Parameter estimation in stochastic rainfall-runoff models
A parameter estimation method for stochastic rainfall-runoff models is presented. The model considered in the paper is a conceptual stochastic model, formulated in continuous-discrete state space form. The model is small and a fully automatic optimization is, therefore, possible for estimating all t...
Main Authors: | , , |
---|---|
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
2006
|
Subjects: | |
Online Access: | https://orbit.dtu.dk/en/publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c |
id |
ftdtupubl:oai:pure.atira.dk:publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c |
---|---|
record_format |
openpolar |
spelling |
ftdtupubl:oai:pure.atira.dk:publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c 2024-05-19T07:42:50+00:00 Parameter estimation in stochastic rainfall-runoff models Jonsdottir, Harpa Madsen, Henrik Palsson, Olafur Petur 2006 https://orbit.dtu.dk/en/publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c eng eng https://orbit.dtu.dk/en/publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c info:eu-repo/semantics/restrictedAccess Jonsdottir , H , Madsen , H & Palsson , O P 2006 , ' Parameter estimation in stochastic rainfall-runoff models ' , Journal of Hydrology , vol. 326 , no. 1-4 , pp. 379-393 . article 2006 ftdtupubl 2024-04-24T00:24:29Z A parameter estimation method for stochastic rainfall-runoff models is presented. The model considered in the paper is a conceptual stochastic model, formulated in continuous-discrete state space form. The model is small and a fully automatic optimization is, therefore, possible for estimating all the parameters, including the noise terms. The parameter estimation method is a maximum likelihood method (ML) where the likelihood function is evaluated using a Kalman filter technique. The ML method estimates the parameters in a prediction error settings, i.e. the sum of squared prediction error is minimized. For a comparison the parameters are also estimated by an output error method, where the sum of squared simulation error is minimized. The former methodology is optimal for short-term prediction whereas the latter is optimal for simulations. Hence, depending on the purpose it is possible to select whether the parameter values are optimal for simulation or prediction. The data originates from Iceland and the model is designed for Icelandic conditions, including a snow routine for mountainous areas. The model demands only two input data series, precipitation and temperature and one output data series, the discharge. In spite of being based on relatively limited input information, the model performs well and the parameter estimation method is promising for future model development. Article in Journal/Newspaper Iceland Technical University of Denmark: DTU Orbit |
institution |
Open Polar |
collection |
Technical University of Denmark: DTU Orbit |
op_collection_id |
ftdtupubl |
language |
English |
description |
A parameter estimation method for stochastic rainfall-runoff models is presented. The model considered in the paper is a conceptual stochastic model, formulated in continuous-discrete state space form. The model is small and a fully automatic optimization is, therefore, possible for estimating all the parameters, including the noise terms. The parameter estimation method is a maximum likelihood method (ML) where the likelihood function is evaluated using a Kalman filter technique. The ML method estimates the parameters in a prediction error settings, i.e. the sum of squared prediction error is minimized. For a comparison the parameters are also estimated by an output error method, where the sum of squared simulation error is minimized. The former methodology is optimal for short-term prediction whereas the latter is optimal for simulations. Hence, depending on the purpose it is possible to select whether the parameter values are optimal for simulation or prediction. The data originates from Iceland and the model is designed for Icelandic conditions, including a snow routine for mountainous areas. The model demands only two input data series, precipitation and temperature and one output data series, the discharge. In spite of being based on relatively limited input information, the model performs well and the parameter estimation method is promising for future model development. |
format |
Article in Journal/Newspaper |
author |
Jonsdottir, Harpa Madsen, Henrik Palsson, Olafur Petur |
spellingShingle |
Jonsdottir, Harpa Madsen, Henrik Palsson, Olafur Petur Parameter estimation in stochastic rainfall-runoff models |
author_facet |
Jonsdottir, Harpa Madsen, Henrik Palsson, Olafur Petur |
author_sort |
Jonsdottir, Harpa |
title |
Parameter estimation in stochastic rainfall-runoff models |
title_short |
Parameter estimation in stochastic rainfall-runoff models |
title_full |
Parameter estimation in stochastic rainfall-runoff models |
title_fullStr |
Parameter estimation in stochastic rainfall-runoff models |
title_full_unstemmed |
Parameter estimation in stochastic rainfall-runoff models |
title_sort |
parameter estimation in stochastic rainfall-runoff models |
publishDate |
2006 |
url |
https://orbit.dtu.dk/en/publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c |
genre |
Iceland |
genre_facet |
Iceland |
op_source |
Jonsdottir , H , Madsen , H & Palsson , O P 2006 , ' Parameter estimation in stochastic rainfall-runoff models ' , Journal of Hydrology , vol. 326 , no. 1-4 , pp. 379-393 . |
op_relation |
https://orbit.dtu.dk/en/publications/d8f4c709-22cc-445a-bcfd-909f11cb0b0c |
op_rights |
info:eu-repo/semantics/restrictedAccess |
_version_ |
1799482540438847488 |