Improved methodology for parameter inference in nonlinear, hydrologic regression models

A new method is developed for the construction of reliable marginal confidence intervals and joint confidence regions for the parameters of nonlinear, hydrologic regression models. A parameter power transformation is combined with measures of the asymptotic bias and asymptotic skewness of maximum likelihood estimators to determine the transformation constants which cause the bias or skewness to vanish. These optimized constants are used to construct confidence intervals and regions for the transformed model parameters using linear regression theory. The resulting confidence intervals and regions can be easily mapped into the original parameter space to give close approximations to likelihood method confidence intervals and regions for the model parameters. Unlike many other approaches to parameter transformation, the procedure does not use a grid search to find the optimal transformation constants. An example involving the fitting of the Michaelis-Menten model to velocity-discharge data from an Australian gauging station is used to illustrate the usefulness of the methodology.

[1]  Keith Beven,et al.  On the generalized kinematic routing method , 1979 .

[2]  George Kuczera,et al.  Estimation of runoff-routing model parameters using incompatible storm data. , 1990 .

[3]  Nonlinear, discrete flood event models, 2. Assessment of statistical nonlinearity — Correction , 1990 .

[4]  D. G. Watts,et al.  Parameter Transformations for Improved Approximate Confidence Regions in Nonlinear Least Squares , 1981 .

[5]  G. J. S. Ross,et al.  The Efficient Use of Function Minimization in Non‐Linear Maximum‐Likelihood Estimation , 1970 .

[6]  George Kuczera,et al.  On the validity of first-order prediction limits for conceptual hydrologic models , 1988 .

[7]  Bryson C. Bates,et al.  Nonlinear, discrete flood event models, 1. Bayesian estimation of parameters , 1988 .

[8]  Richard L. Cooley,et al.  A method of estimating parameters and assessing reliability for models of steady state Groundwater flow: 2. Application of statistical analysis , 1979 .

[9]  D. Ratkowsky.  Statistical Properties of Alternative Parameterizations of the von Bertalanffy Growth Curve , 1986 .

[10]  B. Bates,et al.  Nonlinear, discrete flood event models, 3. Analysis of prediction uncertainty , 1988 .

[11]  Asymmetry of estimators in nonlinear regression , 1987 .

[12]  Philip Hougaard,et al.  Parametrizations of Non‐Linear Models , 1982 .

[13]  J. Witmer,et al.  A Note on Parameter-Effects Curvature , 1985 .

[14]  Robert B. Schnabel,et al.  Computational experience with confidence intervals for nonlinear least squares , 1986 .

[15]  B. Bates.  Nonlinear, discrete flood event models, 2. Assessment of statistical nonlinearity , 1988 .

[16]  P. Hougaard.  The Appropriateness of the Asymptotic Distribution in a Nonlinear Regression Model in Relation to Curvature , 1985 .

[17]  G. P. Clarke,et al.  Approximate Confidence Limits for a Parameter Function in Nonlinear Regression , 1987 .

[18]  R. Clarke.  A review of some mathematical models used in hydrology, with observations on their calibration and use , 1973 .

[19]  J. Witmer,et al.  Nonlinear Regression Modeling. , 1984 .

[20]  S. P. Neuman,et al.  Estimation of Aquifer Parameters Under Transient and Steady State Conditions: 1. Maximum Likelihood Method Incorporating Prior Information , 1986 .

[21]  Philip Hougaard.  The asymptotic distribution of nonlinear regression parameter estimates: improving the approximation , 1988 .

[22]  Chih-Ling Tsai.  Power transformations and reparameterizations in nonlinear regression models , 1988 .

[23]  G. Kuczera.  Improved parameter inference in catchment models: 1. Evaluating parameter uncertainty , 1983 .

[24]  Bryson C. Bates,et al.  Use of parameter transformations in nonlinear, discrete flood event models , 1990 .

[25]  D. H. Pilgrim.  Travel Times and Nonlinearity of Flood Runoff From Tracer Measurements on a Small Watershed , 1976 .

[26]  Soroosh Sorooshian,et al.  The relationship between data and the precision of parameter estimates of hydrologic models , 1985 .

[27]  M. Newson,et al.  Channel studies in the Plynlimon experimental catchments , 1978 .

[28]  Richard L. Cooley,et al.  Simultaneous confidence and prediction intervals for nonlinear regression models with application to a groundwater flow model , 1987 .

[29]  B. Troutman.  Errors and Parameter Estimation in Precipitation‐Runoff Modeling: 1. Theory , 1985 .

[30]  G. Kuczera.  Improved parameter inference in catchment models: 2. Combining different kinds of hydrologic data and testing their compatibility , 1983 .

[31]  S. Sorooshian,et al.  Stochastic parameter estimation procedures for hydrologie rainfall‐runoff models: Correlated and heteroscedastic error cases , 1980 .