Emulation of environmental models using polynomial chaos expansion

Elias C. Massoud

doi:10.1016/j.envsoft.2018.10.008

Emulation of environmental models using polynomial chaos expansion

Elias C. Massoud

Research output: Contribution to journal › Article › peer-review

8 Scopus citations

Abstract

This paper investigates the applicability of model emulation to speed up simulation time of CPU intensive environmental models. Polynomial chaos expansion (PCE) emulators are constructed for three case studies of increasing complexity. The level of emulator training and the order of polynomial necessary to sufficiently build accurate emulators for each model are investigated. Although the PCE emulators shown here do not approximate well the outputs of parameter rich models (80 + parameters), results demonstrate that the emulators mimic closely outputs of relatively simple, low dimensional, simulation models (15 parameters or less). Furthermore, the PCE emulators are tested with applications such as Global Sensitivity Analysis (GSA). Results illustrate the advantages and drawbacks of using classical PCE emulators for treating computational limitation of complex environmental models.

Original language	English
Pages (from-to)	421-431
Number of pages	11
Journal	Environmental Modelling and Software
Volume	111
DOIs	https://doi.org/10.1016/j.envsoft.2018.10.008
State	Published - Jan 2019
Externally published	Yes

Funding

The author gratefully acknowledges support from the SCGSR fellowship of the Department of Energy and the UC-Lab Fees Research Program Award 237285 . This research was performed at the University of California, Irvine, and at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics Space Administration. The author thanks Dr J. Vrugt and Dr C. Xu for assistance with the models used in the paper, and Dr E. Laloy ( [email protected] ) for assistance with the emulation software. Special thanks are given to Los Alamos National Laboratory for allowing the CLM simulations to be made on the Conejo Supercomputing facility. The source code of the PCE method is written in MATLAB and available upon request. The author gratefully acknowledges support from the SCGSR fellowship of the Department of Energy and the UC-Lab Fees Research Program Award 237285. This research was performed at the University of California, Irvine, and at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics Space Administration. The author thanks Dr J. Vrugt and Dr C. Xu for assistance with the models used in the paper, and Dr E. Laloy ([email protected]) for assistance with the emulation software. Special thanks are given to Los Alamos National Laboratory for allowing the CLM simulations to be made on the Conejo Supercomputing facility. The source code of the PCE method is written in MATLAB and available upon request.

Funders	Funder number
U.S. Department of Energy	237285
National Aeronautics and Space Administration
University of California, Irvine

Keywords

Model emulation
Polynomial chaos expansion
Sensitivity analysis
Surrogate modeling

Access to Document

10.1016/j.envsoft.2018.10.008

Cite this

@article{c499d19c21ec4514a3ba7668743060e2,

title = "Emulation of environmental models using polynomial chaos expansion",

abstract = "This paper investigates the applicability of model emulation to speed up simulation time of CPU intensive environmental models. Polynomial chaos expansion (PCE) emulators are constructed for three case studies of increasing complexity. The level of emulator training and the order of polynomial necessary to sufficiently build accurate emulators for each model are investigated. Although the PCE emulators shown here do not approximate well the outputs of parameter rich models (80 + parameters), results demonstrate that the emulators mimic closely outputs of relatively simple, low dimensional, simulation models (15 parameters or less). Furthermore, the PCE emulators are tested with applications such as Global Sensitivity Analysis (GSA). Results illustrate the advantages and drawbacks of using classical PCE emulators for treating computational limitation of complex environmental models.",

keywords = "Model emulation, Polynomial chaos expansion, Sensitivity analysis, Surrogate modeling",

author = "Massoud, {Elias C.}",

note = "Publisher Copyright: {\textcopyright} 2018",

year = "2019",

month = jan,

doi = "10.1016/j.envsoft.2018.10.008",

language = "English",

volume = "111",

pages = "421--431",

journal = "Environmental Modelling and Software",

issn = "1364-8152",

publisher = "Elsevier",

}

TY - JOUR

T1 - Emulation of environmental models using polynomial chaos expansion

AU - Massoud, Elias C.

PY - 2019/1

Y1 - 2019/1

N2 - This paper investigates the applicability of model emulation to speed up simulation time of CPU intensive environmental models. Polynomial chaos expansion (PCE) emulators are constructed for three case studies of increasing complexity. The level of emulator training and the order of polynomial necessary to sufficiently build accurate emulators for each model are investigated. Although the PCE emulators shown here do not approximate well the outputs of parameter rich models (80 + parameters), results demonstrate that the emulators mimic closely outputs of relatively simple, low dimensional, simulation models (15 parameters or less). Furthermore, the PCE emulators are tested with applications such as Global Sensitivity Analysis (GSA). Results illustrate the advantages and drawbacks of using classical PCE emulators for treating computational limitation of complex environmental models.

AB - This paper investigates the applicability of model emulation to speed up simulation time of CPU intensive environmental models. Polynomial chaos expansion (PCE) emulators are constructed for three case studies of increasing complexity. The level of emulator training and the order of polynomial necessary to sufficiently build accurate emulators for each model are investigated. Although the PCE emulators shown here do not approximate well the outputs of parameter rich models (80 + parameters), results demonstrate that the emulators mimic closely outputs of relatively simple, low dimensional, simulation models (15 parameters or less). Furthermore, the PCE emulators are tested with applications such as Global Sensitivity Analysis (GSA). Results illustrate the advantages and drawbacks of using classical PCE emulators for treating computational limitation of complex environmental models.

KW - Model emulation

KW - Polynomial chaos expansion

KW - Sensitivity analysis

KW - Surrogate modeling

UR - http://www.scopus.com/inward/record.url?scp=85055865267&partnerID=8YFLogxK

U2 - 10.1016/j.envsoft.2018.10.008

DO - 10.1016/j.envsoft.2018.10.008

M3 - Article

AN - SCOPUS:85055865267

SN - 1364-8152

VL - 111

SP - 421

EP - 431

JO - Environmental Modelling and Software

JF - Environmental Modelling and Software

ER -

Emulation of environmental models using polynomial chaos expansion

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this