Abstract:
Long term planning in water resource development requires full information about
the water resource and the amount sediment which affects amount and availability of
water.
For this the soil and water assessment tool (SWAT) model was applied to simulate
runoff, the sediment yield and associated uncertainty with the simulated output from
the Dura River sub basin (area 545 km2
), located in North-Western Amhara
Regional State, Ethiopia. The SWAT model was calibrated for the period of 1985 to
1994 and validated for the period of 1995-2004 based on nine parameters identified
during sensitivity analysis for both flow and sediment for which GW_DELAY for
flow and Curve number for sediment were the most sensitive ones with their t-stat
and p-values. The uncertainty analysis was done by using SUFI-2 which is package
for a SWAT –CUP. The calibration and validation of the model was found
satisfactory as performance rating criteria value of coefficient of correlation (R2
)
and Nash-Sutcliffe simulation efficiency (NSE) is found to be 0.85 and 0.82 during
calibration and 0.83 and 0.81 during validation for flow and 0.8 and 0.73 during
calibration and 0.78 and 0.72 during validation for sediment yield prediction
respectively. In the same order from the model uncertainties analysis the percentage
of the observed data within the uncertainty bound is 88% during calibration and
77% during validation for flow and 80% during calibration and 70% during
validation for sediment yield prediction. The water and sediment yield of Dura sub
basin was quantified and also the most sediment yielding part of the basin was
identified. Accordingly, annual average flow of Dura River was estimated to be
18.8M 3
/sec. and annual average sediment yield of the sub basin was estimated to be
10.9 ton per hectare. Substantial sediment contributing sub basins were sub basin 1
(73.36 ton ha-1
yr
-1
), sub basin 6 (78.33 ton ha-1
yr
-1
) and sub basin 9 (96.5ton ha-1
yr
-
1
) identified as severe condition of extent. In conclusion, the SWAT model could be
effectively used to predict runoff and sediment yield in order to effectively design and
plan water related development in absence of gauged information.
