Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Title Distributed Hydrological Modeling at Wadi Samail, Oman
Author(s) Abdel-Fattah, Mohammed; Kantoush, Sameh; Saber,Mohamed; Sumi, Tetsuya
CitationProceedings of the Second International Symposium on FlashFloods in Wadi Systems: Disaster Risk Reduction and WaterHarvesting in the Arab Region (2016): 54-58
Issue Date 2016-10
URL http://hdl.handle.net/2433/228078
Right
Type Presentation
Textversion author
Kyoto University
Distributed Hydrological Modeling at Wadi Samail, Oman Mohammed Abdel-Fattah1,*, Sameh Kantoush1, Mohamed Saber1, Tetsuya Sumi1
1Graduate school of engineering, Kyoto University, Japan 2Disaster prevention institute, Kyoto University, Japan *Corresponding author
Email: [email protected]
Keywords: Flash Flood, Wadi, Hydro-BEAM, RRI
Recently, flash floods are frequently occurring in the arid region as Oman, which counter with
various challenges to the management of wadi flash floods. In the past, Oman hit by cyclone
Gonu in June 2007 causing torrential flooding and severe damages where the economic loss
was about 4 billion USD, as well as nearly 50 deaths. Mitigation measures and warning system
have become more critical given the expected increased extreme events due to climate changes.
Oman is an arid country, where the average annual rainfall, in its capital Muscat, is only 100
mm, while the average of the whole country is only 51 mm/yr varying from less than 20 mm/yr
in the internal desert regions to over 350 mm/yr in the mountain areas. Wadi Samail at the
coastal area of Oman is selected as case study for flash flood hydrological modelling. Rainfall–runoff responses predictions in arid climate as wadi system always presents unique challenges.
One of the main challenges beside data limitation is the hydrological models themselves, where
the majority of models developed for catchments that have different characteristics than other
wadi systems. Hence, the need to evaluate the suitability of alternative modelling approaches
for wadi system and its scarce dataset arises. In that regard, two distributed hydrological models
are selected in this study. The first one is the Hydrological River Basin Environmental
Assessment Model (Hydro-BEAM) and the other is the Rainfall-Runoff-Inundation (RRI)
Model. Another aspect of this contribution is to focus on both structural measures for flash
flood retention as dry dams, and water harvesting. The location of such mitigation structures
must be carefully designed to avoid transferring the problems to the developed downstream
area of the wadi. Moreover, mitigation structures should be designed in a coordinated manner,
to assess their overall effect. This study analyzes the wadi flash flood mitigation for three cases:
1) no dams, 2) distributed small dams all over the catchment in the upstream, and 3) proposed
large dam in the middle or downstream area of the wadi. The effect must be quantified through
a comparison of the consequences with and without mitigation structures over the whole wadi.
Various factors are considered to study and improve the assessment methodologies. The
simulated scenarios highlighted significant differences in calculated hydrographs when using
either distributed or concentrated dams scenarios for wadi Samail. This study is expected to
conclude recommendations for hydrological modelling and management at wadi system in arid
environments. The next questions address how to define dam height and reservoir volume. For
better assessment of several dams options, clear quantification of evaluation factors and cost-
benefit approaches should be included in future. Small dry dams are effective structures. Both,
Hydro-BEAM and RRI models are efficient, and emphasizes the importance of taking into
account the variability and spatial properties of rainfall patterns.
54 25-27 October 2016
The Second International Symposium on Flash Floods in Wadi Systems
Distributed Hydrological Modeling at Wadi Samail,
Oman
M. ABDEL-FATTAH 1, Sameh KANTOUSH2, Mohamed SABER2,Tetsuya SUMI2
1 Graduate School of Engineering, Kyoto University
2 WRRC, DPRI, Kyoto University
1
2nd ISFF
Oct., 2016
• Wadi: Arabic term referring to valley.
• Wadi channel is usually dry except during heavy rain events.
• Flash flood: caused by heavy rainfall in short duration< 6 hrs.24/02/2016
2
Wadi System& Flash Floods
Characteristics of Wadi
24/02/2016 3
0
50
100
150
200
250
300
350
400
Dis
cha
rge
(m
3/s
)
SAMAIL CATCHMENT, WILAYAT AL SEEB
WADI AL KHOUDH
Dischareg(m3\s)
Source: A. Al Barwani, Flash Flood Mitigation and Harvesting Oman case study, First International Symposium on Flash
Floods, 2015, Kyoto, Japan
Damages of Flash Floods Disasters
24/02/2016 4
24/02/2016 5
Cyclones History in OmanDate Name& Type Rainfall (mm)
June 1890 Tropical Cyclones
(TC)
285
May 1963 TC 269
Nov 1966 TC 202
Dec 1971 Cyclonic storm 99
Jun 1977 TC 430
Mar 1999 Low pressure-Sur 122
Oct 1999 Deep Low Pressure 69
May 2002 Cyclonic storm 58
Sep 2004 Low Pressure 116
Jun, 2007 Gonu, TC 626
June, 2010 Phet, TC 603
June, 2015 Ashoba, TC 204
Flash Floods Caused By Cyclones
24/02/2016 6•Ryan L. Sriver, 2011
Gonu 2007 Cyclone
50 killed persons
4 billion USD losses
55 25-27 October 2016
The Second International Symposium on Flash Floods in Wadi Systems
26/10/2016 7
Target Research Area of W. Samail
W. Samail location map in Oman, its digital elevation model and the measurements stations.24/02/2016 8
24/02/2016 9
Satellite and Measured RainfallGSMAP Gauges
0
5
10
15
20
25
30
35
4:00
AM
5:00
AM
6:00
AM
7:00
AM
8:00
AM
9:00
AM
10:00
AM
11:00
AM
12:00
PM
1:00
PM
2:00
PM
3:00
PM
4:00
PM
5:00
PM
6:00
PM
7:00
PM
8:00
PM
9:00
PM
10:00
PM
Pre
cipit
atio
n (
mm
)
Time (h)
GSMap vs Measured Rainfall (Alkhoud station, Oman)
Meas. Rain GSMap GSMap av. Watershed
Gonu 2007 event
Study Motivation
24/02/2016 10
Wadi Flash Floods
Disaster
Unique Features
Water Resources
Hydrological Modelling
Frequency
Wadi
System
Data
Deficiency
Unsuitable
Models
Objectives
• Proposing innovative methodology for management
of flash floods disaster in ungauged wadis.
• Check the applicability of Hydro-BEAM and RRI
models using available data in W. Samail.
• Comparative study by 2 models Hydro-BEAM&RRI
• To find best scenario of DDR (distributed or
concentrated dams)
24/02/2016 11
Methodology
24/02/2016 12
Da
ta C
oll
ecti
on
& A
na
lysi
s
RRI Model
Fie
ld I
nves
tig
ati
on
Topographic Data Rainfall DataLand Use Data
GIS
& G
oo
gle
Ear
th
Dam Location Detection
Thiessen Interp.Watershed Delineation Hydro-shed Data
Surface RunoffInundation
Flash Flood Mitigation Scenarios
Model Calibration& Validation
Model Sensitivity Analysis
Hydrological Modeling
56 25-27 October 2016
The Second International Symposium on Flash Floods in Wadi Systems
RRI Model
• Rainfall-Runoff-Inundation (RRI) model is a two-dimensional model capable of simulating rainfall-runoffand flood inundation simultaneously (Sayama et al.,2012)
24/02/2016 13RRI model scheme overview (Sayama, T., 2013)
24/02/2016 14
(Sayama et al., 2012)
Mass
Balance eq.
Momentum eq.
h water height from local surface,
Q x, y unit width discharges,
u & v flow velocities,
r rainfall intensity,
H water height from the datum,
ρw water density,
g gravitational acceleration,
τ x and τ y shear stresses and
n Manning’s roughness
k lateral saturated hydraulic
conductivity
d soil depth times effective
porosity
Diffusion Wave approximation
RRI Model
24/02/2016 15
Samail_DEM
Elevation (m)
0 - 299
300 - 415
416 - 531
532 - 676
677 - 908
909 - 2,462
Ü
0 8 16 24 324Km
Topography
• 1 sec void filled SRTM data
Input Data Watershed Delineation
Watershed Delineation
Streams
Wadi Boundary Ü
0 8 16 24 324Km24/02/2016 16
DEM
Flow Direction
Flow Accumulat
ionStream Link
Watershed
Stream Order
Stream To Feature
Sink Check
Watershed modelling using GIS flowchart
RRI Parameter Setting
RRI model major parameter setting for the different land uses showing its default
values and ranges.
Parameter (default) Range Alluvium Igneous Sedimentary
nriver (0.03m-1/3s) 0.015~0.04 0.022 0.022 0.022
nslope (0.3 m-1/3s) 0.15 ~ 1.0 0.3 0.35 0.3
d (0.471 m) 0.15 ~ 1.0 - 0.14 0.3
k (0.1ms-1) 0.01-0.3 - 0.05 0.05
kv(5.56*10-7 ms-1) 6.54*10-5 ~ 1.6710-7 4*10-6 - -
φ 0.3 ~ 0.5 0.475 - -
Sf (0.3163 m) 0.0495~0.3163 0.15 - -
24/02/201617
RRI Model Sensitivity Analysis
24/02/2016 18
Over than 1000 simulations (manually)
57 25-27 October 2016
The Second International Symposium on Flash Floods in Wadi Systems
RRI Model Calibration (Gonu-2007)
24/02/2016 19
0
20
40
60
80
1000
200
400
600
800
1000
1200
1-Jun-07 4-Jun-07 7-Jun-07 10-Jun-07 13-Jun-07 16-Jun-07
Rain
fall
(m
m)
Dis
charg
e (m
3/s
)
Date
Ave. meas. rain (mm)
Meas. flow (cms)
Simulated flow (cms)
(a) (b)
6 June, 2007
1:00 PM
Fig. a) RRI model calibration results (Gonu-2007); b) peak flow spatial distribution.
Discharge
(m3/s)
24/02/2016 20
RRI Model Validation (Phet-2010)
0
20
40
60
80
1000
200
400
600
800
1000
1-Jun-10 4-Jun-10 7-Jun-10 10-Jun-10 13-Jun-10 16-Jun-10
Ra
infa
ll (
mm
)
Dis
ch
arg
e (m
3/s
)
Date
Ave. meas. rain (mm)
Meas. flow (cms)
Simulated flow (cms)
(a)
Fig. RRI model validation results (Phet-2007); b) peak flow spatial distribution.
(b)
4 June/2010
7:00 PM
Discharge
(m3/s)
Model Performance Evaluation
24/02/2016 21
( )
( )1
1
*100n
obs sim
i i
i
nobs
i
i
Y Y
PBIAS
Y
=
=
−=
∑
∑
( )
( )
2
1
2
1
1
nobs sim
i i
i
nobs obs
i mean
i
Y Y
NSE
Y Y
=
=
=−
−−
∑
∑
2 2 21 ( 1) ( 1) ( 1)KGE r α β− − − + −= +
( )( )( )
( ) ( )
2
2 1
2 2
1 1
nsim sim obs obs
i mean i mean
i
n nsim sim obs obs
i mean i mean
i i
Y Y Y Y
r
Y Y Y Y
=
= =
− −=
− −
∑
∑ ∑( )max max
max
*100obs sim
sim
Y YPE
Y=
−
Relative hydrograph peak (PE), Coefficient of determination (squared correlation coefficient) (r2),
Percent bias (PBIAS), Nash-Sutcliffe efficiency (NSE)1), Gupta efficiency (KGE) 2)
where r is the linear correlation, Yiobs and Yi
obs are observed and simulated discharge at time step i, Yiobs ,Yi
obs are the mean observed
and simulate discharge, α is a measure of relative variability and equal to the ratio of standard deviation of the simulated and
observed discharge and β represents the bias and equal to the ratio of the mean simulated and observed discharge.
1) Nash, J. E. and Jonh, V. S. : River flow forecasting through conceptual models part I—A discussion of principles, Journal of hydrology, Vol. 10, No. 3, pp. 282-290, 1970.
2) Gupta, H. V., Kling, H., Yilmaz, K. K. and Martinez, G. F. : Decomposition of the mean squared error and NSE performance criteria: Implications for improving
hydrological modelling, Journal of Hydrology, Vol. 377, No. 1, pp. 80-91, 2009.
RRI model performance evaluation for Gonu-2007 and
Phet-2010 extreme flash floods.
Index (Ideal value) Gonu 2007 Phet 2010
PE (0 %) -1.18 -9.05
r2 (1) 0.96 0.89
PBIAS (0%) -14.3 -12.04
NSE (1) 0.93 0.86
KGE (1) 0.81 0.74
24/02/2016 22
Long Term Flash Flood Simulation at W. Samail
0
50
100
150
200
250
300
350
400
Dec, 2005 Jul, 2006 Jan, 2007 Aug, 2007 Feb, 2008 Sep, 2008 Mar, 2009 Oct, 2009 May, 2010
Dis
cha
rge
(cm
s)
Date
Wadi Flow Simulation using RRI
Simulated Discharge
Measured Dischrage
0
100
200
300
400
6/4/2007 6/10/2007 6/16/2007 6/22/2007
Cyclone Gonu Event
Wadi flow measurement station
Flash Flood Inundation Simulation
24/02/2016 23
Discharge vs Inundation
Wadi Samail Dam
24/02/2016 24
Al-Maktoumi, 2014
Afer Gonu 2007, it was clear that This Dam is not
enough and further mitigation structure is needed
58 25-27 October 2016
The Second International Symposium on Flash Floods in Wadi Systems
Different FF Mitigation Scenarios at Wadi Samail
• Mitigation scenario 1: One big concentrated dam
• Mitigation scenario 2: Three smaller distributed dams
24/02/2016 25
Concentrated
Dam
Distributed Dam
1
Distributed Dam
2
Distributed Dam
3
Location: Y
: X
23.554344
58.103665
23.554344
58.103665
23.371503
58.090958
23.368906
58.157186
Reservoir Capacity
(MCM)
75 40 9.6 25.8
Height (m) 50 40 22 22
Proposed Flash Flood Mitigation Dams
24/02/2016 26
Concentrated DamDistributed Dam 1
Distributed Dam 3Distributed Dam 2
Different Mitigation Scenarios in W. Samail
24/02/2016 27
0
100
200
300
400
01. Jun 07 09. Jun 07 17. Jun 07 25. Jun 07
Dis
ch
arge (
cm
s)
Date
Simulated discharge with concentrated and distributed dam
Without dams
With one concentrated dam
With three distributed dams
Conclusion
• Wadi system have very unique features and should be considered in
flash flood hydrological modelling and management.
• RRI model could be calibrated validated efficiently to be used in
wadi system
• Distributed dams and concentrated dams strategies are efficient in
flash flood mitigation.
• Distributed dams have advantage of upstream protection, local
recharge
• More evaluating parameters for the best mitigation scenarios should
be considered as:
• Optimization of each function of dams
• Economical point of view and maintenance28
24/02/2016 29
59 25-27 October 2016
The Second International Symposium on Flash Floods in Wadi Systems