27 2 2008 3 GEO GRA P HICAL RESEA RCH Vol127, No12 Mar1, 2008 1,5, 1,6, 3, 2, 3, 4 (11, 100101 ; 21, 100101 ; 31, 210098 ; 41, 450004 ; 51, 100049 ; 61, 100081) : GL U E,, :, UM, (R 2 ), CS, R 2, B, R 2, : GL U E ; ; ; ; ; : 100020585 (2008) 0220343210 1 [1 ] : 1), ; 2), ; 3),,,,, ( Genetic Algorit hm, GA) [2,3 ] (Simulated Annerling,SA) [4 ] (Artificial Neural Network,ANN) [5,6 ],,, Beven (1992) ( GL U E, Gen2 eralized Likelihood Uncertainty Estimation), [7 ] Beven (2001) [ 8 ] Tawhai ( 01038km 2 ), : 2007206209 ; : 2007211223 : (40671032, 40671033) ; ( CCSF2006230) ; : (19832),,, E2MAIL : shuchang6666 @1631com 3 : (19652),,,, E2MAIL : liusx @igsnrr1ac1cn
344 27 GL U E TOPMOD EL, GL U E (2004) [9 ] GL U E,, GL U E,, [ 10 ] GL U E (2004, 2006), CO2, GL U E [11 ], L ISFLOOD2WB 8 [12 ] ; Gallart (2006) [13 ] Top model,,, ; (2006) [14 ] GL U E, GL U E, [10,15 17 ] [ 18 22 ],, [ 23 25 ], GL U E,,,, 2 211 1 [20 ] Fig11 Xinanjiang model flow chart
2 : 345 : 1) ; 2) : ; 3) ; 4) 4 : :, KC UM L M C ; :, WM B IMP ; :, SM EX KG KI ; :, CI CG CS KE XE KE XE,,, 1 ; 1 [20 ] 1 Tab11 Parameters and their sampled ranges of Xinanjiang model GL U E UM (mm) 5-20 0-50 L M (mm) 60-90 0-150 DM (mm) 0-100 WM (mm) WM = UM + L M + DM B 01 2-01 4 0-1 EX 11 5 015-215 KC 1 0-2 C 01 1-01 2 0-015 SM (mm) 5-45 0-200 KG KI + KG = 017 0 < KI + KG < 1 KI CG 01 95-01995 015-1 CI 01 3-01 8 0-1 CS 0-1 IMP 0101-01 05 0-011 212 GL UE,, ( Equifinality), GL U E, [7 ], [ 9 ] 100000 ( 1),, Nash R 2, GL U E R 2 = 1 - n c i = 1 n c ( Qi - ^Qi) 2 i = 1 ( Qi - gqc) 2 (1) (1) : R 2 ; Qi ; ^Qi ; gqc ;
346 27 nc 3 ( 114 50 115 10, 23 06 23 23 ),,,,, ( 109 40 110 10, 33 45 34 30 ),,,,,,, 34 %, 015 017,, 2 -, [18 ], 2 2 Tab12 Basic hydrological data of two basins (mm) (km 2 ) 18681 73 01 55 385 1978 1987 597142 01 31 4423-1980 1988 2 ( (a) (b) ) Fig12 Relationship between annual rainfall and runoff of (a) Jiuzhou and (b) L ushi 4 411, 3 E = N R2 > m N Total (2) : E ; N R 2 > m R 2 > m ; N Total GL U E ; m R 2 3, R 2 ( R 2 = 015 ) ;, (2)
2 : 347,, R 2 3 Fig13 Comparisons of simulation efficiency in two basins 412,, GL U E,, [7 ], 3 3 Tab13 Several equif inality parameter groups 1 2 3 1 2 3 UM ( mm) 281 54 191 81 161 94 381 27 9157 261 28 L M ( mm) 4119 461 47 132146 3110 261 12 331 92 DM ( mm) 711 13 381 97 421 81 371 11 6151 181 72 B 0158 0134 0151 0103 0117 0115 EX 0182 1112 1120 0164 0171 0191 KC 1114 0194 1138 0115 0144 0110 C 0132 0119 0109 0137 0129 0119 SM ( mm) 100181 851 43 881 71 981 65 381 67 141160 KG 0125 0148 0146 0185 0163 0150 KI 0125 0150 0151 0100 0103 0101 CG 0192 0187 0181 0172 0157 0168 CI 1100 0199 0197 0149 0122 0169 CS 0160 0159 0159 0170 0159 0167 IMP 0109 0102 0107 0106 0105 0104 R 2 0179 0179 0179 0170 0170 0170
348 27, : X ij = X ij m X ij i = 1 (3) : X ij ; X ij ( i, ), m (3) j 4 ( R 2 ), L M, CS, 4 ( (a) ; (b) ) Fig14 Total standardization of equifinality parameter group s in (a) Jiuzhou and (b) L ushi 413, ( 5 7), :,, ( 5) UM L M DM EX CG, R 2, 5 (UM) ( (a) ; (b) ) Fig15 Scatter plots of the likelihood of UM for daily discharge in (a) Jiuzhou and (b) L ushi :,, ( 6) KI + KG KC CS KI ( ), KI + KG ( 6 (a), (b) ), KI + KG [ 20 ],
2 : 349
350 27,, KC ( ) CS ( 6 (c), (d) ),, R 2, : ( 7) B C IMP KG CI WM SM B ( ), C ( 7 (b) ), IMP ( ), KG 1 ( ), WM (WM = UM + L M + DM, ) ( 7 (f) ) CI 1 ( 7 (c) ), SM ( ),,,,,, 5 GL U E : 1) 2) GL U E ( ),, GL U E 3), : UM L M DM EX CG,, R 2 ; KI KI + KG KC CS,, R 2, ; C IMP WM B SM KG CI,,,,, : 1),,,,, 2),, 3),,, 4), GL U E, 5),, [12 ],
2 : 351 : : [ 1 ] Sivapalan M K, Takeuchi S W, Franks S W1 IA HS decade of prediction in ungauged basins ( PUB), 2003-2012 : Shaping an exiting future for t he hydrological sciences1 Hydrological Sciences Journal,2003,48 (6) :857 8791 [ 2 ], 1 1,2006,26 (4) :32 381 [ 3 ],, 1 1,2004, (2) :50 561 [ 4 ], 1 1,2001,20 (1) : 97 1021 [ 5 ],, 1 1,1998,17 (4) :352 3591 [ 6 ], 1 1,1999,18 (4) :382 3901 [ 7 ] Beven K, Binley A M1 The fut ure of distributed models : Model calibration and uncertainty prediction1 Hydrologi2 cal Processes,1992,6 (3) :279 2981 [ 8 ] Keit h Beven, Jim Free1 Equifinality,data assimilation, and uncertainty estimation in mechanistic modeling of com2 plex environmental systems using t he GL U E met hodology1 Journal of Hydrology,2001,249 :11 291 [ 9 ], 1 GL U E 1 : 1 1 :,20041143 1501 [ 10 ],,, 1 1,2007,22 (4) : 649 6551 [ 11 ] Mo Xingguo, Beven K1 Multi2objective parameter conditioning of a t hree2source wheat canopy model1 Agricultural and Forest Meteorology, 2004,122 : 39 631 [ 12 ] MO Xingguo, et al1 Parameter conditioning and prediction uncertainties of t he L ISFLOOD2WB distributed hydro2 logical model1 Hydrological Sciences2Journal2des Sciences Hydrologiques,2006,51 (1),45 651 [13 ] Gallart F,et al1 Using internal catchment information to reduce t he uncertainty of discharge and baseflow predic2 tions1 Advances in Water Resources,2006 :1 161 [ 14 ] Zhang Danrong, Beven K, Mermoud A1 A comparison of non2linear least square and GL U E for model calibration and uncertainty estimation for pesticide t ransport in soils1 Advances in Water Resources, 2006, 29 ( 12) : 1924 19331 [ 15 ] Free J, Beven K J,Ambroise B1 Bayesian estimation of uncertainty in runoff prediction and t he value of data : An application of t he GL U E approach1 Water Resources Research,1996,32 (7) :2161 21731 [ 16 ] Keit h B, Zak S1 Equifinality, sensitivity and uncertainty in t he estimation of critical load1 Science of t he Total En2 viron ment,1999,236 :191 2141 [ 17 ] Romanowicz R J,Beven K J1 Comment s on generalized likelihood uncertainty estimation1 Reliability Engineering and System Safety,2006,91 :1315 13211 [ 18 ] 1 1 :,19861 140 1541 [ 19 ] 1 1 :,19841 11 1481 [ 20 ], 1 1,1988, 6 :2 91 [ 21 ] 1 1 :,19951 1 231 [ 22 ], 1 ( ) 1,1989,17 (4) :65 691 [ 23 ], 1 1,2006,28 (4) :519 5251 [ 24 ], 1 GL U E 1,2006,24 (259) :31 471 [ 25 ], 1 1 : 1 1 :,20041 151 1551
352 27 Uncertainty analysis of Xinanjiang model parameter SHU Chang 1,5, L IU Su2xia 1,6, 3, MO Xing2guo 2, L IAN G Zhong2min 3, DA I Dong 4 (11 Key Lab of Water Cycle & Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research ( IGSNRR), CAS, Beijing 100101, China ; 21 Key Lab of Ecological Net Observation and Modeling, IGSNRR, CAS, Beijing 100101, China ; 31 State Key Laboratory of Hydrology2Water Resources and Hydraulic Engineering, Hohai University, Nanjing 210098,China ; 41Bureau of Hydrology, Yellow River Conservancy Commission, Zhengzhou 450004, China ; 51 Graduate School of the Chinese Academy of Sciences,Beijing 100049, China ; 61 National Meteorological Center, China Meteorological Administration, Beijing 100081, China) Abstract :The uncertaint y p ro blem in hydrological model is an impo rtant issue of scientific research at p resent, which cover s t hree aspect s of data, model st ruct ure and parame2 ters1 Parameter is one of t he key roles in analyzing model uncertainty problem1 The value of parameters depends on characteristics of a basin, but in fact it is difficult to obtain be2 cause t here are few o bservatio n statio ns1 In general, it needs to co nfirm parameter s by several calibratio n met hods including Genetic Algo rit hm, Simulated Annerling and Artifi2 cial Neural Network1 So t here exist s parameter uncertainty problem1 The generalized like2 lihood uncertainty estimation ( GL U E) met hodology is an effective approach to st udy un2 certainty of parameters1 In this paper, t he uncertainty in Xinanjiang model is examined by employing GL U E1 Based on t he simulation result s of daily data f rom Jiuzhou (1978 1987) and L ushi (1980 1988) basins, it is found t hat t he p henomenon of equifinality exist s among parameters group s for bot h of t he basins1 According to comparison result of scatter plot s, parameters of Xinanjiang model can be classified into t hree group s : sensitivity pa2 rameters such as UM, EX ; non2sensitivity parameters such as KC, CS and regional sensi2 tivity parameters such as B, WM1 The conclusion is favorable for understanding parame2 ter s of Xinanjiang model so as to p rovide valuable scientific informatio n for simulating hydrological processes1 Finally it p ut s forward t he main content s on f ut ure uncertainties research in hydrological modeling1 Key words :GL U E met hodology ; Xinanjiang model ; equifinality ; uncertainty ;L ushi basin ; Jiuzhou basin