Documente Academic
Documente Profesional
Documente Cultură
=
= = =
m p
N
i
N
i
m
N
i
p m P
S S N
Q Q Q Q
R
(1)
DR =
m
P
Q
Q
log
(2)
Accuracy = 1 (accurate, if DR lies between -0.1 to 0.1),
otherwise 0. (3)
where, Q
p
is the predicted discharge and, Q
m
is the
measured discharge, S
p
and S
m
are standard deviations of
predicted and measured values, respectively and N is the
number of observation. From the Eq.(1), it is clear that
DR = 0 suggests that the predicted value of the discharge
is identical to the measured discharge; if DR > 0, the
predicted value of the discharge overestimates, and if
DR< 0, it underestimates. The performance indices of
newly derived expressions utilizing GA along with those
employing RMCD are given in Table (4). Table (4)
concludes the best predictive model in the present case is
the fourth model with one input parameter, i.e., index
velocity (U
i
). The model has the highest accuracy at
89.4% and the largest R
2
at 0.98, The prediction by the
second model with input parameters U
i
and A is also
satisfactory However, the first and the second models are
found to be inadequate in forecasting discharge in a tidally
affected river. Here, both accuracy and R
2
are less than
satisfactory.
CONCLUSION
Knowledge of accurate discharge through a river is
essential for water allocation, water diversion, flood
fighting strategies and water rights the traditional rating
curves and available empirical expressions for estimating
discharge in tidally affected rivers are evaluated and were
found inadequate. In the present work, implementing
genetic algorithm on published field data, new empirical
expressions for estimating discharge in tidally affected
rivers have been derived. The proposed expressions use
few hydraulic and geometric characteristics of a river
which are normally available to field engineers. A case
study has been made on Samrangjin gauging station of
River Nakdong in South Korea. Performance of new
expressions is compared with those of Lee and Cheong
(2009). Based on various performance indices, i.e.,
coefficient of determination, discrepancy ratio and
accuracy, new expressions are found superior to other
considered expressions. The best performing GA model,
with only mean index velocity as input, has coefficient of
determination equal to 0.98 and accuracy as 88.9%.
241 Predicting Discharge in Tidally Affected Rivers Using Genetic Algorithm
International Journal of Earth Sciences and Engineering
ISSN 0974-5904, Volume 04, No 06 SPL, October 2011, pp 239-242
Table 2. Geometric and hydraulic characteristics of Nakdong River at Samrangjin station in S. Korea ( Lee and Cheong 2009)
* Datasets used for derivation set.
Date
Q
m
3
/s
H
m
T
m
D
m
W
s
m
A m
2
U
i
m/s
U
m/s
*13-01-2005 119.35 0.83 0.35 0.48 356 2589.65 0.044 0.05
*21-01-2005 83.620 0.73 0.12 0.61 355 2570.13 0.026 0.03
31-01-2005 91.940 0.81 0.72 0.09 355 2588.38 0.018 0.04
21-02-2005 143.42 0.81 0.52 0.29 355 2552.25 0.073 0.06
*02-03-2005 111.75 0.80 0.39 0.41 355 2596.88 0.047 0.04
10-03-2005 166.02 0.75 0.13 0.62 356 2406.53 0.092 0.07
*18-03-2005 225.27 0.77 0.33 0.44 356 2488.00 0.077 0.09
28-03-2005 352.51 0.82 0.18 0.64 356 2525.55 0.179 0.14
*06-04-2005 196.34 0.70 -0.11 0.81 356 2502.00 0.098 0.08
*14-04-2005 262.67 0.74 0.15 0.59 355 2536.75 0.136 0.10
25-04-2005 271.81 0.66 -0.41 0.80 356 2531.25 0.132 0.11
*02-05-2005 120.97 0.76 0.23 0.53 357 2523.40 0.061 0.05
11-05-2005 270.80 0.77 0.26 0.51 357 2501.25 0.145 0.11
19-05-2005 175.72 0.68 0.03 0.65 356 2496.90 0.104 0.07
27-05-2005 147.80 0.75 0.52 0.23 357 2512.95 0.085 0.06
*07-06-2005 274.69 0.70 0.07 0.63 357 2501.65 0.133 0.11
*15-06-2005 257.27 0.68 -0.17 0.85 357 2486.05 0.155 0.10
23-06-2005 166.55 0.68 0.11 0.57 357 2488.35 0.101 0.07
01-07-2005 554.30 0.77 0.19 0.58 357 2551.25 0.337 0.22
*11-07-2005 1093.7 1.17 0.29 0.88 361 2823.50 0.543 0.39
19-07-2005 440.93 0.77 0.31 0.46 356 2438.25 0.232 0.18
*27-07-2005 301.97 0.67 -0.11 0.78 356 2143.10 0.193 0.14
04-08-2005 1805.0 1.58 0.67 0.91 365 2893.00 0.766 0.62
*12-08-2005 982.30 0.93 0.91 0.02 357 2450.08 0.450 0.40
*23-08-2005 2197.9 2.26 1.33 0.93 357 2808.18 0.888 0.78
30-08-2005 615.40 0.85 0.41 0.44 350 2303.75 0.308 0.27
*08-09-2005 483.49 1.01 0.21 0.80 360 2325.75 0.173 0.21
16-09-2005 263.48 0.87 0.57 0.30 357 2238.98 0.131 0.12
*27-09-2005 439.75 0.70 0.15 0.55 350 2352.00 0.232 0.19
06-10-2005 319.36 0.75 0.26 0.49 352 2258.75 0.172 0.14
14-10-2005 169.71 0.76 0.48 0.28 352 2256.40 0.090 0.08
*24-10-2005 114.72 0.83 0.64 0.19 352 2274.70 0.077 0.05
*02-11-2005 135.16 0.74 -0.02 0.76 352 2222.30 0.088 0.06
*09-11-2005 99.570 0.73 0.54 0.19 352 2216.70 0.043 0.04
17-11-2005 163.15 0.66 0.25 0.41 352 2186.55 0.066 0.07
25-11-2005 133.70 0.74 0.36 0.38 352 2207.80 0.086 0.06
*05-12-2005 123.65 0.75 0.46 0.29 352 2202.50 0.085 0.06
*13-12-2005 129.08 0.76 0.29 0.47 352 2231.40 0.086 0.06
21-12-2005 161.58 0.75 0.29 0.46 343 2222.05 0.084 0.07
242
Saurabh Sharma, Prithvi Raj Jain, Rajeev Ranjan Sahay
International Journal of Earth Sciences and Engineering
ISSN 0974-5904, Volume 04, No 06 SPL, October 2011, pp 239-242
Table 3. Statistical characteristics of datasets
Fig.3. Measured and predicted discharges by GA models
(Verification set) x
REFRENCES
[1] ASCE (1996). Task Committee on Hydrology
Handbook of Management Group D., 2
nd
Ed., New
York, 752-765.
[2] Deb K (2002). Multi-Objective optimization using
evolutionary algorithms. John Wiley and Sons Asia.
[3] Goldberg DE (2001) Genetic Algorithms In: Search,
Optimization and Machine Learning. Addison-
Wesley New York.
[4] Lee, S. And Cheong, T. S. (2009). Development of
regression equations for the water discharge
estimation n tidally affected rivers. KSCE journal of
civil eng. 13(3), 195-203.
[5] Michalewicz Z (1992). Genetic algorithm + data
structures = evolutionary programs. Springer, New
York.
Equations
Discharge
Ranges
Estimates
R
2
Accuracy
(%)
c d
Q= H
D
GA 83-950 1521.4 4.118 0.793 - - 0.35 55.5
RMCD
83-950 1502.3 3.692 0.893 - - 0.34 44.4
Q =(H+ c)
(D+ d)
GA
83-950 1510.2 4.784 0.857 0.038 0.002 0.39 44.4
RMCD
83-950 1794.7 4.816 0.956 0.02 0.03 0.49 38.9
Q = ( +U
i
)A
GA
83-2200 -0.002 0.828 - - - 0.98 84.2
RMCD
83-2200 -0.002 0.829 - - - 0.98 84.2
Q = + c( +U
i
)
GA
83-2200 -0.096 46.46 0.023 49.04 - 0.98 88.9
RMCD 83-2200 -0.002 0.829 -1.417 2411.8 - 0.98 94.4