Page - 268 - in Book of Full Papers - Symposium Hydro Engineering

Thirdly, three time-trend function are included: (1) Stationary, not varied with time (S) (2) linear form of a trend (L) (3) parabolic form (P) The Akaike Information Criterion (AIC) is applied to identify the optimum model, which possesses the minimum AIC value. AIC = -2 ln ML + 2k [1] Where, ML denotes the maximum likelihood, k is the number of fitted parameters. The results for curve fitting of flood peak at Shigu, Jinjiangjie and Panzhihua hydrological stations are as follows[15]: Table1 AIC value of Gumbel, Lognormal and Gamma distribution for flood peaks at Sigu, Jinjiangjie and Panzhihua hydrological stations model Gumbel LogNo Gamma Gumbel LogNo Gamma Gumbel LogNo Gamma AIC value of Shigu station AIC value of Jinjiangjie station AIC value of Panzhihua station ASBS 930.56 917.18 917.27 910.64 896.04 895.42 900.82 883.69 883.73 ALBS 932.51 919.17 919.27 912.64 897.63 897.08 902.81 885.53 885.53 APBS 933.08 918.53 918.92 912.44 898.02 897.15 900.60 881.19 881.59 ASBL 932.21 919.14 919.24 912.04 898.02 897.34 901.65 885.40 885.65 ALBL 934.20 921.11 921.22 913.72 899.54 898.90 903.25 887.36 887.52 APBL 935.08 920.52 920.90 912.83 899.35 898.33 901.15 883.15 883.42 ASBP 934.14 921.02 920.98 913.87 899.26 899.03 903.48 887.35 887.39 ALBP 936.14 923.01 922.98 915.66 901.18 900.81 904.92 889.25 888.99 APBP 936.72 922.15 922.40 914.83 901.07 900.22 902.90 884.44 884.68 The results indicate that the best curve fitting for flood peaks of Shigu, Jinjiangjie and Panzhihua stations are stationary LogNormal, stationary Gamma distribution and nonstationary Lognormal distribution with linear trend of mean value and stationary standard deviation. The 1st to 3rd reservoirs are located between Shigu and Jinjiangjie. Meanwhile, the 4th reservoir lies between Jinjiangjie and Panzhihua, it is much 268