Quantifying simulator discrepancy in discrete-time dynamical simulators

Richard D. Wilkinson, Michail Vrettas, Dan Cornford, Jeremy E. Oakley

Research output: Contribution to journalArticle

Abstract

When making predictions with complex simulators it can be important to quantify the various sources of uncertainty. Errors in the structural specification of the simulator, for example due to missing processes or incorrect mathematical specification, can be a major source of uncertainty, but are often ignored. We introduce a methodology for inferring the discrepancy between the simulator and the system in discrete-time dynamical simulators. We assume a structural form for the discrepancy function, and show how to infer the maximum-likelihood parameter estimates using a particle filter embedded within a Monte Carlo expectation maximization (MCEM) algorithm. We illustrate the method on a conceptual rainfall-runoff simulator (logSPM) used to model the Abercrombie catchment in Australia. We assess the simulator and discrepancy model on the basis of their predictive performance using proper scoring rules. This article has supplementary material online.
Original languageEnglish
Pages (from-to)554-570
Number of pages17
JournalJournal of Agricultural, Biological, and Environmental Statistics
Volume16
Issue number4
DOIs
Publication statusPublished - 1 Dec 2011

Fingerprint

Uncertainty
Discrepancy
simulator
Discrete-time
Simulator
uncertainty
Simulators
Likelihood Functions
runoff
rain
prediction
methodology
Specification
Specifications
Monte Carlo Algorithm
Expectation-maximization Algorithm
Particle Filter
Rainfall
Runoff
Scoring

Bibliographical note

The definitive version is available at onlinelibrary.wiley.com
© Journal, International Biometric Society and Wiley-Blackwell

Keywords

  • model error
  • rainfall-runoff model
  • Monte Carlo EM algorithm

Cite this

Wilkinson, Richard D. ; Vrettas, Michail ; Cornford, Dan ; Oakley, Jeremy E. / Quantifying simulator discrepancy in discrete-time dynamical simulators. In: Journal of Agricultural, Biological, and Environmental Statistics. 2011 ; Vol. 16, No. 4. pp. 554-570.
@article{db16125fee7c4f08953a2b9234a2cfe5,
title = "Quantifying simulator discrepancy in discrete-time dynamical simulators",
abstract = "When making predictions with complex simulators it can be important to quantify the various sources of uncertainty. Errors in the structural specification of the simulator, for example due to missing processes or incorrect mathematical specification, can be a major source of uncertainty, but are often ignored. We introduce a methodology for inferring the discrepancy between the simulator and the system in discrete-time dynamical simulators. We assume a structural form for the discrepancy function, and show how to infer the maximum-likelihood parameter estimates using a particle filter embedded within a Monte Carlo expectation maximization (MCEM) algorithm. We illustrate the method on a conceptual rainfall-runoff simulator (logSPM) used to model the Abercrombie catchment in Australia. We assess the simulator and discrepancy model on the basis of their predictive performance using proper scoring rules. This article has supplementary material online.",
keywords = "model error, rainfall-runoff model, Monte Carlo EM algorithm",
author = "Wilkinson, {Richard D.} and Michail Vrettas and Dan Cornford and Oakley, {Jeremy E.}",
note = "The definitive version is available at onlinelibrary.wiley.com {\circledC} Journal, International Biometric Society and Wiley-Blackwell",
year = "2011",
month = "12",
day = "1",
doi = "10.1007/s13253-011-0077-3",
language = "English",
volume = "16",
pages = "554--570",
journal = "Journal of Agricultural, Biological, and Environmental Statistics",
issn = "1085-7117",
publisher = "Springer",
number = "4",

}

Quantifying simulator discrepancy in discrete-time dynamical simulators. / Wilkinson, Richard D.; Vrettas, Michail; Cornford, Dan; Oakley, Jeremy E.

In: Journal of Agricultural, Biological, and Environmental Statistics, Vol. 16, No. 4, 01.12.2011, p. 554-570.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Quantifying simulator discrepancy in discrete-time dynamical simulators

AU - Wilkinson, Richard D.

AU - Vrettas, Michail

AU - Cornford, Dan

AU - Oakley, Jeremy E.

N1 - The definitive version is available at onlinelibrary.wiley.com © Journal, International Biometric Society and Wiley-Blackwell

PY - 2011/12/1

Y1 - 2011/12/1

N2 - When making predictions with complex simulators it can be important to quantify the various sources of uncertainty. Errors in the structural specification of the simulator, for example due to missing processes or incorrect mathematical specification, can be a major source of uncertainty, but are often ignored. We introduce a methodology for inferring the discrepancy between the simulator and the system in discrete-time dynamical simulators. We assume a structural form for the discrepancy function, and show how to infer the maximum-likelihood parameter estimates using a particle filter embedded within a Monte Carlo expectation maximization (MCEM) algorithm. We illustrate the method on a conceptual rainfall-runoff simulator (logSPM) used to model the Abercrombie catchment in Australia. We assess the simulator and discrepancy model on the basis of their predictive performance using proper scoring rules. This article has supplementary material online.

AB - When making predictions with complex simulators it can be important to quantify the various sources of uncertainty. Errors in the structural specification of the simulator, for example due to missing processes or incorrect mathematical specification, can be a major source of uncertainty, but are often ignored. We introduce a methodology for inferring the discrepancy between the simulator and the system in discrete-time dynamical simulators. We assume a structural form for the discrepancy function, and show how to infer the maximum-likelihood parameter estimates using a particle filter embedded within a Monte Carlo expectation maximization (MCEM) algorithm. We illustrate the method on a conceptual rainfall-runoff simulator (logSPM) used to model the Abercrombie catchment in Australia. We assess the simulator and discrepancy model on the basis of their predictive performance using proper scoring rules. This article has supplementary material online.

KW - model error

KW - rainfall-runoff model

KW - Monte Carlo EM algorithm

UR - http://www.scopus.com/inward/record.url?scp=83555177262&partnerID=8YFLogxK

U2 - 10.1007/s13253-011-0077-3

DO - 10.1007/s13253-011-0077-3

M3 - Article

AN - SCOPUS:83555177262

VL - 16

SP - 554

EP - 570

JO - Journal of Agricultural, Biological, and Environmental Statistics

JF - Journal of Agricultural, Biological, and Environmental Statistics

SN - 1085-7117

IS - 4

ER -