAZARD AND FLOOD
RISK ASSESSMENT USING REMOTE SENSING
TECHNIQUES
Case s udy: Kha oum S a e, Sudan
Abee Awad Abdulmagied Khai y
ii
SPATIOTEMPORAL FLOOD HAZARD AND FLOOD RISK
ASSESSMENT USING REMOTE SENSING TECHNIQUES IN
KHARTOUM STATE, SUDAN
Disse a ion supe ised by
PhD, Ana C is ina Ma inho da Cos a
PhD, Filibe o Pla Banon
PhD, Ma co Oc á io T indade Painho
Feb ua y 2020
iii
DECLARATION OF ORIGINALITY
I decla e ha he wo k desc ibed in his documen is my own and no
om someone else. All he assis ance I ha e ecei ed om o he people
is duly acknowledged and all he sou ces (published o no published)
a e e e enced.
This wo k has no been p e iously e alua ed o submi ed o NOVA
In o ma ion Managemen School o elsewhe e.
Lisbon, Feb ua y 14 h, 2020
Abee Awad Abdulmagied Khai y
The signed o iginal has been a chi ed by he NOVA IMS se ices
i
ACKNOWLEDGMENTS
O Allah, ou Lo d, o hee be he p aise in all he hea ens, and in all he ea h,
and in all ha i pleases Thee o c ea e a e wa ds. O hou Who a wo hy o p aise
and glo y, mos wo hy o wha a se an says, and we a e all hy se an s, no one
can wi hhold wha Thou gi es o gi e wha Thou wi hholdes . I is only h ough
he guidance o Allah ha his wo k has been accomplished, and o Him all he
p aise is due.
The ea e , I would like o exp ess my since e app ecia ion o my supe iso s,
whom, wi hou hei con inuous suppo and encou agemen I wouldn’ ha e been
able o accomplish his wo k. I am g a e ul o and hono ed by all he ene gy and
ime you ha e pu h ough his wo k, o you insigh ul obse a ions and
commen s ha helped me imp o e on my wo k. I would like o hank he Eu opean
Commission o acili a ing he E asmus Mundus Join Mas e P og am and hus
allowing me o bene i om his unique oppo uni y. I also wan o ex end my
g a i ude o he s a and acul y membe s o NOVA In o ma ion Managemen
School IMS o p o iding me wi h endless suppo h oughou he cou se o he
hesis wo k.
I would like o hank my iend B endan O'Han ahan, o his encou agemen
and help in ge ing ex a in o ma ion, and Abdallah Almokash i and Ghazi El ayah
o hei suppo .
I am humbled by he amoun o emo ional and physical suppo ha my small
and ex ended amily ha e gi en me o ca y h ough and comple e his wo k.
Fo e e I will be in g a i ude o my sis e s, my husband Ahmed, and my in-laws,
who, h ough hei wo ds o ad ice and suppo ga e me he s eng h o push
o wa d.
Finally, I dedica e his wo k o my daugh e , my mo he , and my la e a he , all
ha I am and all ha I do is due o you and o you, you a e my o e.
Spa io empo al lood haza d and lood isk assessmen using
emo e sensing echniques
Case s udy: Kha oum S a e, Sudan
ABSTRACT
The s a e o Kha oum being he mos popula ed s a e in Sudan, aces he
consequences o loods ecu ence almos annually du ing ainy season. Policy
make s and on g ound NGOs need o ackle he haza d o loods in an e ec i e and
e icien manne . Recen esea ch demons a ed he capabili ies and po en ials o
emo e sensing in lood haza d and isk mapping. This s udy aims o map lood
haza d and assess he isk o loods in s a e o Kha oum, Sudan. In o de o iden i y
he lood haza d in s a e coun ies, an inunda ion indica o is used, namely he
ela i e equency o inunda ion (RFI).
Flood e en s ha occu ed om 1988 o 2018 we e mapped using Landsa sa elli e
images, and maximum lood ex en was hen delinea ed. RFI was ob ained using
maximum lood ex en maps and se ed as he lood haza d map. We de eloped a
Land Co e Land Use (LCLU) map using Landsa 8 o iden i y a ec ed u ban and
c oplands a eas in he s a e o Kha oum. RFI alues was used along wi h LCLU
map o assess s a e coun ies, and o assess he ulne abili y o public acili ies
(heal h and educa ional acili ies) using zonal s a is ics. I was demons a ed ha , in
e ms o a e age RFI alues o LCLU classes pe coun y, c oplands had he
highes lood haza d, and U ban a eas ca ied a ela i ely mode a e lood haza d.
The esul s o his s udy indica e ha c oplands on he i e banks a e he mos
inunda ed a eas in he s a e o Kha oum, and he mos u banized coun ies ha e he
highes lood haza d.
i
KEYWORDS
Flood Haza d Mapping
Flood Risk Assessmen
Remo e Sensing
Land Co e Land Use
ii
ACRONYMS
AHP – Analy ical Hie a chy P ocess
CCD – Cold Cloud Du a ion
IDW – In e se Dis ance Weigh ing
LCLU – Land Co e Land Use
MCA – Mul i-C i e ia Analysis me hods
NDBI – No malized Di e ence Buil -up Index
NDVI – No malized Di e ence Vege a ion Index
NDWI – No malized Di e ence Wa e Index
OSM – Open S ee Maps
RFI – Rela i e F equency o Inunda ion
SAR – Syn he ic Ape u e Rada
UNDRR – Uni ed Na ion O ice o Disas e Risk Reduc ion
UNITAR – Uni ed Na ion Ins i u e o T aining and Resea ch
UNOSAT – UNITAR’s Ope a ional Sa elli e Applica ion P og amme
UNOCHA – Uni ed Na ion’s O ice o Coo dina ion o Humani a ian A ai s
USGS - Uni ed S a es Geological Su ey
VHR – Ve y High Resolu ion
WHO – Wo ld Heal h O ganiza ion
iii
INDEX OF THE TEXT
ACKNOWLEDGMENTS .................................................................................................................. IV
ABSTRACT ........................................................................................................................................ V
KEYWORDS ..................................................................................................................................... VI
ACRONYMS ................................................................................................................................... VII
1. INTRODUCTION ...................................................................................................................... 1
1.1. MOTIVATION ..................................................................................................................... 1
1.2. AIM AND OBJECTIVES ........................................................................................................ 1
1.3. THESIS OUTLINE ................................................................................................................ 2
2. LITERATURE REVIEW ........................................................................................................... 3
2.1. FLOOD RISK ASSESSMENT USING REMOTELY SENSED DATA ............................................... 3
2.2. FLOOD RISK ASSESSMENT IN SUDAN ................................................................................. 5
3. METHODS ................................................................................................................................. 6
3.1. STUDY AREA ..................................................................................................................... 8
3.2. DATA ................................................................................................................................. 9
3.2.1. Remo ely Sensed Da a .................................................................................................. 9
3.2.1. Auxilia y Da a ............................................................................................................ 11
3.3. MAPPING FLOOD EXTENT ................................................................................................ 11
3.4. MAPPING MAXIMUM FLOOD EXTENT .............................................................................. 12
3.5. RELATIVE FREQUENCY OF INUNDATION .......................................................................... 13
3.6. FLOOD RISK ASSESSMENT ............................................................................................... 14
3.6.1. LCLU Map.................................................................................................................. 14
3.6.2. Zonal S a is ics ........................................................................................................... 15
4. RESULTS AND DISCUSSION ............................................................................................... 17
4.1. FLOOD EXTENT MAPS ...................................................................................................... 17
4.1.1. Scan-Line Co ec o (SLC) p oblem .......................................................................... 17
4.2. MAXIMUM FLOOD EXTENT MAPS .................................................................................... 18
4.3. RELATIVE FREQUENCY OF INUNDATION MAP .................................................................. 20
4.4. FLOOD RISK ASSESSMENT ............................................................................................... 21
4.4.1. LCLU Map.................................................................................................................. 21
4.4.2. Zonal S a is ics ........................................................................................................... 24
4.5. HUMANITARIAN MAPPING RESPONSE FOR AUGUST 2013 AND JULY 2016 FLOOD EVENTS
25
5. CONCLUSION ........................................................................................................................ 30
5.1. LIMITATIONS.................................................................................................................... 31
5.2. RECOMMENDATIONS ........................................................................................................ 31
REFERENCES ................................................................................................................................... 32
ANNEX 1: SPECTRAL SPECIFICATIONS FOR LANDSAT IMAGES ........................................ 36
ANNEX 2: MAXIMUM FLOOD EXTENT MAP FOR 1991 .......................................................... 38
ANNEX 3: MAXIMUM FLOOD EXTENT FOR 1992 .................................................................... 39
ix
ANNEX 4: MAXIMUM FLOOD EXTENT FOR 1993 .................................................................... 40
ANNEX 5: MAXIMUM FLOOD EXTENT FOR 1994 .................................................................... 41
ANNEX 6: MAXIMUM FLOOD EXTENT FOR 1996 .................................................................... 42
ANNEX 7: MAXIMUM FLOOD EXTENT FOR 1997 .................................................................... 43
ANNEX 8: MAXIMUM FLOOD EXTENT FOR 1998 .................................................................... 44
ANNEX 9: MAXIMUM FLOOD EXTENT FOR 1999 .................................................................... 45
ANNEX 10: MAXIMUM FLOOD EXTENT FOR 2001 .................................................................. 46
ANNEX 11: MAXIMUM FLOOD EXTENT FOR 2003 .................................................................. 47
ANNEX 12: MAXIMUM FLOOD EXTENT FOR 2005 .................................................................. 48
ANNEX 13: MAXIMUM FLOOD EXTENT FOR 2006 .................................................................. 49
ANNEX 14: MAXIMUM FLOOD EXTENT FOR 2007 .................................................................. 50
ANNEX 15: MAXIMUM FLOOD EXTENT FOR 2009 .................................................................. 51
ANNEX 16: MAXIMUM FLOOD EXTENT FOR 2013 .................................................................. 52
ANNEX 17: MAXIMUM FLOOD EXTENT FOR 2016 .................................................................. 53
ANNEX 18: MAXIMUM FLOOD EXTENT FOR 2017 .................................................................. 54
ANNEX 19: MAXIMUM FLOOD EXTENT FOR JUNE 2018 ....................................................... 55
ANNEX 20: MAXIMUM FLOOD EXTENT FOR AUGUST - SEPTEMBER 2018 ...................... 56
ANNEX 21: MAXIMUM FLOOD EXTENT FOR NOVEMBER 2018 .......................................... 57
ANNEX 22: ERROR MATRIX AND UNCERTAINTY AND CONFIDENCE ANALYSIS FOR
LCLU MAP ........................................................................................................................................ 58
ANNEX 23: NDWI FOR 15/09/2016 LANDSAT IMAGE .............................................................. 59
ANNEX 24: RFI ZONAL STATISTICS FOR KHARTOUM STATE COUNTIES ........................ 60
ANNEX 25: RFI DESCRIPTIVE STATISTICS FOR HEALTH AND EDUCATIONAL
FACILITIES IN KHARTOUM STATE ............................................................................................ 62
5
independen ly o cloud co e (Ma cus & Fons ad, 2008; Pul i en i e al., 2011;
Sca pino e al., 2018; Schumann e al., 2009), and he a ailabili y o SAR images
hanks o Sen inel mission (Pou sanidis & Ch ysoulakis, 2017).
2.2. Flood Risk Assessmen in Sudan
Wi hin he A ican con inen , Sudan is conside ed o be a lood-p one coun y
(Li e al., 2016). As a coun y associa ed wi h loods, i also exhibi s a pa e n o
ex eme lash loods, ollowing a ecu ing pa e n since he ea ly nine een hs and
wen ie h cen u ies (Da ies & Walsh, 1997). In hei s udy, (Su cli e e al., 1989;
Walsh e al., 1994) had also indica ed ha ex eme lood damage was no only
caused by he Nile lood in G ea e Kha oum a ea ( he main h ee ci ies in
Kha oum S a e; Kha oum, Kha oum No h – also e e ed o as Bah i, and
Omdu man), bu also due o he hea y ain s o ms ha las ed o 4-5 consecu i e
days causing a uno , and he epheme al wa e cou ses h oughou he a ea. One o
he ea lies analysis o he loods impac o e Sudan was ca ied ou by (Su cli e e
al., 1989) o he 1988 lood e en , whe e Me eosa Cold Cloud Du a ion (CCD)
measu emen s we e used along wi h ae ial pho og aphy. Recen s udies conduc ed
on lood haza d mapping in Sudan included geo-s a is ical da a analysis using
In e se Dis ance Weigh ing (IDW) me hod o ain all in si u-s a ions, as well as
quan i a i e in e iews o calcula e a social ulne abili y index o Kha oum S a e
o 2013 – 2014 loods (Mahmood e al., 2017), also he opinions o expe s on he
same lood e en s h ough a quali a i e app oach (Ho n & Elagib, 2018) which
sugges ed a lood managemen amewo k, ha included lood isk mapping as a
pa o da a esou ce enhancemen . The sho age in lood haza d mapping and lood
isk assessmen in Sudan, is he mo i e behind his esea ch.
E en hough apid lood mapping has been ca ied o e he las decade by he
Uni ed Na ion Ins i u e o T aining and Resea ch (UNITAR) p og am, UNITAR’s
Ope a ional Sa elli e Applica ions P og am (UNOSAT) in a semi eal- ime analysis
(Sudan maps | UNITAR, n.d.), Kha oum s a e s ills su e s om he consequences
o lash loods.
6
3. Me hods
Figu e 1 illus a es he me hodology ha was ollowed du ing his s udy. In he
low cha lood e en s a e mapped using Landsa 5, Landsa 7, and Landsa 8
images. Wi hin he lood e en , pe each Landsa image, a lood ex en map was
p oduced using image classi ica ion, and inally all lood ex en maps a e used o
calcula e he maximum lood ex en map pe lood e en . In his s udy he imeline
o mapping loods s a s om 1988 un il 2018, spanning 30 yea s. The maximum
lood ex en maps a e hen used o p oduce he lood haza d map using Rela i e
F equency o Inunda ion RFI Index. Fo lood isk assessmen a Land Co e Land
Use (LCLU) map is p oduced, he lood haza d map is analyzed using he LCLU
map as well as heal h acili ies and educa ional acili ies da a om Open S ee
Maps (OSM). The ollowing sec ions de ails all he s eps o he me hodology.
7
Figu e 1: Me hodology Flow Cha
8
3.1. S udy A ea
Sudan is a coun y loca ed in he no heas e n pa o he A ican con inen . The
coun y is bo de ed by Egyp , Libya, Cen al A ican Republic, Chad, Sou h Sudan,
E hiopia, E i ea, and he Red Sea.
Kha oum s a e is he mos popula ed s a e o he 18 s a es o Sudan, al hough
he smalles in e m o a ea. I s capi al is Kha oum ci y, which is also he na ional
capi al o Sudan. Kha oum is si ua ed be ween 31˚E and 35˚E longi ude and 15˚N
and 17˚N la i ude. Kha oum s a e coun s o 17% o he o al popula ion o Sudan,
Figu e 2 shows he bounda y o he s a e in addi ion o he s a e coun ies. The s a e
consis s o 7 coun ies:
• Kha oum
• Jebel Awliya
• Umdu man
• Oumbada
• Ka a i
• Bah i
• Sha g Alneel
Figu e 2: S udy A ea
9
The Nile i e , he wo ld’s longes i e is o med in he ci y o Kha oum, by
he joining o he Blue Nile ha o igina e om Tana Lake in E hiopia, and he
Whi e Nile ha o igina e om Vic o ia lake in Uganda. The s a e has an a id
clima e, wi h a ainy season om July o Sep embe . Kha oum has a his o y o
ecu en looding, plu ial and lu ial (Da ies & Walsh, 1997), and ex emely
ulne able due o i s high popula ion.
3.2. Da a
3.2.1. Remo ely Sensed Da a
Fo his esea ch 70 emo ely sensed images acqui ed Landsa di e en
missions we e eely ob ained o he pe iod 1988 – 2018 and downloaded om he
Uni ed S a es Geological Su ey (USGS) Ea hExplo e websi e
h ps://www.ea hexplo e .usgs.go , de ails abou spec al band speci ica ions o
each Landsa senso used a e a ailable in Annex 1. The images we e selec ed based
on he Da mou h Flood Obse a o y’s Ac i e A chi e o La ge Floods
(B aken idge, n.d.). A o al o 21 lood e en s we e eco ded. When acqui ing he,
images o he lood e en on Oc obe 1997 we e no a ailable on he Ea hExplo e
a chi e. Table 1 shows mo e de ails abou acqui ed images.
Table 1: Remo ely Sensed Images speci ica ions
Acquisi ion
da e
Pa h/Raw
Senso
Spa ial
esolu ion
Landsa
Numbe
o bands
25/08/1988
173/049
TM
30 m
Landsa 5
7
02/09/1988
173/049
TM
30 m
Landsa 4
7
10/09/1988
173/049
TM
30 m
Landsa 5
7
18/09/1988
173/049
TM
30 m
Landsa 4
7
18/08/1991
173/049
TM
30 m
Landsa 5
7
05/09/1992
173/049
TM
30 m
Landsa 5
7
23/08/1993
173/049
TM
30 m
Landsa 5
7
10/08/1994
173/049
TM
30 m
Landsa 5
7
26/08/1994
173/049
TM
30 m
Landsa 5
7
16/09/1996
173/049
TM
30 m
Landsa 5
7
02/08/1997
173/049
TM
30 m
Landsa 5
7
18/08/1997
173/049
TM
30 m
Landsa 5
7
03/09/1997
173/049
TM
30 m
Landsa 5
7
10
19/09/1997
173/049
TM
30 m
Landsa 5
7
05/10/1997
173/049
TM
30 m
Landsa 5
7
21/10/1997
173/049
TM
30 m
Landsa 5
7
06/09/1998
173/049
TM
30 m
Landsa 5
7
22/09/1998
173/049
TM
30 m
Landsa 5
7
08/08/1999
173/049
TM
30 m
Landsa 5
7
24/08/1999
173/049
TM
30 m
Landsa 5
7
09/09/1999
173/049
TM
30 m
Landsa 5
7
13/08/2001
173/049
TM
30 m
Landsa 5
7
29/08/2001
173/049
TM
30 m
Landsa 5
7
14/09/2001
173/049
TM
30 m
Landsa 5
7
11/08/2003
173/049
ETM+
30 m
Landsa 7
8
17/09/2005
173/049
ETM+
30 m
Landsa 7
8
19/08/2006
173/049
ETM+
30 m
Landsa 7
8
04/09/2006
173/049
ETM+
30 m
Landsa 7
8
20/09/2006
173/049
ETM+
30 m
Landsa 7
8
06/10/2006
173/049
ETM+
30 m
Landsa 7
8
07/09/2007
173/049
ETM+
30 m
Landsa 7
8
23/09/2007
173/049
ETM+
30 m
Landsa 7
8
09/10/2007
173/049
ETM+
30 m
Landsa 7
8
27/08/2009
173/049
ETM+
30 m
Landsa 7
8
12/09/2009
173/049
ETM+
30 m
Landsa 7
8
06/08/2013
173/049
ETM+
30 m
Landsa 7
8
07/09/2013
173/049
ETM+
30 m
Landsa 7
8
05/07/2016
173/049
OLI_TIRS
30 m
Landsa 8
11
13/07/2016
173/049
ETM+
30 m
Landsa 7
8
21/07/2016
173/049
OLI_TIRS
30 m
Landsa 8
11
29/07/2016
173/049
ETM+
30 m
Landsa 7
8
06/08/2016
173/049
OLI_TIRS
30 m
Landsa 8
11
14/08/2016
173/049
ETM+
30 m
Landsa 7
8
22/08/2016
173/049
OLI_TIRS
30 m
Landsa 8
11
30/08/2016
173/049
ETM+
30 m
Landsa 7
8
07/09/2016
173/049
OLI_TIRS
30 m
Landsa 8
11
15/09/2016
173/049
ETM+
30 m
Landsa 7
8
23/09/2016
173/049
OLI_TIRS
30 m
Landsa 8
11
25/08/2017
173/049
OLI_TIRS
30 m
Landsa 8
11
02/09/2017
173/049
ETM+
30 m
Landsa 7
8
10/09/2017
173/049
OLI_TIRS
30 m
Landsa 8
11
18/09/2017
173/049
ETM+
30 m
Landsa 7
8
26/09/2017
173/049
OLI_TIRS
30 m
Landsa 8
11
17/06/2018
173/049
ETM+
30 m
Landsa 7
8
25/06/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
11
03/07/2018
173/049
ETM+
30 m
Landsa 7
8
11/07/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
19/07/2018
173/049
ETM+
30 m
Landsa 7
8
27/07/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
04/08/2018
173/049
ETM+
30 m
Landsa 7
8
20/08/2018
173/049
ETM+
30 m
Landsa 7
8
12/08/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
28/08/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
05/09/2018
173/049
ETM+
30 m
Landsa 7
8
13/09/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
21/09/2018
173/049
ETM+
30 m
Landsa 7
8
29/09/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
08/11/2018
173/049
ETM+
30 m
Landsa 7
8
16/11/2018
173/049
OLI_TIRS
30 m
Landsa 8
11
24/11/2018
173/049
ETM+
30 m
Landsa 7
8
3.2.1. Auxilia y Da a
Heal h acili ies and educa ional acili ies ac oss he coun y we e a ailable by
OSM h ough Uni ed Na ions O ice o he Coo dina ion o Humani a ian A ai s
(UNOCHA)’s humani a ian esponse online po al
h ps://www.humani a ian esponse.in o.
OSM p o ides eely a ailable c owdsou ced spa ial da a, hese da a hough may
no be comp ehensi e, ye a e egula ly upda ed.
3.3. Mapping Flood Ex en
In o de o p oduce lood ex en maps Landsa images we e classi ied o he
h ee ca ego ies sugges ed by (Skakun e al., 2014); wa e o he h ee Nile i e s
and looded a eas, no wa e o all d y lands, and no da a o a eas co e ed wi h
clouds, cloud shadows o missing da a. The alues co esponding o each class is
illus a ed in Table 2, his in o ma ion is impe a i e o unde s anding u he wo k.
Since images we e no cloud ee, aining samples had o be collec ed pe image
o he h ee classes.
A e wa ds, he T ain Suppo Vec o Machine Classi ie ool was ained
using Landsa image and aining samples shape ile. The ou pu is a Classi ie
De ini ion ile. A e ha he classi ie de ini ion ile is used as an inpu in addi ion
12
o he Landsa image o he Classi y Ras e ool, which ou pu is a classi ied
image, ha is he lood ex en map. The ools a e wi hin he Segmen a ion and
Classi ica ion oolbox in he Spa ial Analys Tools in A cMap 10.6.
Table 2: Classes and he co esponding alues
Class Name
Value
Wa e
0
No Wa e
1
No Da a
2
3.4. Mapping Maximum Flood Ex en
A e ob aining a map o lood ex en o all he acqui ed images, a maximum
lood ex en map is p oduced o e e y lood e en , in which lood ex en maps
wi hin he lood e en a e used. E e y pixel on he map is assigned one o he h ee
classes men ioned abo e. A pixel is assigned wa e i a leas he same pixel was
assigned wa e in one o he lood ex en maps, i is assigned no wa e i he same
pixel was assigned as no wa e in all he lood ex en maps, and inally i is assigned
no da a i i was in all he lood ex en maps we e assigned no wa e o no da a.
In o de o calcula e he maximum lood ex en o each lood e en , Ras e
Calcula o ool in he Map Algeb a oolbox was used. A simple mul iplica ion o all
lood ex en maps was pe o med o ob ain an in e media e map wi h wa e class,
no wa e class and o he classes o alues mul iples o he numbe 2. The maximum
lood ex en map is inally p oduced a e assigning all o he classes o he no da a
class alue 2, by using Reclassi y ool in he Reclass oolbox. Bo h he Map
Algeb a oolbox and Reclass oolbox a e in he Spa ial Analys Tools in A cMap
10.6.
13
3.5. Rela i e F equency o Inunda ion
In s a is ics, he equency o an e en is he numbe o imes his e en has
occu ed in a da ase , and ela i e equency is he a io o he equency o an e en
occu ing in a da ase o he numbe o all e en s occu ing in he same da ase .
F om his we can induce ha he equency o inunda ion is o be calcula ed pe
pixel om he maximum lood ex en maps p oduced o all lood e en s. The
equency o inunda ion equals he numbe o imes a pixel was classi ied as wa e .
In o de o calcula e he ela i e equency o inunda ion, he equency o
inunda ion pe pixel is he numbe o imes i was classi ied as wa e di ided by he
o al numbe o imes he pixel was classi ied as wa e and no wa e . To calcula e
RFI alue i s a cons an as e named Wa e cons an as e wi h 0 alue was
c ea ed, ano he cons an as e was c ea ed, named No Wa e cons an as e and i s
alue was 1, hese wo as e iles we e c ea ed using C ea e Cons an Ras e ool in
he Ras e C ea ion Toolbox. A e ha he ool Equal To F equency in he Local
oolbox was used o calcula e he Wa e F equency Ras e and No Wa e F equency
Ras e , in each u n one o he cons an as e iles was used in addi ion o all
maximum lood ex en maps. A o al equency as e was c ea ed using he Ras e
Calcula o ool by adding he wa e and no wa e equency as e iles o each
o he . In o de o calcula e he RFI alue any pixel in he wa e equency ile equal
o 0 was se o null using he Se Null ool in he Condi ional Toolbox. Las ly, in
o de o ob ain he RFI map he Ras e Calcula o ool was used one mo e ime o
di ide he wa e equency as e by he o al equency as e , all he ools used
eside in he Spa ial Analys Tools in A cMap 10.6. The p ocess is illus a ed in
Figu e 3.
14
Figu e 3: RFI Mapping P ocess
3.6. Flood Risk Assessmen
3.6.1. LCLU Map
Since he e we e no LULC maps o he s udy egion eadily a ailable,
we p oduced one as desc ibed below. The LULC map o A ica disclosed
by FAO
(h p://www. ao.o g/geone wo k/s /en/main.sea ch? i le=land%20co e ;
e ie ed 10 Janua y 2020) is e y coa se. (Salman e al., 2008) published a
LULC map o g ea e Kha oum-Sudan, bu i s digi al e sion is no
a ailable, and he au ho s did no answe o he eques o sha e i in due
ime.
The Kha oum s a e LCLU map was p oduced o iden i y lood p one a eas, o
be used la e along wi h he lood haza d map p oduced om he RFI index. The
LCLU classes we e inspi ed om (B ox on e al., 2014)’s LCLU Map o A ica.
The h ee main classes we e Wa e Bodies, U ban, C opland and Vege a ion, in
addi ion o hese classes, he emaining un-u banized a eas ha a e a om he
i e s’ banks we e classi ied as Ba en Land.
In his s ep a Landsa 8 image om 28/12/2016 was acqui ed and clipped in o
he s udy a ea ex en . In o de o ob ain high accu acy o classi ica ion he
21
Figu e 6: Kha oum S a e Flood Haza d Map using he ela i e equency o inunda ion (RFI) de i ed
om Landsa images acqui ed om 1988 o 2018
This map, along wi h he LCLU map o he s a e, and poin ea u es o heal h
acili ies and educa ional acili ies, a e used o assess he deg ee o haza d in a eas
o Kha oum s a e in sec ions below. The signi icance o his map elies on i s use
as a base o u he in ensi e lood isk assessmen o he s a e.
4.4. Flood Risk Assessmen
4.4.1. LCLU Map
In his s ep he LCLU map and he lood haza d map a e used o assess he isk
o loods on Kha oum S a e. The LCLU map is shown in Figu e 7. To assess he
accu acy o he map, Google Ea h® was used as he g ound u h. The o e all
accu acy o he map is 84.2%.
22
Figu e 7: Kha oum S a e LCLU Map
De ailed accu acy assessmen esul s a e shown in Annex 22. The na u e o he
s a e’s a id clima e and u ban s uc u e ha can be seen in Figu e 8, as well as he
mode a e esolu ion (30 m) o Landsa images, led o con usion be ween U ban,
C opland/Vege a ion, and Ba en Land classes. Bu when compa ed o Open S ee
Maps OSM da a as shown in Figu e 9 and Figu e 10, he bounda ies o u ban a eas
o s a e coun ies in ela ion o he oad ne wo k da a we e sa is ac o y.
23
Figu e 8: U ban S uc u e in Aljazee a Eslang, Ka a i Coun y (sou ce: Google Ea h®)
Figu e 9: Kha oum Coun y oad ne wo k (Open S ee Maps OSM) combined wi h LCLU map
24
Figu e 10: Sha g Alneel and Jebel Awlyia coun ies oad ne wo k (Open S ee Maps OSM) combined
wi h LCLU map
4.4.2. Zonal S a is ics
The esul s o he zonal s a is ics o each coun y is shown in Annex 24, in
which he LCLU map was used o zoning. Wi h ega ds o RFI alues o U ban
a eas, Umdu man a e aged 0.23 ± 0.24 as he highes a e age be ween all se en
coun ies, Kha oum and Jebel a e aged 0.21 ± 0.16 and 0.21 ± 0.15 espec i ely.
Ombadah had he lowes a e age o 0.12 ± 0.06. In a eas o c opland and
ege a ions, Umdu man a e aged he highes be ween coun ies wi h 0.66 ± 0.32,
coming a e ha , Ka a i and Kha oum wi h 0.64 ± 0.27 and 0.52 ± 0.35
espec i ely. The lowes RFI a e age o c oplands and ege a ions a ea was also in
Sha g Alneel wi h 0.23 ± 0.17. Ombadah coun y anked a lowe RFI o c oplands
and ege a ions a ea bu was no conside ed because c oplands and ege a ion
ep esen only 1% o coun y a ea.
Since he esul s o zonal s a is ics o heal h and educa ional acili ies we e pe
poin , he a e age RFI alue o each acili y was he used o de i e desc ip i e
s a is ics. Ou o 715 heal h acili ies in he s a e o Kha oum, a o al o 511 heal h
25
acili ies had RFI alue associa ed wi h hem, a e aging 0.19 ± 0.09. As o he
educa ional acili ies, ou o 216 acili ies, a o al o 165 acili ies wi h RFI a e age
alue 0.18 ± 0.09 we e analyzed. De ailed esul s a e shown in Annex 25.
4.5. Humani a ian Mapping Response o Augus 2013 and July
2016 Flood E en s
In Augus 2013, hea y ains ook on se e al s a es o Sudan leading o lash
loods, along wi h Nile i e looding. The c isis led o he dea h o 45 people and
o e 70 inju ed. The se e e damages in in as uc u e has a ec ed a ound 150,000
people ac oss he coun y. Kha oum s a e, being he mos popula ed s a e in Sudan,
was he mos a ec ed, wi h mo e han 84,000 people a ec ed by he loods,
acco ding o epo s by he Wo ld Heal h O ganiza ion (WHO)
(h ps://www.who.in /hac/c ises/sdn/sudan loods2013si ep2.pd ; e ie ed 24
Janua y 2020) and UNOCHA
(h ps://disas e scha e .o g/documen s/10180/13939/OCHASudanFlashUpda eFloo
ds4.pd /a d50da8-cdac-4d22-a906-073e87e47876? e sion=1.0; e ie ed 24
Janua y 2020).
In esponse o a eques by UNITAR/UNOSAT on behal o UNDP o ice in
Sudan, he In e na ional Cha e Space and Majo Disas e s was ac i a ed o
p o ide sa elli e image y, in o de o assess in elie wo k. The cha e is a
collabo a ion be ween 17 space agencies and space esea ch ins i u es. In addi ion
o he cha e membe s, 19 o he en i ies con ibu e o he cha e in e ms o
disas e moni o ing, sa elli e image p o ision, and image analysis and maps
p oduc ion.
Along wi h he cha e , he Da mou h Flood Obse a o y also con ibu ed in
delinea ing he lood e en h ough he analysis o MODIS Te a sa elli e images.
Figu e 11 illus a es wo k done by UNOSAT h ough he cha e and he
obse a o y. I can be clea ha when compa ed o he maximum lood ex en map
p oduced using Landsa 7/Landsa 8 o he same lood e en in Annex 16, he
lood ex en in u ban and c oplands and ege a ions a eas is exceedingly la ge han
wha was cap u ed h ough he Landsa images. This is due o he na u e o he
26
di e en sa elli es used in he wo maps, since Fo mosa 2 and Rada sa 2 bo h
p oduce SAR images ha a e cloud ee. In addi ion, he images we e wi hin days
o he lood e en . Landsa images used in his s udy we e o e 30 days apa , hence
hey we e no able o p ope ly de ec he maximum lood ex en o his e en .
Figu e 11: UNOSAT Flood Ex en o 2013 lood e en using images om Fo mosa 2, Rada sa 2, and
MODIS Te a
As can be seen in he igu e abo e, al hough Sha g Alneel and Bah i coun ies
a e mos inunda ed coun ies, ye mos o hese a eas a e ba en. When o e laying
ba en a ea on op o he lood ex en , he a ec wi h ega ds u ban and c opland
and ege a ion a eas, which is illus a ed in Figu e 12 below. . The maximum lood
ex en map ob ained by Landsa images a e o e laid he cha e and Da mou h
Flood obse a o y lood ex en in Figu e 13 below.
27
Figu e 12: UNOSAT Flood Ex en o 2013 lood e en wi h o e laying ba en a ea
Figu e 13: UNOSAT Flood Ex en o 2013 lood and MODIS Te a wi h he maximum lood ex en o
augus 2013 by Landsa 7, Landsa 8.
28
In 2016 an ea ly ain season s a ed in June wi h hea y ains a ec ing 80,000
people na ionwide, causing loods in 13 ou o 18 s a es. Du ing hese loods
Kha oum s a e was no he mos a ec ed a ea, hence he e was no ex ensi e
mapping o loods o he s a e. The cha e was no ac i a ed o he s a e un il he
9 h o Augus . Images we e analyzed by UNOSAT ocused on he lood ex en on
he banks o he Whi e Nile and Blue Nile i e s. Sa elli e images used o mapping
we e Te aSAR-X o 3 m esolu ion, and SPOT-7 o 1.5 m esolu ion. Bo h maps
a e illus a ed in Figu e 14 and Figu e 15 below. The maximum lood ex en map
o 2016 is based on 11 images collec ed be ween July 5 h and Sep embe 23 d.
Hence, i cap u es almos he whole e en , al hough i s spa ial esolu ion is coa se .
Because o he ex en o he images, a s a e wise compa ison wi h he maximum
lood ex en map o 2016 (Annex 17) is no possible.
Figu e 14: UNOSAT Flood Ex en map on 11 Augus 2016 using Te aSAR-X sa elli e image
29
Figu e 15: UNOSAT Flood Ex en map on 15 Augus 2016 using SPOT-7 sa elli e image
UNITAR has an online po al o hei lood analysis p oduc s unde he
ollowing URL h p:// loods.unosa .o g/geopo al/. The po al is a aluable sou ce
o geospa ial in o ma ion, wi h one downside o he exclusion o o iginal images,
due o copy igh s, which does no allow o analysis g ow h o esea ch pu poses.
30
5. Conclusion
This s udy aimed a mapping lood haza d and assessing lood isk using mul i-
empo al sa elli e image y. Using images om Landsa 4,5,7 and 8, and h ough
speci ic objec i es his s udy was able o iden i y a eas - on coun y le el - ha a e
exposed o loods in Kha oum s a e, Sudan. A eas we e analyzed based on hei
use, and we e and classi ied o u ban, c opland and ege a ion, ba en, and wa e
bodies using a LCLU map ha was p oduced as a pa o he s udy.
Flood haza d was iden i ied using he ela i e equency o inunda ion RFI. By
mapping he RFI indica o , his s udy wen a s ep o wa d in lood isk assessmen
o he s a e o Kha oum, Sudan. A s ep ha was sugges ed by (Ho n & Elagib,
2018) in hei lood managemen amewo k o he capi al ci y Kha oum. This
map can also be used as he haza d ac o when calcula ing lood isk.
Using zonal s a is ics, i was ound ha he g ea es haza d was associa ed wi h
c oplands, due o hei p oximi y o i e banks, anking Umdu man and Ka a i
wi h he highes RFI alues. As o u ban a eas, Jebel Awlyia and Kha oum
coun ies, he mos u banized coun ies had he highes RFI alues. When haza d
deg ee was in es iga ed o public acili ies such as heal h and educa ional
acili ies, hese acili ies we e ound o be wi hin he u ban a eas haza d le el.
Heal h and educa ional acili ies lood haza d a e aged 0.19 and 0.18
consecu i ely, which was calcula ed o 511 heal h acili ies and 216 educa ional
acili ies wi h no RFI ze o alue.
The absence o a Land Co e Land Use LCLU map wi h a mode a e esolu ion
o Kha oum s a e has led o addi ional con ibu ion o his s udy. A LCLU map
was p oduced wi h an o e all accu acy o 84%, whe e he s a e was classi ied o as
u ban, c oplands and ege a ion, ba en, and wa e bodies. This map was an
essen ial componen in de ec ing u ban and c opland a eas wi h high exposu e o
loods.
The a ailabili y o sa elli e images wi h mode a e spa ial esolu ion o delinea e
maximum lood ex en was an essen ial componen o he s udy. The quali y o
acqui ed images, be i he ime o acquisi ion wi h ega ds o he lood e en o he
37
Table 3: Landsa 8 (OLI - TIRS) bands speci ica ions (Sou ce: U.S. Geological Su ey)
Bands
Wa eleng h (µm)
Resolu ion (m)
Band 1 - Coas al ae osol
0.43 – 0.45
30
Band 2 – Blue
0.45 – 0.51
30
Band 3 – G een
0.53 – 0.59
30
Band 4 – Red
0.64 – 0.67
30
Band 5 – Nea In a ed (NIR)
0.85 – 0.88
30
Band 6 – SWIR 1
1.57 – 1.65
30
Band 7 – SWIR 2
Band 8 - Panch oma ic
Band 9 - Ci us
Band 10 – The mal In a ed (TIRS) 1
Band 11 – The mal In a ed (TIRS) 2
2.11 – 2.29
0.50 – 0.68
1.36 – 1.38
10.6 – 11.19
11.50 – 12.51
30
15
30
100
100
38
Annex 2: Maximum Flood Ex en Map o 1991
Table 1: Change ma ix o 1991 lood e en
1991
Wa e
No Wa e
LCLU
U ban
0.96%
99.04%
Wa e Bodies
95.05%
4.95%
C opland/Vege a ion
10.00%
90.00%
Ba en
0.35%
99.65%
39
Annex 3: Maximum Flood Ex en o 1992
Table 1: Change ma ix o 1992 lood e en
1992
Wa e
No Wa e
No Da a
LCLU
U ban
1.60%
67.31%
31.09%
Wa e Bodies
82.15%
5.77%
12.1%
C opland/Vege a ion
12.02%
71.61%
16.38%
Ba en
0.54%
63.19%
36.27%
40
Annex 4: Maximum Flood Ex en o 1993
Table 1: Change ma ix o 1993 lood e en
1993
Wa e
No Wa e
No Da a
LCLU
U ban
1.24%
54.36%
44.40%
Wa e Bodies
94.71%
1.40%
3.9%
C opland/Vege a ion
9.29%
34.74%
55.98%
Ba en
0.86%
83.95%
15.19%
41
Annex 5: Maximum Flood Ex en o 1994
Table 1: Change ma ix o 1994 lood e en
1994
Wa e
No Wa e
No Da a
LCLU
U ban
4.01%
17.84%
78.15%
Wa e Bodies
72.26%
0.19%
27.6%
C opland/Vege a ion
12.75%
3.66%
83.59%
Ba en
1.05%
19.65%
79.30%
42
Annex 6: Maximum Flood Ex en o 1996
Table 1: Change ma ix o 1996 lood e en
1996
Wa e
No Wa e
No Da a
LCLU
U ban
2.29%
94.17%
3.55%
Wa e Bodies
95.50%
4.36%
0.1%
C opland/Vege a ion
9.08%
90.49%
0.44%
Ba en
0.13%
99.26%
0.61%
43
Annex 7: Maximum Flood Ex en o 1997
Table 1: Change ma ix o 1997 lood e en
1997
Wa e
No Wa e
No Da a
LCLU
U ban
1.23%
7.95%
90.83%
Wa e Bodies
95.13%
0.26%
4.6%
C opland/Vege a ion
7.49%
5.01%
87.50%
Ba en
3.88%
17.58%
78.55%
44
Annex 8: Maximum Flood Ex en o 1998
Table 1: Change ma ix o 1998 lood e en
1998
Wa e
No Wa e
No Da a
LCLU
U ban
7.07%
34.03%
58.89%
Wa e Bodies
98.78%
0.57%
0.7%
C opland/Vege a ion
36.83%
35.17%
28.00%
Ba en
2.05%
58.54%
39.42%
45
Annex 9: Maximum Flood Ex en o 1999
Table 1: Change ma ix o 1999 lood e en
1999
Wa e
No Wa e
No Da a
LCLU
U ban
13.83%
6.13%
80.05%
Wa e Bodies
90.46%
0.16%
9.4%
C opland/Vege a ion
32.25%
7.61%
60.14%
Ba en
10.40%
16.89%
72.71%
46
Annex 10: Maximum Flood Ex en o 2001
Table 1: Change ma ix o 2001 lood e en
2001
Wa e
No Wa e
No Da a
LCLU
U ban
1.40%
96.11%
2.49%
Wa e Bodies
97.60%
1.25%
1.1%
C opland/Vege a ion
10.44%
69.13%
20.43%
Ba en
0.03%
92.73%
7.23%
53
Annex 17: Maximum Flood Ex en o 2016
Table 1: Change ma ix o 2016 lood e en
2016
Wa e
No Wa e
No Da a
LCLU
U ban
8.83%
0.50%
90.67%
Wa e Bodies
99.93%
0.00%
0.1%
C opland/Vege a ion
24.92%
0.12%
74.96%
Ba en
4.25%
3.39%
92.36%
54
Annex 18: Maximum Flood Ex en o 2017
Table 1: Change ma ix o 2017 lood e en
2017
Wa e
No Wa e
No Da a
LCLU
U ban
6.93%
30.41%
62.67%
Wa e Bodies
99.86%
0.01%
0.1%
C opland/Vege a ion
26.11%
22.06%
51.83%
Ba en
3.42%
46.99%
49.59%
55
Annex 19: Maximum Flood Ex en o June 2018
Table 1: Change ma ix o June 2018 lood e en
June 2018
Wa e
No Wa e
No Da a
LCLU
U ban
5.91%
11.99%
82.10%
Wa e Bodies
98.57%
0.11%
1.3%
C opland/Vege a ion
13.87%
16.32%
69.81%
Ba en
4.30%
30.91%
64.79%
56
Annex 20: Maximum Flood Ex en o Augus -
Sep embe 2018
Table 1: Change ma ix o Augus 2018 lood e en
Augus 2018
Wa e
No Wa e
No Da a
LCLU
U ban
10.35%
16.76%
72.89%
Wa e Bodies
99.91%
0.01%
0.1%
C opland/Vege a ion
35.17%
8.32%
56.51%
Ba en
7.01%
23.38%
69.61%
57
Annex 21: Maximum Flood Ex en o No embe
2018
Table 1: Change ma ix o No embe 2018 lood e en
No embe 2018
Wa e
No Wa e
No Da a
LCLU
U ban
4.91%
59.54%
35.54%
Wa e Bodies
99.54%
0.29%
0.2%
C opland/Vege a ion
11.83%
59.51%
28.65%
Ba en
2.49%
67.94%
29.57%
58
Annex 22: E o Ma ix and Unce ain y and
Con idence Analysis o LCLU Map
Table 1: Accu acy assessmen esul s o LCLU Map
Table 2: Con idence analysis o accu acy assessmen o LCLU Map
Use ’s
P oduce ’s
U ban
0.7±0.16
0.91±0
Wa e Bodies
1±0
1±0
C opland/Vege a ion
0.9±0.11
0.77±0.14
Ba en Land
0.77±0.15
0.72±0.16
O e all Con idence
0.84 ± 0.07
Use ’s
P oduce ’s
U ban
70.0%
91.0%
Wa e Bodies
100.0%
100.0%
C opland/Vege a ion
90.0%
77.0%
Ba en Land
77.0%
72.0%
O e all Accu acy
84.2%
59
Annex 23: NDWI o 15/09/2016 Landsa image
60
Annex 24: RFI Zonal S a is ics o Kha oum S a e
coun ies
Bah i
MINIMUM
MAXIMUM
MEAN
STANDARD DEVIATION
U ban
0.06
1
0.15
0.10
Wa e Bodies
0.08
1
0.95
0.11
C opland/Vege a ed Land
0.06
1
0.40
0.28
Ba en Land
0.05
1
0.11
0.06
Sha g Alneel
MINIMUM
MAXIMUM
MEAN
STANDARD
DEVIATION
U ban
0.06
1
0.18
0.14
Wa e Bodies
0.07
1
0.94
0.14
C opland/Vege a ed Land
0.06
1
0.23
0.17
Ba en Land
0.05
1
0.12
0.09
Jebel Awlyia
MINIMUM
MAXIMUM
MEAN
STANDARD DEVIATION
U ban
0.06
1
0.21
0.15
Wa e Bodies
0.08
1
1.00
0.05
C opland/Vege a ed Land
0.06
1
0.29
0.21
Ba en Land
0.07
1
0.35
0.25
Kha oum
MINIMUM
MAXIMUM
MEAN
STANDARD DEVIATION
U ban
0.07
1
0.21
0.16
Wa e Bodies
0.23
1
0.99
0.05
C opland/Vege a ed Land
0.06
1
0.52
0.35
Ba en Land
0.08
1
0.40
0.32
Umdu man
MINIMUM
MAXIMUM
MEAN
STANDARD DEVIATION
U ban
0.06
1
0.23
0.24
Wa e Bodies
0.15
1
0.99
0.05
C opland/Vege a ed Land
0.06
1
0.66
0.32
Ba en Land
0.06
1
0.13
0.08
61
Ombadah
MINIMUM
MAXIMUM
MEAN
STANDARD DEVIATION
U ban
0.06
1
0.12
0.06
Wa e Bodies
0.10
0.94
0.65
0.27
C opland/Vege a ed Land
0.06
0.85
0.13
0.07
Ba en Land
0.06
1
0.18
0.13
Ka a i
MINIMUM
MAXIMUM
MEAN
STANDARD DEVIATION
U ban
0.06
1
0.15
0.13
Wa e Bodies
0.12
1
0.98
0.06
C opland/Vege a ed Land
0.06
1
0.64
0.27
Ba en Land
0.06
1
0.15
0.11
62
Annex 25: RFI Desc ip i e S a is ics o Heal h and
Educa ional Facili ies in Kha oum S a e
Heal h Facili ies
Mean
0.19
S anda d E o
0.00
Median
0.18
Mode
0.11
S anda d De ia ion
0.09
Sample Va iance
0.01
Ku osis
0.53
Skewness
0.93
Range
0.47
Minimum
0.08
Maximum
0.55
Sum
97.48
Coun
511.00
Educa ional Facili ies
Mean
0.18
S anda d E o
0.01
Median
0.17
Mode
0.18
S anda d De ia ion
0.09
Sample Va iance
0.01
Ku osis
2.54
Skewness
1.33
Range
0.52
Minimum
0.08
Maximum
0.60
Sum
29.83
Coun
165.00