You are on page 1of 9

ECOHYDROLOGY

HYDROBIOLOGY
DOI: 10.2478/v10104-011-0020-9
Vol. 10
No. 2-4, 191-200
2010

SWAT based runoff and sediment yield


modelling: a case study of the Gumera
watershed in the Blue Nile basin

Ecohydrology for
water ecosystems
and society in Ethiopia

Meqaunint Tenaw Asres1, Seleshi B. Awulachew2


Ministry of Water Resource, P.O. Box 15497, Addis Ababa,
e-mail: mequanntt@yahoo.com
2
International Water Management Institute, P.O. Box 5689, Addis Ababa, Ethiopia,
e-mail: s.bekele@cgiar.org
1

Abstract
Land degradation caused by soil erosion (sheet and rill erosion) and soil fertility decline
is a serious threat in the Ethiopian highlands, especially in the Gumara watershed. In this
study the SWAT (Soil and Water Assessment Tool) model was applied to the Gumara
watershed to predict sediment yield and runoff, to establish the spatial distribution of
sediment yield and to test the potential of watershed management measures to reduce
VHGLPHQWORDGLQJVIURPKRWVSRWDUHDV7KHPRGHOZDVFDOLEUDWHGXVLQJYH\HDUVRI
RZDQGVHGLPHQWUHFRUGVDQGYDOLGDWHGXVLQJGDWDIRUWKHQH[WWKUHH\HDUV2XWRI
6:$7VXEEDVLQVVXEZDWHUVKHGV  ZHUHLGHQWLHGDVHURVLRQSURQHDUHDV
contributing to a mean annual sediment load ranging from 11 to 22 t ha-1 yr-1. The model
ZDVXVHGWRHYDOXDWHWKHSRWHQWLDORIYHJHWDWLRQOWHUVWULSVRIYDULRXVZLGWKVWRUHGXFH
VHGLPHQWSURGXFWLRQIURPFULWLFDOPLFURZDWHUVKHGV7KHLQVWDOODWLRQRIYHJHWDWLRQOWHU
VWULSVRQYXOQHUDEOHODQGZDVVKRZQWRUHVXOWLQDWRUHGXFWLRQLQVHGLPHQW
\LHOGIRUVWULSZLGWKVRIPDQGPUHVSHFWLYHO\
Key words: land degradation, soil erosion, sediment yield, SWAT, critical sub waterVKHGVYHJHWDWLRQOWHUVWULSV

1. Introduction
Ethiopia possesses huge amount of potential
natural resources, which include an annual
DYDLODELOLW\RIELOOLRQP3 of surface water and
2.6 billion m3 of ground water as well as 3.7 million
ha of potentially irrigable land, that could be used
WRLQFUHDVHDJULFXOWXUDOSURGXFWLRQDQGSURGXFWLYLW\
(Awulachew et al.0R:5 +RZHYHU
despite this potential resource base, agricultural
production is low in some parts of the country, as

DUHVXOWRIHQYLURQPHQWDOGHJUDGDWLRQZKLFKLV
manifested primarily by soil erosion and loss of
soil fertility (Teketay 2004).
Sheet and rill erosion are, by far, the most
widespread types of accelerated water erosion,
constituting a principal cause of land degradation
LQWKHFRXQWU\7KHLUFRPELQHGLPSDFWVLJQLFDQWO\
DIIHFWVDJULFXOWXUDOSURGXFWLRQDQGSURGXFWLYLW\
(Constable 1984). A rapidly increasing population,
GHIRUHVWDWLRQ RYHUFXOWLYDWLRQ H[SDQVLRQ RI
FXOWLYDWLRQDWWKHH[SHQVHRIODQGXQGHUFRPPXQDO

192

M.T. Asres, S.B. Awulachew

use rights (grazing and woody biomass resource 2. Materials and methods
H[SORLWDWLRQ FXOWLYDWLRQRIPDUJLQDODQGVWHHSODQGV
RYHUJUD]LQJDQGRWKHUVRFLDOHFRQRPLFDQGSROLWLFDO 2.1. Description of the study area
IDFWRUVKDYHUHVXOWHGLQVHULRXVVRLOHURVLRQZLWKLQWKH
The Gumara watershed is located in the North
country in general and within the Gumara watershed
West part of Ethiopia in Amhara Regional State;
in particular (BCEOM 1998; MoARD 2004).
south Gondar zone (Fig. 1). It is situated to the
Despite the recognition of the problem, based VRXWKHDVWRI/DNH7DQDDQGFRYHUVWKHGUDLQDJH
on estimates of gross soil loss and sediment measure- area of about 1464 km2.
PHQWVDWZDWHUVKHGRXWOHWVIHZVWXGLHVKDYHEHHQ
The major landforms of the watershed include
conducted to determine the spatial and temporal DWJHQWO\VORSLQJWRXQGXODWLQJSODLQVKLOOVDQG
YDULDELOLW\RIUXQRIIDQGVHGLPHQW\LHOGDWWKHPLFUR mountains. The upper and middle parts of the
ZDWHUVKHGOHYHO6WXGLHVDWWKLVVFDOHDUHLPSRUWDQW catchment are characterized by mountainous, highly
EHFDXVHPDQ\RIWKHVROXWLRQVWRHQYLURQPHQWDO rugged and dissected topography with steep slopes
problems, such as soil erosion and non-point source DQGWKHORZHUSDUWLVFKDUDFWHUL]HGE\YDOOH\RRUV
pollution, will require changes in management on ZLWKDWWRJHQWOHVORSHV(OHYDWLRQLQWKHZDWHUVKHG
the landscape scale (Wilson et al. 2000).
UDQJHVIURPWRPDERYHVHDOHYHOZLWK
Measurement of sediment transport at a water- DPHDQHOHYDWLRQRIPDVO0RUHWKDQWKUHH
VKHGRXWOHWSURYLGHVXVHIXOLQIRUPDWLRQFRQFHUQLQJ TXDUWHUVRIWKHZDWHUVKHGLVLQWHQVLYHO\FXOWLYDWHG
ZKDWLVKDSSHQLQJLQWKHZDWHUVKHG+RZHYHULWGRHV and teff (Eragrostis tef), maize, barley, and wheat
not indicate which part of the watershed is susceptible are the major crops. Bush or shrub land, grazing
to erosion and contributing more sediment to the out- land, forest/wood land and wetland/swamp are
let. Furthermore, lack of procedures for transferring WKHRWKHUPDLQODQGFRYHUW\SHVLQWKHZDWHUVKHG
LQIRUPDWLRQIURPULYHUJDXJLQJVWDWLRQVWRSURSRVHG (WWDSE 2007).
UHVHUYRLUVLWHVDQGPLFURZDWHUVKHGVUHSUHVHQWVDQ%DVHGRQWKH)$2FODVVLFDWLRQV\VWHPVL[
RWKHUSUREOHPLQHVWLPDWLQJRZDQGVHGLPHQWORDGV VRLOW\SHVQDPHO\+DSOLF/XYLVROV&KURPLF/Xat the location of interest (Admasu 2006).
YLVROV/LWKLF/HSWRVROV(XWULF9HUWLVROV(XWULF
One of the possible solutions to the problem )OXYLVROVDQG&KURPLF&DPELVROVDUHWKHFRPof land degradation due to soil erosion is therefore, mon soil types in the watershed (BCEOM 1998;
to understand the processes causing erosion at the 0R$5'::'6( 5DLQIDOORYHUWKH
PLFURZDWHUVKHGOHYHODQGWRLPSOHPHQWZDWHUVKHG watershed is unimodal and most of the rainfall is
PDQDJHPHQWPHDVXUHV(IIHFWLYHZDWHUVKHGSODQ- concentrated in the season extending from June
ning requires information on runoff and erosion WR 6HSWHPEHU ZLWK D YLUWXDO GURXJKW RFFXUULQJ
rates at the plot, hill slope, and small watershed IURP1RYHPEHUWKURXJK$SULO7KHIRXUZHWWHVW
VFDOHDQGDQXQGHUVWDQGLQJRIKRZWKHVHYDU\DFURVV months contribute 85 percent of the total annual
the landscape. In addition, there is
DQHHGWRLGHQWLI\DUHDVWKDWKDYH
the potential for high erosion, so
WKDWFRUUHFWLYHDFWLRQVFDQEHWDNHQ
to reduce sediment production from
these areas.
,QUHFHQWGHFDGHVVHYHUDOVLPXODWLRQPRGHOVKDYHEHHQGHYHORSHGIRU
estimating soil erosion and enhancing understanding of the spatial and
temporal complexities of catchment
response. Such models facilitate asVHVVPHQWRISURFHVVHVRSHUDWLQJRYHU
large areas, in order to identify and
target priority management areas.
7KH PDLQ REMHFWLYHV RI WKLV
study were to determine the spaWLDOYDULDELOLW\RIVHGLPHQW\LHOGWR
identify critical micro watersheds
DQGWRHYDOXDWHYDULRXVFRQVHUYDtion scenarios for reducing sediment
yield, based on the simulation results
SURYLGHGE\WKHSK\VLFDOO\EDVHGDQG
spatially distributed SWAT model.
Fig. 1.0DMRUULYHUEDVLQVLQ(WKLRSLDDQGORFDWLRQRI*XPDUDZDWHUVKHG

SWAT based runoff and sediment yield modeling

rainfall. The dry season (October to May) has a total


rainfall of about 15% of the mean annual rainfall
(WWDSE 2007).

2.2. Methods
The SWAT 2005 model integrated with geographic information system (GIS) techniques was
used to simulate runoff and sediment yield in this
study. SWAT is a physically-based and computaWLRQDOO\HIFLHQWK\GURORJLFDOPRGHOZKLFKXVHV
UHDGLO\DYDLODEOHLQSXWV,WZDVGHYHORSHGWRSUHGLFW
the impact of land management practices on water,
sediment and agricultural chemical yields in large
FRPSOH[ZDWHUVKHGVZLWKYDU\LQJVRLOVODQGXVH
DQGPDQDJHPHQWFRQGLWLRQVRYHUORQJSHULRGVRI
time (Neitsch et al. 2005).
'LJLWDOHOHYDWLRQPRGHO '(0 GDWDSRO\JRQ
FRYHUDJHRIVRLOVDQGODQGXVHDQGSRLQWFRYHUDJH
of weather stations were used as basic input to the
model. Other inputs include daily rainfall, miniPXPDQGPD[LPXPWHPSHUDWXUHUHODWLYHKXPLGLW\
solar radiation, and wind speed. We delineated the
watershed using a 90 m 90 m resolution DEM
and digitized stream networks for the study area.
After watershed delineation, it was partitioned
LQWRK\GURORJLFUHVSRQVHXQLWV +58 ZKLFKKDYH
unique soil and land use combinations within the
watershed to be modeled. Accordingly, multiple
HRU with a 20% land use threshold and a 10% soil
threshold were adopted.
For modelling surface runoff and sediment
\LHOGWKH6RLO&RQVHUYDWLRQ6HUYLFHFXUYHQXPEHU
PHWKRG 6&6 DQGWKHPRGLHG8QLYHUVDO
6RLO/RVV(TXDWLRQZHUHXVHGUHVSHFWLYHO\,QRUGHU
WRLGHQWLI\WKHPRVWLPSRUWDQWRUVHQVLWLYHPRGHO
SDUDPHWHUVEHIRUHFDOLEUDWLRQPRGHOVHQVLWLYLW\
analysis was carried out using a built-in SWAT
VHQVLWLYLW\DQDO\VLVWRROWKDWXVHVWKH/DWLQ+\SHUcube One-factor-At-a-Time (LH-OAT) procedure
9DQ*ULHQVYHQ 2QFHWKHPRVWVHQVLWLYH
SDUDPHWHUVKDYHEHHQLGHQWLHGYDOXHVRIVHOHFWHG
PRGHOSDUDPHWHUVZHUHYDULHGLWHUDWLYHO\RYHUDQ
DSSURSULDWHUDQJHGXULQJYDULRXVFDOLEUDWLRQUXQV
XQWLODVDWLVIDFWRU\DJUHHPHQWEHWZHHQREVHUYHG
DQGVLPXODWHGVWUHDPRZDQGVHGLPHQWGDWDZHUH
obtained.
Daily precipitation, maximum and minimum
WHPSHUDWXUHVRODUUDGLDWLRQUHODWLYHKXPLGLW\DQG
wind speed data collected from 9 weather stations
were used as an input for the model. The missing
metrological data from these weather stations were
OOHGE\DZHDWKHUJHQHUDWRUPRGHOHPEHGGHGLQ
SWAT. Daily discharge data were collected and
the gap of daily suspended sediment was estimated
XVLQJVHGLPHQWUDWLQJFXUYHZLWKFRUUHODWLRQFRHIFLHQWRI

193

Before calibration proceeds, the performance


RIWKHPRGHOZDVHYDOXDWHGIURPWKHLQLWLDOVLPXODWLRQUXQVZLWKPRGHOGHIDXOWSDUDPHWHUYDOXHV
HOGPHDVXUHGSDUDPHWHUVDQGSDUDPHWHUVFROOHFWHG
from literature. The results of the model from the
LQLWLDOUXQZDVQRWZRUVHDQGQHHGVQHWXQLQJ
RIWKHPRVWVHQVLWLYHSDUDPHWHUVXQWLODFFHSWDEOH
calibration statics was obtained. Therefore, the
manual calibration procedure was used in this study.
For each calibration run and parameter change, the
corresponding model performance statistics (R and
Ens) were calculated. This procedure continued until
WKHDFFHSWDEOHFDOLEUDWLRQVWDWLFVZDVDFKLHYHG7KH
SHUFHQWGLIIHUHQFHEHWZHHQVLPXODWHGDQGREVHUYHG
GDWD ' WKHFRUUHODWLRQFRHIFLHQW 52) and the
Nash and Sutcliffe simulation efficiency index
(Ens  1DVK6XWFOLIIH ZHUHXVHGWRHYDOXDWH
the models performance during calibration and
YDOLGDWLRQSURFHVVHV
7KHPRGHOZDVFDOLEUDWHGDQGYDOLGDWHGXVLQJRZDQGVXVSHQGHGVHGLPHQWGDWDPHDVXUHG
LQRQHJDXJLQJVWDWLRQFRYHULQJDERXWRIWKH
WRWDOZDWHUVKHG)LYH\HDUVRIPHDVXUHGGDLO\RZ
DQGVHGLPHQW\LHOGGDWDZHUHXVHGIRURZDQG
sediment calibration. The model with calibrated
SDUDPHWHUVZDVYDOLGDWHGE\XVLQJDQLQGHSHQGHQW
VHWRIPHDVXUHGRZDQGVHGLPHQW\LHOGGDWDZKLFK
were not used during model calibration. We used
WKUHH\HDUVRIPHDVXUHGRZDQGVHGLPHQWGDWD
IRU YDOLGDWLRQ SURFHVVHV )ROORZLQJ FDOLEUDWLRQ
DQGYDOLGDWLRQRIWKH6:$7PRGHOWZRVFHQDULR
DQDO\VHV OWHUVWULSVRIYDULRXVZLGWKV ZHUHUXQ
to assess the potential reduction in sediment loads
from critical sub-watersheds.

3. Results
3.1. Sensitivity analysis
6HQVLWLYLW\DQDO\VLVZDVXQGHUWDNHQWRLGHQWLI\
VHQVLWLYH SDUDPHWHUV WKDW VLJQLILFDQWO\ DIIHFWHG
VXUIDFHUXQRIIEDVHRZDQGVHGLPHQW\LHOG&XUYH
QXPEHU &1,, DYDLODEOHZDWHUFDSDFLW\ 62/B
$:& DYHUDJHVORSHVWHHSQHVV 6/23( VDWXUDWHG
K\GUDXOLFFRQGXFWLYLW\ 62/B. PD[LPXPFDQRS\
VWRUDJH FDQP[ VRLOGHSWK 62/B= DQGWKHVRLO
HYDSRUDWLRQ FRPSHQVDWLRQ IDFWRU (6&2  ZHUH
IRXQGWREHUHODWLYHO\VHQVLWLYHSDUDPHWHUVWKDW
VLJQLFDQWO\DIIHFWVXUIDFHUXQRII
The threshold water depth for flow in the
VKDOORZDTXLIHU *:401 WKHEDVHRZ$OSKD
IDFWRU $/3+$B%) DQGWKHGHHSDTXLIHUSHUFRODWLRQ
IUDFWLRQV UFKUJGS ZHUHRWKHUVHQVLWLYHSDUDPHWHUV
WKDWPDLQO\LQXHQFHEDVHRZ 7DEOH9, 
:HDOVRLGHQWLHGDYHUDJHVORSHVWHHSQHVV
WKH8QLYHUVDO6RLO/RVV(TXDWLRQ 86/( FRYHU
or management factor, the USLE support practice

194

M.T. Asres, S.B. Awulachew

IDFWRUDYHUDJHVORSHOHQJWKWKHOLQHDUIDFWRUIRU
channel sediment routing and the exponential factor
IRUFKDQQHOVHGLPHQWURXWLQJDVWKHPRVWVHQVLWLYH
SDUDPHWHUVWKDWVLJQLFDQWO\DIIHFWVHGLPHQW\LHOG
WDEOH9,, 

3.2. Calibration and validation


Flow calibration resulted in a NashSuttcliffe
VLPXODWLRQHIFLHQF\LQGH[ (ns) of 0.76, a correlaWLRQFRHIFLHQW 52 RIDQGDPHDQGHYLDWLRQRI
3.29% showing a good agreement between measured

DQGVLPXODWHGPRQWKO\RZV )LJ7DEOH, )RU


WKHYDOLGDWLRQDJRRGPDWFKEHWZHHQPRQWKO\PHDVXUHGDQGVLPXODWHGRZVZDVGHPRQVWUDWHGE\WKH
FRUUHODWLRQFRHIFLHQW 52) of 0.83, a NashSutcliffe
VLPXODWLRQHIFLHQF\LQGH[ (ns) of 0.68 and a mean
GHYLDWLRQRIPHDVXUHGDQGVLPXODWHGPRQWKO\RZV
of -5.4% (Table II).
In calibrating the sediment response, a good
agreement between simulated and measured monthly
sediment yield was demonstrated by a correlation
coefficient (R2) of 0.85, a Nash-Sutcliffe model
HIFLHQF\LQGH[ (ns RIDQGDPHDQGHYLDWLRQ

Fig. 2.&DOLEUDWLRQUHVXOWVIRUPRQWKO\PHDQPHDVXUHGDQGVLPXODWHGRZV

Fig. 3.9DOLGDWLRQUHVXOWVIRUPRQWKO\PHDQPHDVXUHGDQGVLPXODWHGRZ
Table I.&DOLEUDWLRQDQGYDOLGDWLRQVWDWLVWLFVIRUPRQWKO\PHDQPHDVXUHGDQGVLPXODWHGVWUHDPRZV
Parameter

Calibrated (1998-2002)

Validated (2003-2005)

0.87
0.76
3.29

0.83
0.68
-5.4

R  FRUUHODWLRQFRHIFLHQW
Ens 1DVK6XWFOLIIHPRGHOHIFLHQFLHV
' GHYLDWLRQRIPHDQGLVFKDUJH
2

Table II.&RPSDULVRQRIPRQWKO\PHDQPHDVXUHGDQGVLPXODWHGRZV
Period
Calibration (1998-2002)
9DOLGDWLRQ 

0HDQRZ P3 s-1)
Measured
Simulated
31.63
32.69
33.98
32.15

SWAT based runoff and sediment yield modeling

' RI 7DEOH,,,)LJ :KHQYDOLGDWLQJ


the model for sediment, a good match between
simulated and measured sediment was demonstrated by
DFRUUHODWLRQFRHIFLHQW 52) of 0.79, a NashSutcliffe
PRGHOHIFLHQF\LQGH[ (ns DQGPHDQGHYLDWLRQ
' RI 7DEOH,9 

3.3. The spatial pattern of sediment source


areas
$IWHUFDOLEUDWLRQDQGYDOLGDWLRQWKHPRGHOZDV
run for a period of 10 years from 1996 to 2005. From

195

the model simulation output, the main sediment


VRXUFHDUHDVLQWKHZDWHUVKHGZHUHLGHQWLHG7KH
measured mean annual suspended sediment yield
for the 10 year period generated from the sediment
UDWLQJFXUYHZDVWKD-1 year-1 and the mean
annual suspended sediment yield simulated by the
SWAT model was 16.2 t ha-1 year-1.
The spatial distribution of sediment yield within
the Gumara watershed is presented in Figure 6. The
spatial distribution of sediment yields indicates that,
out of the total 30 SWAT sub-basins, 18 sub-basins
produce mean annual sediment yields ranging from

Table III. &DOLEUDWLRQDQGYDOLGDWLRQUHVXOWVRIPRQWKO\PHDQPHDVXUHGDQGVLPXODWHGVHGLPHQW\LHOG


Parameter
R  FRUUHODWLRQFRHIFLHQW
Ens 1DVK6XWFOLIIHPRGHOHIFLHQFLHV
' 'HYLDWLRQRIPHDQGLVFKDUJH
2

Calibrated (1998-2002)
0.85
0.74
-14.2

Validated (2003-2005)
0.79
0.62
-16.9

Fig. 4. Calibration results for monthly mean measured and simulated sediment yield.

Fig. 5.9DOLGDWLRQUHVXOWVIRUPRQWKO\PHDQPHDVXUHGDQGVLPXODWHGVHGLPHQW\LHOG
Table IV. Comparison of simulated and measured mean monthly and annual sediment yields for the calibration
DQGYDOLGDWLRQSHULRGV
Simulation periods
Calibration (1998-2002)
9DOLGDWLRQ 

Mean measured sediment yield


(t ha-1 m-1)
(t ha-1 yr-1)
1.80
21.6
1.91
22.92

Mean simulated sediment yield


(t ha-1 m-1)
(t ha-1 yr-1)
1.55
18.6
1.60
19.2

196

M.T. Asres, S.B. Awulachew

Fig. 6. The spatial distribution SWAT simulated annual sediment yield classes by sub-basin (t ha-1 yr-1). Numbers
1-30 indicate a sub-basin number.

11-22 t ha-1 year-1, while most of the lowland and


wetland areas are characterized by sediment yields
in the range of 0-10 t ha-1 year-1.

3.4. Scenario analysis


2QFHWKHPRGHOKDGEHHQYDOLGDWHGDQGWKH
results considered acceptable, the model could be
parameterized to explore the scenarios of interest
HJWRHYDOXDWHWKHSRWHQWLDOLPSDFWRIODQGXVH
FKDQJHPDQDJHPHQWDQGFRQVHUYDWLRQSUDFWLFHV 
$IWHUGHWDLOHGDQDO\VLVRIWKHSUREOHPVDQGEHQHWV
RIWKHH[LVWLQJSK\VLFDOFRQVHUYDWLRQSUDFWLFHVLQ
WKHZDWHUVKHGZHWHVWHGWKHPRGHOZLWKDOWHUQDWLYH
VFHQDULRVRIYHJHWDWLRQOWHUVWULSVZLWKYDU\LQJ
width to reduce sediment production from critical
VXEZDWHUVKHGV,QHYDOXDWLQJWKHHIIHFWRIOWHU
strips, three management scenarios were considered
and simulated:
, EDVHFDVH QROWHUVWULS
,, OWHUVWULSVPZLGHRQDOO+58VLQVHOHFWHG
sub-watersheds; and
,,,OWHUVWULSVPZLGHRQDOO+58VLQVHOHFWHG
sub-watersheds.
7KHLQVWDOODWLRQOWHUVWULSVZDVVKRZQWRUHduce the mean annual sediment yield by 58% to
62% with 5 m buffer strip and 74.2 to 74.4% with
POWHUVWULSV 7DEOH9)LJDQG 5HODWLYHO\
high reductions in sediment yield occurred when 5
PZLGHOWHUVWULSVZHUHVLPXODWHG+RZHYHUGRXEOLQJWKHZLGWKRIWKHOWHUVWULSVIURPPWRP
GLGQRWSURGXFHWKHVDPHOHYHORIUHGXFWLRQDVZDV
REVHUYHGIRUWKHDQGPVLWXDWLRQ7KLVVXJJHVWV
WKDWWKHEHQHWVGHULYHGIURPLQVWDOOLQJOWHUVWULSV
ZLOOWDSHURIIIRUIXUWKHULQFUHDVHVLQOWHUZLGWK

4. Discussion
(YHQWKRXJKWKHSUREOHPRIVRLOHURVLRQLV
FOHDUO\GHPRQVWUDWHGE\DYDLODEOHHVWLPDWHVRIJURVV
HURVLRQDQGHOGREVHUYDWLRQVWKHUHLVDQHHGIRU
TXDQWLWDWLYHLQIRUPDWLRQDWWKHPLFURZDWHUVKHG
OHYHOWRGHYHORSZDWHUVKHGPDQDJHPHQWSODQVDQG
for decision making. In this study, an attempt was
made to characterize the Gumera watershed in
WHUPVRIVHGLPHQW\LHOGLGHQWLFDWLRQRISRWHQWLDO
VHGLPHQWVRXUFHDUHDVDQGHYDOXDWLRQRIDOWHUQDWLYHPDQDJHPHQWPHDVXUHVWRUHGXFHWKHRQVLWH
and offsite effects of soil erosion in the watershed.
:HHYDOXDWHGWKHSHUIRUPDQFHRIWKH6:$7
PRGHOXVLQJVWDQGDUGFDOLEUDWLRQDQGYDOLGDWLRQ
statistics. A good agreement between simulated
DQGPHDVXUHGPRQWKO\VWUHDPRZZDVGHPRQVWUDWHGE\WKHFRUUHODWLRQFRHIFLHQW 52 = 0.87),
WKH1DVK6XWFOLIIHPRGHOHIFLHQF\ (ns = 0.76)
DQGWKHPHDQGHYLDWLRQSHUFHQW '  IRUWKH
FDOLEUDWLRQSHULRGDQGYDOXHVRI52 = 0.83, Ens = 0.8
DQG' IRUWKHYDOLGDWLRQSHULRG7KHPRGHO
RYHUHVWLPDWHGRZVE\DQGXQGHUHVWLPDWHG
flows by 5.4% for the calibration (1998-2002)
DQGYDOLGDWLRQSHULRGV  UHVSHFWLYHO\
+RZHYHUWKHVHYDOXHVDUHZLWKLQWKHDFFHSWDEOH
range of 15%. In general, the time series trend of
WKHJDXJHGRZLVZHOOPDWFKHGE\WKHVLPXODWHG
RZLQERWKFDOLEUDWLRQDQGYDOLGDWLRQSHULRGV
In simulating sediment yield, good agreement
between simulated and measured flows for the
calibration period was again obtained, as demonVWUDWHGE\WKHFRUUHODWLRQFRHIFLHQW 52 = 0.85),
WKH1DVK6XWFOLIIHPRGHOHIFLHQF\ (ns = 0.74.)
DQGWKHPHDQGHYLDWLRQSHUFHQW '  7KH
HTXLYDOHQWYDOXHVIRUWKHYDOLGDWLRQSHULRGZHUH

SWAT based runoff and sediment yield modeling

197

Table V. 0HDQDQQXDOFKDQJHVLQVHGLPHQW\LHOGDVVRFLDWHGZLWKWKHLQVWDOODWLRQRIOWHUVWULSVRIYDU\LQJZLGWKVLQ
selected critical sub-watersheds.
Selected
critical
sub-watersheds
11
16
17
22
24
28
29

Mean annual sediment yield (t ha-1 yr-1) (1996-2005)


Filter strip width (m)
Base case
QROWHUVWULS
5m
10 m
11.800
4.5
3.03
18.200
7.6
4.68
12.100
4.6
3.11
17.600
6.8
4.54
21.300
8.2
5.48
12.700
4.9
3.28
19.200
7.4
4.95

Reduction in sediment yield (%)


Filter strip width (m)
5m
10 m
-62
-74.35
-58
-74.30
-62
-74.29
-61
-74.23
-62
-74.29
-61
-74.19
-61
-74.23

Fig. 7. Reduction in sediment yield (t ha-1 yr-1 GXHWRLQVWDOODWLRQRIOWHUVWULSVRIGLIIHUHQWZLGWKVDV


compared to the base case.

Fig. 8. Per cent of reduction in simulated mean annual suspended sediment yield (t ha-1 yr-1) due to
LQVWDOODWLRQRIOWHUVWULSV

198

M.T. Asres, S.B. Awulachew

Table VI.6XPPDU\RINH\LQLWLDODQGQDOFDOLEUDWHGRZSDUDPHWHUYDOXHV
Parameter description

Parameter
code

Range

,QLWLDO6&6&1,,YDOXH
CN2
25%
$YDLODEOHZDWHUFDSDFLW\ PPZDWHUPPVRLO
62/B$:& 25%
$YHUDJHVORSHVWHHSQHVV PP-1)
SLOPE
25%
62/B.
25%
6DWXUDWHGK\GUDXOLFFRQGXFWLYLW\ PPK-1)
7KUHVKROGZDWHUGHSWKLQVKDOORZDTXLIHUIRURZ PP GWQMN 0-5000
Maximum canopy storage (mm)
canmx
0-10
%DVHRZ$OSKDIDFWRU GD\V
$/3+$B%)
0-1
Deep aquifer percolation fraction
UFKUJBGS
0-1
6RLOHYDSRUDWLRQFRPSHQVDWLRQIDFWRU
ESCO
0-1
Soil depth (mm)
62/B=
25

Initial
(default)
value
*
**
**
0.0
0.0
0.048
0.05
0.95
**

Adjusted
parameter value
+13%
-20 %
12%
+25
0.0
0.1
0.048
0.03
0.90
-12%

&DQP[FDOLEUDWHGIRUIRUHVWDQGEUXVKODQGVDQGIRURWKHUODQGXVHVYDOXHIRUFDQP[LV 6:$7GHIDXOWSDUDPHWHU
YDOXHVIRUHDFKODQGXVHDQGVRLOW\SHLQHDFKVXEEDVLQ ,QLWLDOSDUDPHWHUYDOXHVIURPHOGPHDVXUHPHQWDQG
OLWHUDWXUHUHYLHZ
Table VII. 6XPPDU\RILQLWLDODQGQDOVHGLPHQWFDOLEUDWLRQSDUDPHWHUYDOXHV
Parameter Description

Range (g)

86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  WHII


0.003-0.5
86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  EDUOH\DQGZKHDW
0.001-0.5
86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  PDL]H 
0.001-0.5
86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  EXVKRUVKUXE
0.001-0.5
USLE (c factor) (forest)
0.001-0.5
86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  GHQVHJUDVV
0.001-0.5
86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  VRUJKXP
0.001-0.5
86/(FRYHURUPDQDJHPHQWIDFWRU 86/(B&  QJHUPLOOHW 
0.001 -0.5
$YHUDJHVORSHOHQJWK P 6/68%%61
25%
$YHUDJHVORSHVWHHSQHVV PP-1): SLOPE
25%
Linear factor for channel sediment routing (SPCON)
0.00001-0.01
Exponential factor for channel sediment routing (SPEXP)
1.0-2.0
86/(VXSSRUWSUDFWLFHIDFWRU 86/(B3
0-1

Initial
value
0.25(a)
0.15(b)
0.10(c)
0.02(d)
0.003(e)
0.01(f)
0.10
0.15
**
**
0.0001
1.0
1

Adjusted
parameter value
0.35
0.25
0.18
0.09
0.004
0.05
0.18
0.25
+15%
+12%
0.0035
1.1
0.85

D  E  F  G  H  I SDUDPHWHUYDOXHVGHQHGIRU(WKLRSLDE\+XUQL   J UDQJHVJLYHQLQ6:$7XVHUV


PDQXDO 6:$7GHIDXOWDQGGHULYHGSDUDPHWHUYDOXHV

R2 = 0.79, Ens = 0.62 and D = -16.9%. For both


tKHFDOLEUDWLRQDQGYDOLGDWLRQSHULRGVVLPXODWHG
sediment yield were underestimated by 14.2% and
UHVSHFWLYHO\EXWWKHVHYDOXHVIDOOZLWKLQ
WKHDFFHSWDEOHUDQJHRI2YHUDOOWKHWLPH
series trend of the measured sediment yield is well
matched by the simulated sediment yield for both
WKHFDOLEUDWLRQDQGYDOLGDWLRQSHULRGV
Considering the acceptable limits of the statistiFDOPRGHOHYDOXDWLRQFULWHULDWKHVHUHVXOWVLQGLFDWH
a good match between measured and simulated
sediment yield. A good performance of the model
IRUWKHYDOLGDWLRQSHULRGLQGLFDWHVWKDWWKHWWHG
parameters for the calibration period can be taken
DVSURYLGLQJDUHSUHVHQWDWLYHVHWRISDUDPHWHUVIRU
the Gumara watershed, and further simulation and
HYDOXDWLRQRIDOWHUQDWLYHVFHQDULRDQDO\VLVFDQEH
carried out for other periods using the SWAT model.

After calibration, it was possible to successfully


GHYHORSDPRGHOIRU*XPHUDZDWHUVKHGWRSUHGLFW
UXQRIIDQGVHGLPHQW\LHOGDQGWRHYDOXDWHWKHLPpact of broadly adoptable watershed management
LQWHUYHQWLRQVLHYHJHWDWLYHOWHUVWULSV)ROORZLQJ
FDOLEUDWLRQDQGYDOLGDWLRQRI6:$7PRGHOWZR
scenario analyses were tested to reduce sediment
loads from critical sub-watersheds. The simulation
UHVXOWVRIWKHWZRVFHQDULRDQDO\VLV XVLQJYHJHWDWLRQOWHUVWULSVRIPDQGPZLGH LQGLFDWHG
WKDWLPSOHPHQWLQJOWHUVWULSVFDQUHGXFHVHGLPHQW
yield by 58% to 74%. The simulation results are
consistent with results from soil and water conserYDWLRQUHVHDUFKSORWVLQWKH$ED\5LYHUEDVLQWKDW
indicated that soil loss was reduced by 55% to 84%
DIWHULQVWDOODWLRQRIYHJHWDWLRQOWHUVVWULSVPDLQO\
grass strips (SCRP 1996).

SWAT based runoff and sediment yield modeling

The SWAT model predictions demonstrated that


about 72% of the Gumara watershed is erosion prone
and contributes high sediment yields, exceeding
the tolerance limit (soil formation rate) in the study
DUHD2YHUDOO6:$7SHUIRUPHGZHOOLQVLPXODWLQJ
runoff and sediment yield on monthly basis at the
watershed scale and thus can be used as a planning
tool for watershed management. The study can be
further extended to similar watersheds in the Abay
5LYHUEDVLQRWKHUVLPLODUDUHDVDQGFDQEHXVHGWR
address the lack of information on processes operating at scales between the micro watershed and
ODUJHZDWHUVKHGV6LJQLFDQWLPSURYHPHQWVLQZDWHU
TXDOLW\DQGVHGLPHQWUHGXFWLRQFRXOGEHDFKLHYHG
by targeting critical areas of the Gumara watershed.

References
Admasu, T. 2006. Evaluation of sediment yield and
reservoir sedimentation in Angereb watershed. MSc
WKHVLV$UED0LQFK8QLYHUVLW\(WKLRSLD
Awulachew, S.B., Yilma, A.D., Luelseged, M., Loiskandl, W., Ayana, M., Alamirew, T. 2007. Water
resources and irrigation development in Ethiopia.
Working paper 123, International Water Management
Institute, Colombo, Sirilanka, pp. 78.
BCEOM 1998. Abbay River basin integrated development
master plan project9ROXPH;,,,(QYLURQPHQW
Ministry of Water Resource, Addis Ababa, Ethiopia.
Constable, M. 1984. Resource of rural development in
Ethiopia. Ethiopian Highlands Reclamation Study.
FAO/Ministry of Agriculture, Addis Ababa, Ethiopia.
Hurni, H. 1985. Soil conservation manual for Ethiopa.
Ministry of Agriculture, Addis Ababa, Ethiopia.

199

MoWR 2002. Ethiopian water sector strategy. Ministry


of Water Resources (MoWR), Addis Ababa, Ethiopia.
MoARD 2004. Woody biomass inventory and strategic
planning project, a national strategic plan for biomass
HQHUJ\VHFWRUQDOUHSRUW, Ministry of Agriculture and
5XUDO'HYHORSPHQW 0R$5' $GGLV$EDED(WKLRSLD
1DVK-6XWFOLIIH-5LYHURZIRUHFDVWLQJWKURXJK
conceptual models part 1 a discussion of principles.
Journal of Hydrology 10, 282-290.
1HLWVFK6/$UQROG-*.LQLU\-5:LOOLDPV-5
2005. Soil and water assessment tool, theoretical
documentation: version 2005. USDA Agricultural
5HVHDUFK6HUYLFHDQG7H[DV$ 0%ODFNODQG5HVHDUFK
Center, Temple, TX.
SCRP 1996. Supplementary notes on soil erosion assessment and recommendation on soil and water conservation6RLO&RQVHUYDWLRQ5HVHDUFK3URMHFW 6&53 
8QLYHUVLW\RI%HUQH6ZLW]HUODQGLQDVVRFLDWLRQZLWK
0LQLVWU\RI$JULFXOWXUH*RYHUQPHQWRI(WKLRSLD
SCS 1972. National engineering handbook. Section 4
+\GURORJ\86'$6RLO&RQVHUYDWLRQ6HUYLFH 6&6 
Teketay, D. 2004. Forestry research in Ethiopia: past,
SUHVHQWDQGIXWXUH,Q%DOFKD*<HVKLWHOD.
Bekele, T. [Eds] Proceedings of a national conference
on forest resources of Ethiopia: status, challenges and
opportunities, 27-29 November 2002. Addis Ababa,
Ethiopia, pp. 1-39.
9DQ*ULHQVYHQ$Sensitivity, auto-calibration,
uncertainty and model evaluation in SWAT2005.
Wilson, J., Gallant, C. 2000. Digital Terrain Analysis.
In: Wilson, J.P., Gallant, J.C. [Eds] Terrain analysis:
principles and applications, Wiley.
WWDSE 2007. Catchment development plan, Gumara
irrigation project. Ministry of Water Resources, Addis
Ababa, Ethiopia.

You might also like