System identification and validation for output prediction of a dynamic hydrologic process

The paper presents an identification procedure for a dynamic model of am hydrologic process. The process involves solute transport in streams subject to aquifer interaction and unsteady flows and the intended use of the model is prediction. Detailed assumptions and results are provided to illustrate...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of forecasting Ročník 10; číslo 3; s. 319 - 346
Hlavní autoři: Jakeman, A. J., Thomas, G. A., Dietrich, C. R.
Médium: Journal Article
Jazyk:angličtina
Vydáno: Chichester John Wiley & Sons, Ltd 01.05.1991
Wiley
Wiley Periodicals Inc
Témata:
Algorithms
Forecasts
Hydrology
Identification
Instrumental variables
Monte Carlo simulation
Predictions
Recursion
Specification
Statistical analysis
Stochastic models
Studies
System identification
Theory
Validation
Validity
Water resources
ISSN:0277-6693, 1099-131X
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The paper presents an identification procedure for a dynamic model of am hydrologic process. The process involves solute transport in streams subject to aquifer interaction and unsteady flows and the intended use of the model is prediction. Detailed assumptions and results are provided to illustrate the level of comprehensive analysis required to assess model adequacy. The assessment procedure easily generalizes to any dynamic model which is linear‐in‐the‐parameters. As a fundamental tool, instrumental variable algorithms can be adopted which have a number of attractive features. These algorithms make both model‐order identification and specification among alternatives a straightforward task. They are known to be consistent estimators in the presence of a wide class of errors. It is seen that they can be made stable and robust in the presence of data outliers. Instrumental variable algorithms can also be used which are asymptotically efficient and provide a covariance matrix of parameter estimates. The paper shows how they aid the quantification of predictive uncertainty and investigates the validity of the underlying assumptions. Further, it illustrates that, when instrumental variable algorithms are used in recursive mode, they can be used not only as an additional tool to access model inadequacy but also as an aid to model improvements.
Bibliografie:ArticleID:FOR3980100307
istex:56CAD57607513513704B7ED3B289FA5CCF0435D2
ark:/67375/WNG-CP9X681N-T
ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 14
ObjectType-Article-1
content type line 23
ISSN:0277-6693
1099-131X
DOI:10.1002/for.3980100307