INDIAN INSTITUTE OF TECHNOLOGY ROORKEE

Assessment of Runoff Generation at Rift Valley

Lakes Basin of Ethiopia for present and future
climate scenario

Dr. Manoj K. Jain and Dr. Negash Wagesho

Department of Hydrology Urba Minch University
Indian Institute of Technology, Roorkee Ethiopia
INTRODUCTION
• The natural environment is under tremendous
stress as a consequence of various demands of
increasing population.
• Temporal and spatial patterns of precipitation and
extreme flows have been changing in the recent
past.
• Global water balance is highly influenced…
• Increased earth and sea surface temp., melting of
polar ice, rise in average global ocean level…
(Loaiciga et al., 1996; Singh et al., 1997; Barnett et al., 2005; Steele-
Dunne et al., 2008; IPCC,2007)
2
2
 Introduction…

• Avg. temp. rise of 0.6 ± 0.2 oC (1860-2000) versus

0.74°C ± 0.18°C (1906-2000) (IPCC,2001; 2007)
T = 0.07°C ± 0.02°C per decade

1906 1950 2005

T = 0.13°C ± 0.03°C
• Anthropogenic induced land use/cover changes have
transformed 1/3rd -1/4th of ice-free surface of our planet
(Vitousek, 1994; Vitousek et al., 1997).
• Land use/land cover changes significantly altered bulk
water yield from the watershed.
3
3
 Introduction…

• Land use changes are significant in tropical regions

(Piao et al. 2007).
• Currently, 1/4th of the population of Africa is facing
high water stress and this magnitude is expected to
increase (2-3 fold ) in the next 40 years (Boko et al.,
2007)
• Adaptation mechanisms developed by African
farmers are not sufficient enough to cope up with
current and future climate variability (IPCC, 2007)

4
4
 Introduction…
• The sub-Saharan region of Africa has been challenged by
both man-made and natural stresses …. (Tadross et al.,
2005; You and Ringler, 2010).
• Key Features - prolonged drought
- severe and unprecedented flood
- poor economic developments
- poor institutional developments
• Ethiopia is a victim of such global challenges
• Example: 48 flood and 12 drought events over the last
Century (EM-DAT, 2010).

5
5
 Introduction…
Ethiopia: Agricultural sector accounts ETHIOPIA

• 45% of the country’s GDP

• 85% of the export revenue
• Employs 80% of labour force.
• Major source of subsistence and household
income for majority of the rural population.
• Agricultural production is mainly rain-fed.

6
6
 Introduction…

• Owing to limited data availability, little attempt has

been made, in the past, to understand the
watershed dynamics of Ethiopian basins in
response to changing climate and catchment
conditions.
• The Rift Valley lakes and rivers system of Ethiopia
has undergone major changes in recent past.
• Agricultural, water supply, and hydropower sectors
are affected by variable climate patterns.

7
7
Key Objectives

• Explore the impact of large scale present and future

climatic variables under different greenhouse gas
forcing on local climate at Rift Valley lakes basin of
Ethiopia.
• Simulate runoff at desired locations using bias
corrected precipitation and temperature in two snow
free, agricultural watersheds in the basin.
• Forecast the likely future climate implications in the
basin.

8
8
STUDY AREA
• The study mainly focuses on Rift Valley lakes Basin of
Ethiopia.
• Geog. location: 4o24’29’’ to 8o26’38‘’ N latitude and
36o35’45’’ to 39o23‘31’’ E longitude.
• Mean annual rainfall: 600-1220 mm
• Temperature: Avg. temp. varies from 10.3 oC (min.) to 30.6 oC (max.)
• Climate Condition: sub-humid to moderate tropical semi-arid.
• Gemorphology: lowland plateau & escarpments
• Study Watersheds: Bilate (5330 km2) and Hare (166.5 km2)

9
9
River Basins of Ethiopia & Study Watersheds

Bilate

Hare

Bilate (5330km2) and

12 major river Basins Hare (166.5 km2)

10
Methodology
Rift Valley Lakes Basin (Study Basin)

2 GCMs selected
Collection of Data and - BCCR-BCM2.0
Preprocessing Selection of GCMs and - CSIRO-MK3.0
Temp., PCP., Spatial Data (DEM, Emission Scenarios
soil, land use) Emission Scenario
- A1B , A2 , CC

Predictands (observed Predictors (large scale Predictors

Temp. and PCP.) atmospheric variables) - Shum, Ux, Vx, Z, Slp

Downscaling 1990-99 (CC)

[ Temp and PCP ] 2081-90 (A1B, A2)

- Linear correction Bias Correction for

- Non-linear correction [ PCP and Temp ]

Runoff Simulation
Hydrologic Modelling
[ Present climate ]
[ SWAT Model ]
[ Future scenario ]
11
Dataset Used

• Observed daily precipitation and temperature data from

1980-2009.
• Daily Global Climate Model (GCM) Variables for current
climate (1971-1999) and
• Daily GCM Variables for future climate condition (2081-
2090).

13
13
 Data Used ( Bilate and Hare watersheds)

• Topographic Data (90m SRTM DEMs )

• Hydrometerological Data ( Daily RF, Temp., RH, Sunshine and
Streamflow)
• Soil Data ( FAO soil data); Land Use Data ( Landsat image )
(a)

Soil
Land use Slope

(a)

14
14
Soil Land use and Slope classes

(b)

(b)

Soil Land Use Slope

15
15
GCMs Selected and Scenario Used
• BCCR-BCM2.0 (Bjerknes Center for Climate Research
Version 2.0 of Norway )
- Atmospheric resolution of T63 (1.9ox1.9o)
- Oceanic resolution of 0.5o-1.5o x 1.5o
• CSIRO-MK3.0 (Commonwealth Scientific and Industrial
Research Organization of Australia)
- Atmospheric resolution of T63 (1.9ox1.9o)
- Oceanic resolution of 0.9ox1.9o.
• Emission Scenarios Used: A1B, A2 and Current Climate
condition

A1B – medium forcing, CO2 conc. 720 ppm in 2100 (Balanced World)
A2 – high forcing , CO2 conc. 820ppm in 2100 (Heterogeneous World)
16
16
Temperature downscaling

• Downscaled temperature show modest agreement to the

observed series.
• Regression , quantile plot and Q-Q plot is conducted to verify
the validity of simulated current climate condition.

17
17
Observed Temp. versus downscaled
(Regression)
• .

18
18
Quantile plot of temp. for two GCM
realizations
34 34
32 A1B Scenario 32 A2 Scenario
30 30
TMAX (oC)

TMAX (oC)
28 28
26 26
24 Observed Observed
24
MK3.0 MK3.0
22
BCM2.0 22 BCM2.0
20
20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Quantiles
Quantiles

34
32 Current Climate
30
TMAX (oC)

28
The quantile plots are S-shaped
26
24
Observed and are characteristic example
MK3.0
22 BCM2.0 of bell-shaped distribution.
20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Quantiles
19
19
Current climate Precip. (Predicted, Observed, Bias
corrected
160
Precipitation (mm) 140 (a) BCM2.0
120
100
80
60 Observed
Uncorrected
40 Linear corrected
20 Power transformed
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

180
160 (b) MK3.0
140
Precipitation (mm)

120
100
80
Observed
60
40
Uncorrected
20 Linear corrected
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 20
20
 Future climate Precip. (Predicted, Observed, Bias
corrected
A1B scenario Precip. A2 scenario Precip.
200
180 BCM2.0 350
160 Linear corrected
BCM2.0
Precipitation (mm)

300
140 Power transformed

Precipitation (mm)
250 Observed
120
Uncorrected
100 200
80
150
60 Observed
40 Uncorrected 100
Linear corrected
20 Power transformed 50
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
180 180
160 MK3.0 160 MK3.0
Precipitation [mm]

140 140

Precipitation (mm)
120 120
100 100
80 80
60 Observed 60 Observed
40 Uncorrected 40 Uncorrected
Linear corrected Linear corrected
20 20
Power transformed Power transformed
0 0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

21
21
Hydrologic Modelling

• SWAT model is calibrated and validated for observed data

at both watersheds.
• Runoff is simulated for current climate and future scenarios
at two watersheds.
• Simulated daily runoff is aggregated to monthly series for
further analysis.

22
22
 Discharge (m3/s) Runoff (m3/s)

10

2
4

0
6
8
0
30
40

10
20
50
60
Jan-95 Jan-92
Apr-95
Jun-95 May-92
Sep-95
Sep-92
Dec-95
Mar-96 Jan-93

observed
simulated
Jun-96

Hare
May-93
Sep-96
Dec-96 Aug-93
Mar-97
Dec-93
Bilate

Jun-97
Sep-97 Apr-94
Dec-97
Aug-94
Mar-98
Jun-98 Dec-94
Sep-98
Apr-95
Dec-98
Mar-99 Aug-95
1992-96

Jun-99 1995-00 Dec-95

Sep-99
Dec-99 Apr-96
Model Calibration

Mar-00
Aug-96
Jun-00
observed
simulated
Sep-00 Dec-96
Dec-00
0
0

800
600
400
200
800
400
200

600

1000

Rainfall (mm) RF (mm)

simulated discharge
Simulated discharge (m3/s) (m3/s)
0
1
2
3
4
5
0
20
30
40
50

10

0
0

1
10

2
R² = 0.81
20
R² = 0.92

3
Qsim = 0.8687*Q obs + 0.207
30
Qsim = 0.963Qobs - 0.674

4
3

3
observed discharge (m /s)
40

observed discharge (m /s)

5

23
(a)

(b)
50

23
 Discharge (m3/s)
Discharge (m3/s)

0
20
40
60
80

0
2
4
6
Jan-03 Jan-98

Apr-03 May-98

Jul-03 Sep-98

Oct-03 Dec-98

Jan-04 Apr-99
Apr-04 Aug-99
Jul-04 Dec-99

Hare Watershed
Oct-04 Apr-00
Jan-05
Bialte Watershed

Aug-00
Apr-05
Dec-00

observed
simulated
Jul-05
Apr-01
Oct-05
Aug-01
Jan-06
Model Validation

2003-06
Dec-01
Apr-06

(a)
Apr-02
1998-02

Jul-06
Aug-02
Oct-06
observed
Simulated

Dec-02
0
0

800
600
400
200
800
600
400
200

Rainfall (mm)
RF (mm)

simulated discharge (m 3/s) simulated discharge (m3/s)

1
2

0
3
4
10
20
30
40
50

0
0

1
10

R² = 0.81
20
R² = 0.82

2
Qsim = 1.0679*Qobs - 0.3046
Qsim = 0.95*Qobs + 1.30

30

observed discharge
3

3
(m3/s)
observed discharge (m /s)
40

(b)

24
50

24
Model performance indices

Model efficiency Bilate basin Hare-basin

indices Calibration Validation Calibration Validation
( 1995-2000) (2003-2006) ( 1994-2000) (2003-2006)
R2 0.92 0.82 0.88 0.81
bR2 0.89 0.78 0.71 0.86
NSE 0.91 0.79 0.87 0.96
PBIAS 8.93 -9.05 1.36 8.83
RSR 0.09 0.21 0.19 0.32
p-factor 0.82 0.78 0.81 0.78
r-factor 0.72 0.88 1.40 1.80

25
25
 Runoff (mm) Runoff (mm)

20
40
60
80

0
100

60

20
40
80

0
100
120
Jan-90 Jan-90
( 1990-99)

Jul-90 Jul-90

(b)
(a)
Jan-91 Jan-91
Jul-91 Jul-91
Jan-92 Jan-92
Jul-92 Jul-92
Jan-93 Jan-93
Jul-93 Jul-93
Jan-94 Jan-94
Jul-94 Jul-94
Jan-95 Jan-95
Observed

Jul-95 Jul-95
Observed

Jan-96 Jan-96
Jul-96 Jul-96
Jan-97 Jan-97
BCM2.0

Jul-97 Jul-97
MK3.0

Jan-98 Jan-98
Jul-98 Jul-98
Jan-99 Jan-99
Runoff Simulated – Current climate condition

MK3.0

Jul-99 Jul-99
BCM2.0

Jan-00 Jan-00
Hare
Bilate

26
26
 Runoff (mm) Runoff (mm)

0
100

40
60
80

20

20
80

40
60

0
100
Jan-2081 Jan-2081
Jul-2081 Jul-2081
Jan-2082 Jan-2082
Jul-2082 Jul-2082
Jan-2083 Jan-2083
Jul-2083 Jul-2083
Jan-2084 Jan-2084
Jul-2084 Jul-2084
Jan-2085 Jan-2085

MK3.0 (A2)
MK3.0 (A1B)

Jul-2085 Jul-2085
Jan-2086 Jan-2086
Bilate

Jul-2086 Jul-2086
Jan-2087 Jan-2087
Jul-2087 Jul-2087
Jan-2088 Jan-2088
Jul-2088 Jul-2088
Jan-2089 Jan-2089
Jul-2089 BCM2.0 (A2) Jul-2089
BCM2.0 (A1B)

Jan-2090 Jan-2090
Jul-2090 Jul-2090 Runoff (mm)
20
40
60
80

0
100
120

0
100
120

20
40
60
80

Jan-2081 Jan-2081
Jul-2081 Jul-2081
Jan-2082 Jan-2082
Jul-2082 Jul-2082
Jan-2083 Jan-2083
Jul-2083 Jul-2083
Jan-2084 Jan-2084
Jul-2084 Jul-2084
MK3.0 (A1B)

Jan-2085 Jan-2085
BCM2.0 (A2)

Jul-2085
Hare

Jul-2085
Jan-2086 Jan-2086
Jul-2086 Jul-2086
Jan-2087 Jan-2087
Jul-2087 Jul-2087
Jan-2088 Jan-2088
Jul-2088 Jul-2088
Jan-2089 Jan-2089
Runoff Simulated – Future climate condition (2081-90)

Jul-2089 Jul-2089
BCM2.0 (A1B)

MK3.0 (A2)

Jan-2090 Jan-2090
Jul-2090 Jul-2090
27
27
 Average monthly Runoff Simulated
Bilat

Bilate
Hare
Hare

A2 scenario A1B scenario Current climate

28
28
Box plots of simulated Runoff (mm)
A B

29
29
Conclusion
• Runoff simulated for the current climate (1990-1999) using
bias corrected precipitation and temperature modestly
reproduced effects similar to that of observed weather
variables at both watersheds.
• The overall NSE and coefficient of determination model
performance indices ranges between 0.79 and 0.96 during
calibration and validation period at both watersheds; other
indices are at acceptable limit.
• The simulated annual water yield is within ±3.4% error to the
observed annual stream flow volume at the same outlet.

30
30
Conclusions

• Significant amount of variability is observed in downscaled

precipitation for current climate whereas the associated
variability for temperature is very less.
• Increased extreme precipitation and temperature events
prevail for future scenarios.
• Average dry-spell length found to increase between October
and February whereas it remains stable from March –
September months for both emission scenarios.
• Bias correction improved both the mean and CV to a
reasonable degree

31
31
 Conclusions…

• Simulated average annual water yield shows slight

variation between GCMs.
• The simulated future runoff is characterized by
higher number of extreme events.

32
32
Thanks

33