3.1 Flow Sensitivity Analysis
Initially twelve parameter were identified to select the most sensitive parameters during flow calibration. Monthly streamflow input data from 1993 to 2010 was used for flow sensitivity parameter identification.
From 100-iteration output, eight parameters were assigned as sensitive for further calibration process. The highest flow sensitive parameters were the curve number (CN2), soil evaporation compensation factor (ESCO), available water capacity of the soil layer (SOL_AWC).
Table 2
Grid view of Stream flow sensitive parameter.
|
Parameter
|
Description of parameter
|
P-Value
|
t-Stat
|
Range
|
Fitted value
|
rank
|
|
Minimum
|
Maximum
|
|
1
|
R-CN2.mgt
|
SCS curve number II
|
0
|
12.7
|
-0.1
|
0.8
|
0.34
|
1
|
2
|
V-GW_REVAP
gw
|
Ground water revap coefficient
|
0.18
|
-1.34
|
0.1
|
0.2
|
0.17
|
3
|
3
|
V-OV_N.hr
|
Manning's "n" value for overland flow
|
0.64
|
0.46
|
0.1
|
1
|
0.69
|
7
|
4
|
R-REVAPMN.
gw
|
Threshold depth of water in the shallow aquifer
|
0.24
|
1.16
|
0
|
0.2
|
0.04
|
5
|
5
|
A-CANMX.hr
|
maximum canopy storage
|
0.23
|
-1.19
|
0
|
1
|
0.33
|
4
|
6
|
V-ALPHA_BF.
gw
|
Base flow alpha factor
|
0.38
|
0.86
|
0
|
0.3
|
0.07
|
6
|
7
|
R-ESCO.bsn
|
Soil evaporation compensation factor
|
0.89
|
0.90
|
0
|
0.6
|
0.08
|
8
|
8
|
R-SOL_AWC.
sol
|
Available water capacity of the soil layer
|
1.58
|
0.12
|
1
|
1.6
|
1.52
|
2
|
3.2 Flow Calibration
The available Stream flow data from the period January 1993 to December 2003 used for calibration and the obtained results for coefficient of determination (R²0.81) was very good, Nash Sutcliffe Efficiency (NSE) was very good and Percent bias (Pbias) which was between The observed and simulated streamflow during calibration period was 4.68 m3/sec and 3.83 m3/sec.
Table 3
Observed and simulated flow at the calibration period
Time (year)
|
Average Streamflow (m³/s)
|
Model Efficiency (Monthly)
|
1993-2003
|
Observed
|
Simulated
|
R²
|
NSE
|
PBIAS\(\left(\mathbf{\%}\right)\)
|
4.68
|
3.83
|
0.81
|
0.76
|
18
|
3.3 Flow Validation
The model validation carried out from January 2004 to December 2010 and the obtained results for coefficient of determination (R²0.81) was very good, Nash Sutcliffe Efficiency (NSE) was good and Percent bias (Pbias) which was between. The observed and simulated average stream flow of Meki gauging station during calibration period was 5.3m3/sec and 4.4m3/sec respectively.
Table 4
Observed and simulated flow at the validation period
Time (year)
|
Average Streamflow (m³/s)
|
Model Efficiency (Monthly)
|
2004-2010
|
Observed
|
Simulated
|
R²
|
NSE
|
PBIAS\(\left(\%\right)\)
|
5.3
|
4.4
|
0.81
|
0.74
|
17.1
|
3.4 Sediment Rating Curve Development
The sediment rating curve is a relationship between the river discharge and sediment concentration. It is widely used to estimate the sediment load being transported by a river. Generally, a sediment rating curve may be plotted showing average sediment concentration or load as a function of discharge averaged over different periods. Since the sediment measurement in the watershed is less, a rating curve is developed to estimate sediment yield from flow measurement (Aga et al., 2018). The Sediment flow measurement in the Meki River was not in continuous time step; so that by using stream flow and measured sediment data can generate sediment rating curve.
3.5 Sediment Computation for the Model Performance Evaluation
From the Fig. of 7 and 8, the results show that model performance was very good (R2\(=0.8743>0.8\)) and well agreement between sediment obtained from rating curve and simulated sediment obtained from SWAT. The total annual sediment yield from watershed in to the Lake Batu during Simulation period was estimated by using SWAT model was 75.896 ton/ha/yr.
3.6 Runoff Spatial Distribution
Fig. 9 shows the different surface runoff within all 35 sub-basins as annual averages over the 25-year simulation period (1993–2017). The result shows not only the streamflow at the gauge was an important variable for analyzing the water balance of a watershed, but also the spatial patterns within the entire watershed was important. Study of ( Hurni 1988) confirmed that sub basins can be classified as none to slight (), slight (80-130), medium (130-220), high (220-612), and very high () mm for runoff.
From SWAT simulation the runoff produced at the Meki River Watershed was 367.95 mm. As shown in Fig. 9, runoffs have no impact on sediment yield at some sub basin. For instance, in sub-basins (1-8), 10, 11,12,13,18, 20-26, 31 and 34 there was high runoff but less sediment yield. This was may be the response of LU/LC, soil resistance to erosion, slope and other management practice founded in the watershed. The sub-watersheds that produce moderate surface runoff were 9, 14, 19, 28, 32, 33 and 35. The sub-watersheds that produce slight surface runoff were 16, 17, 27 and 30 besides none to slight surface runoff were in 15 and 29. These simulation results show the relative variations of Runoff level within a sub-basin. Moreover, these results showed that Runoff to Meki River watershed was mainly from sub-basins of (1-8), 10, 11, 12, 13, 18, 20-26, 31 and 34. These sub-basin, which fall under high Runoff, was characterized by intensive cultivated land which leads to high Runoff susceptibility of the watershed.
3.7 Sediment Yield Spatial Distribution
The sediment yield of all 35 sub-basins, represented as annual values over 25 years (1993–2017) was shown in Fig.10. From SWAT simulation the sediment yield produced at the Meki River watershed was 75.896 ton/ha/yr. These simulation results show the relative variations of soil erosion level within a sub-basin. Study of Hurni (1988) confirmed that sub basins can be classified as none to slight (), slight (10-21), medium (22-61), high (62-109), and very high () ton/ha/yr. for sediment. The moderate sediment yielding sub-basins were (1-8), 10, 11, (14-27) and (32-34) of Meki River watershed corresponds to moderate erosion level (22–61 ton/ha/yr.). These sub-basins were covered most areas by intensive cultivated lands followed by moderate cultivated land.
The sediment yield distribution, for instance, sub-basins 9, 30 and 35 were high and characterized by maximum sediment yield distribution to Meki River watershed. The slight sediment yielding sub-basins were 12, 28, and 31. They deliver least sediment may be due to well-covered land use/cover. The erosion level was indicated as none to slight sediment in sub-basin 29. The sub-basins 9, 30 and 35, which fall under high erosion class, led to high sediment susceptibility of the watershed.