46
4.2.2 Model Calibration and Verification
There are three years paired observed rainfalls and discharge data were used to calibrate and validate the model. The calculation methods that have been
employed to the model are: NRCS curve number method to calculate loss, Snyder for transform method, and recession for base-flow method. Table 4.8 shows the
adjusted parameters during calibration period. Table 4.8 Set of calibrated parameters on Palu Catchment
Parameter Range
Value Initial abstraction scale factor
0 – 0.5 0.05
Baseflow threshold for sub-basin 1
0 – 1 0.22
Baseflow threshold for sub-basin 2
0 – 1 0.06
Baseflow threshold for sub-basin 3
0 – 1 0.06
Baseflow threshold for sub-basin 4
0 – 1 0.1
Baseflow threshold for sub-basin 5
0 – 1 0.04
Baseflow threshold for sub-basin 6
0 – 1 0.22
Baseflow threshold for sub-basin 7
0 – 1 0.05
Comparison between simulated and observed hydrograph during calibration process shows the good fit, even though there are some deviations or
errors. During calibration process by using one-year daily data, the accuracy achieved is quite good with 0.758 of R
2
. It might be due to the areal rainfall pattern spatially doesn’t represent the real pattern in each sub-basin area and also
due to the simplification of some parameters under the model.
5 10
15 20
25 30
50 100
150 200
250 300
350 400
450
Di sch
a rge
m
3
s
Date
Precipita\on mm Simulated
Observed
47
Figure 4.8 Simulation versus observation hydrograph, and correlation between simulation and observed discharges during calibration
process. Performance of the model was objectively evaluated by using Nash-
Sutcliffe efficiency and relative volume error, in which it gave good efficiency value of 0.81 and 42.9 of relative volume errors. By using three tests it can be
stated that the model is satisfactory accepted. For advanced evaluation, the calibrated parameters was verified using two years independent paired observed
rainfall-discharge data on 2006 and 2007.
R² = 0.75882 100
200 300
400 500
50 100
150 200
250 300
350 400
450
Q
Sim ulate
d
m
3
s
Q
Observed
m
3
s
5 10
15 20
25 30
35 40
50 100
150 200
250 300
350 400
450
Preci p
it a
Uo n
mm
Di sch
a rge
m
3
s
Date
Precipita\on Simulated
Observed
48
Figure 4.9 Simulation versus observation hydrograph, and correlation between simulation and observed discharges during verification
process. Figures 4.9 demonstrate verification of calibrated parameters that have been done
previously. Here the performance of calibrated parameters was evaluated using Nash-Sutcliffe efficiency and relative volume errors. Overall it gave values 0.9 of
Nash-Sutcliffe efficiency and 12.1 of relative volume errors. The correlation coefficient of R
2
during verification process was quite low, achieving only 0.58 of RMS value. From the correlation graph between observed and simulation
discharges shows that both values have strong correlation.
4.2.3 Changes in Seasonal Stream Flow
In order to analyze the annual seasonal stream flow, twenty years of daily hydrograph was simulated. The simulation used twenty years rainfall data that
recorded by eight rain gauges spread out on the catchment. Based on the simulated hydrograph, it can be indicated that annual peak flow occurred in 1994,
2000, and 2007. The dash-circle indicates the peak flow of the years.
R² = 0.57642
100 200
300 400
100 200
300 400
Q
Sim ulate
d
m
3
s
Q
Observed
m
3
s
49
Figure 4.10 Twenty years simulated hydrograph in Palu River. Wet season high flow; in order to analyze the seasonal stream flow, period
of April - June and October – December was selected as the wettest months, while the peak on May and November. January-March and August-September were
considered as driest month. During the wet season, rainfall increased and reached the peak on May and November. Due to the increasing of rainfall intensity, large
amount of stream flow was generated from surface runoff. During the 1990 to 2009 period, the stream flows fluctuated about 73 to 288 m
3
s, and reach the peak flow on May 2007 as 2007 big flood was occurring in Palu.
Figure 4.11. Variations between high wet seasons flow, low dry season flow and peak to low ratio.
5 10
15 20
25 30
35 40
50 100
150 200
250 300
350
P mm Discharge m
3
s
Years
50 100
150 200
250 300
350
1990 1991
1992 1993
1994 1995
1996 1997
1998 1999
2000 2001
2002 2003
2004 2005
2006 2007
2008 2009
Di sch
a rge
M
3
s
Years
Wet Dry
Ra\o