Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
1
Modelling Runoff with Satellite Data
Nyandwaro Gilbert Nyageikaro
Patrick Willems
Joel Kibiiy
2
Outline
• Background information
• SWAT model development
• The model sensitivity analysis and
calibration
• Calibration of satellite weather data
• Conclusions
3
• Both floods and droughts have negative effects…
• Floods cause destruction of property and
displacement of people.
• Hydrological modeling can help mitigate these
effects.
• Lack of sufficient data is a challenge
• Aim: to develop a calibration technique for
satellite-based weather data.
• Research based on nzoia basin (about 12,800km2)Western kenya.
Background Information
4
Background Information…
35°0'0"E
35°0'0"E
34°0'0"E
34°0'0"E
1°0'0"N 1°0'0"N
0°0'0" 0°0'0"
5
• Basin part of Lake Victoria of the larger Nile basin
• Located mainly in agricultural zone.
• Basin is mainly tropical humid.
• Mean annual rainfall varies from about 1070 mm to
2200 mm
• Mean monthly rainfall trend of two maxima over the
year: April-May and July-November respectively
• Altitudes: 4000m-1000m a.s.l
Background Information…
6
Hydrological challenges facing Nzoia basin:
– Flooding in the lower reach (popularly known as Budalangi area)
– Floods almost on annual Occurrence
– Breaching of dykes – old and poorly maintained
– Deaths and displacement of people; Destruction of property.
Background Information…
7
Background Information…
35°0'0"E
35°0'0"E
34°0'0"E
34°0'0"E
1°0'0"N 1°0'0"N
0°0'0" 0°0'0"
Floodplain
8
“Budalang'i residents edgy as river Nzoia breaks banks”
“People are worried because of the volumes of water which have already burst the River banks. Floods are likely to occur if the water overpowers the dykes,” ……
[Daily Nation, April 14, 2013]
Background Information…
9
2008 Flood: Destruction of property 2006 Flood: Damaged dyke
2006 Flood: Destruction of Property
2006 Flood: Displaced people
Fig.: Flooding at the Lower Reaches of Nzoia
10
2003 Flood
2008 Flood
2006 Flood: Destruction of Property
2006 Flood: Displacement of People
Fig.: Flooding at the Lower Reaches of Nzoia
11
Repair of the broken dyke in Budalang'i
12During a Survey to a breached dyke in Jan 2012
13
– Weak disaster management capacity
• Facilities
• Information
• Manpower
• Funding
– Lack of sufficient data
– Aim: To develop a calibration technique for satellite-based weather data.
Background Information…
14
• Development of Rainfall-Runoff model-SWAT
model for Nzoia basin.
• Model sensitivity analyses and calibration
• Calibration of satellite weather data
Methodology
15
• Watershed: 12, 800 sq.km.
• 27 Subbasins.
• 211 HRUs- Landuse, soil and slope
• 13 Rainfall stations
• Simulation: 1974-1984
Development of Nzoia SWAT Model
16
Development of Nzoia SWAT Model
Calibration Station
17
Sensitivity analysis
• Most sensitive parameters for calibration
• LH procedure, sensitivity of parameters dependent on interval used
• Rank….groups (sensitive and non-sensitive)
• Some parameters low in ranking….crucial (GW_DELAY)…judgement needed!
• Sensitive parameters: CN2, ESCO, Sol_AWC, Surlag, GWQMN, GW_REVAP, ALPHA_BF… and GW_DELAY
18
• Based on observed discharge at downstream station.
• Models runs for the period 1974-1984.
– Warm-up: 1974-1975
– Calibration: 1980-1984
– Validation: 1976-1979
• The models performance assessed through:
– NSE
– Validation of Maxima and Minima (Box-Cox transformations)…….
Model Calibration
[Box and Cox, 1964]
19
• Water Engineering Time Series PROcessing
tool (WETSPRO) for calibration (Willems,
2008).
• Multi-criteria objecive employed.
NSE
• NSE: 0.89 Calibration; 0.81 Validation
Model Calibration
20
Model Calibration…
0
50
100
150
200
250
300
350
400
450
29000 29500 30000 30500 31000 31500Number of time steps
OBS
Filtered baseflow
SWAT
General agreement:
21
Model Calibration…Maxima (peaks):
4
6
8
10
12
14
16
4 6 8 10 12 14 16
BC
( S
imu
late
d m
axim
a )
BC( Observed maxima )
SWAT
bisector
mean deviation
standard deviation
standard deviation
22
Model Calibration…
Cumulative Vol:
0
50000
100000
150000
200000
250000
0 500 1000 1500 2000
Cu
mu
lative d
ischarg
e
Time
OBS SWAT
23
Model Calibration…Extreme maxima:
0
50
100
150
200
250
300
350
400
450
0.1 1 10
Dis
ch
arg
e
Return period [years]
OBS SWAT
24
Model Calibration…Extreme minima:
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.1 1 10
1 / D
isch
arg
e
Return period [years]
OBS SWAT
25
• USGS RFE Satellite rainfall estimates for
calibration against raingause data.
• Resolution: 8km
• Satellite data extracted at 13 points (at
raingauge stations)
• Calibration: Quartile-pertubation approach at
individual stations
• Period: 2001-2008
– 2001-2004: Calibration of Satellite data
– 2005-2008: Validation
• Analysis: Mass curves Runoff extremes.
Calibration of satellite data
26
• USGS RFE Satellite rainfall estimates for
calibration against raingause data.
• Satellite data extracted at 13 points (at
raingauge stations)
• Calibration: Quartile-pertubation approach at
individual stations
• Analysis: Runoff extremes.
Calibration of satellite data
27
Calibration of satellite data…
0
100
200
300
400
500
600
0.1 1 10
Dis
ch
arg
e
Return period [years]
OBS
UNCALIB_RFE
CALB_RFE
28
• Satellite data can be valuable where observed
data is insufficient
• Calibration to local scale can better improve
runoff estimation with satellite data
• Effect of using both satellite rainfall and pet data
to be studied
Calibration of satellite data
29
THANK YOU
Suggestions and Questions