Comparison of Newton-type and direct search algorithms for calibration of conceptual rainfall-runoff models

作者: Jene' D. Hendrickson , Soroosh Sorooshian , Larry E. Brazil

DOI: 10.1029/WR024I005P00691

关键词:

摘要: An examination of the calibration aspect conceptual rainfall-runoff models was undertaken using Sacramento soil moisture accounting model and a study comparing performance Newton-type optimization algorithm with that direct search algorithm. Results indicate is more robust two because susceptible to poor conditioning response surface. Graphical studies surface model's parameter space confirmed presence discontinuities rough-textured surface, particularly in derivatives.

参考文章(32)
Norman Holmes. Crawford, Ray E. Linsley, DIGITAL SIMULATION IN HYDROLOGY' STANFORD WATERSHED MODEL 4 Calif.] Stanford University. ,(1966)
George A. Bekey, Sami F. Masri, Random search techniques for optimization of nonlinear systems with many parameters Mathematics and Computers in Simulation. ,vol. 25, pp. 210- 213 ,(1983) , 10.1016/0378-4754(83)90094-0
R. J. Moore, R. T. Clarke, A distribution function approach to rainfall runoff modeling Water Resources Research. ,vol. 17, pp. 1367- 1382 ,(1981) , 10.1029/WR017I005P01367
P. R. Johnston, D. H. Pilgrim, Parameter optimization for watershed models Water Resources Research. ,vol. 12, pp. 477- 486 ,(1976) , 10.1029/WR012I003P00477
R. Fletcher, M. J. D. Powell, A Rapidly Convergent Descent Method for Minimization The Computer Journal. ,vol. 6, pp. 163- 168 ,(1963) , 10.1093/COMJNL/6.2.163