Modeling and prediction of nonlinear environmental system using Bayesian methods

Majdi Mansouri, Benjamin Dumont, Marie France Destain

Research output: Contribution to journalArticle

23 Citations (Scopus)

Abstract

An environmental dynamic system is usually modeled as a nonlinear system described by a set of nonlinear ODEs. A central challenge in computational modeling of environmental systems is the determination of the model parameters. In these cases, estimating these variables or parameters from other easily obtained measurements can be extremely useful. This work addresses the problem of monitoring and modeling a leaf area index and soil moisture model (LSM) using state estimation. The performances of various conventional and state-of-the-art state estimation techniques are compared when they are utilized to achieve this objective. These techniques include the extended Kalman filter (EKF), particle filter (PF), and the more recently developed technique variational filter (VF). Specifically, two comparative studies are performed. In the first comparative study, the state variables (the leaf-area index LAI , the volumetric water content of the soil layer 1, HUR1 and the volumetric water content of the soil layer 2, HUR2) are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error (RMSE) with respect to the noise-free data. In the second comparative study, the state variables as well as the model parameters are simultaneously estimated. In this case, in addition to comparing the performances of the various state estimation techniques, the effect of number of estimated model parameters on the accuracy and convergence of these techniques are also assessed. The results of both comparative studies show that the PF provides a higher accuracy than the EKF, which is due to the limited ability of the EKF to handle highly nonlinear processes. The results also show that the VF provides a significant improvement over the PF because, unlike the PF which depends on the choice of sampling distribution used to estimate the posterior distribution, the VF yields an optimum choice of the sampling distribution, which also accounts for the observed data. The results of the second comparative study show that, for all techniques, estimating more model parameters affects the estimation accuracy as well as the convergence of the estimated states and parameters. However, the VF can still provide both convergence as well as accuracy related advantages over other estimation methods.

Original languageEnglish
Pages (from-to)16-31
Number of pages16
JournalComputers and Electronics in Agriculture
Volume92
DOIs
Publication statusPublished - 2013
Externally publishedYes

Fingerprint

Bayesian theory
filter
Extended Kalman filters
State estimation
prediction
comparative study
modeling
Kalman filter
Water content
leaf area index
Sampling
Variational techniques
Soils
methodology
Soil moisture
soil water content
Mean square error
water content
Nonlinear systems
Dynamical systems

Keywords

  • Extended Kalman filter
  • Leaf area index and soil moisture model
  • Nonlinear environmental system
  • Particle filter
  • State and parameter estimation
  • Variational filter

ASJC Scopus subject areas

  • Forestry
  • Animal Science and Zoology
  • Agronomy and Crop Science
  • Computer Science Applications
  • Horticulture

Cite this

Modeling and prediction of nonlinear environmental system using Bayesian methods. / Mansouri, Majdi; Dumont, Benjamin; Destain, Marie France.

In: Computers and Electronics in Agriculture, Vol. 92, 2013, p. 16-31.

Research output: Contribution to journalArticle

@article{c8a147e3e6654e0f9fa8e13c66555c35,
title = "Modeling and prediction of nonlinear environmental system using Bayesian methods",
abstract = "An environmental dynamic system is usually modeled as a nonlinear system described by a set of nonlinear ODEs. A central challenge in computational modeling of environmental systems is the determination of the model parameters. In these cases, estimating these variables or parameters from other easily obtained measurements can be extremely useful. This work addresses the problem of monitoring and modeling a leaf area index and soil moisture model (LSM) using state estimation. The performances of various conventional and state-of-the-art state estimation techniques are compared when they are utilized to achieve this objective. These techniques include the extended Kalman filter (EKF), particle filter (PF), and the more recently developed technique variational filter (VF). Specifically, two comparative studies are performed. In the first comparative study, the state variables (the leaf-area index LAI , the volumetric water content of the soil layer 1, HUR1 and the volumetric water content of the soil layer 2, HUR2) are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error (RMSE) with respect to the noise-free data. In the second comparative study, the state variables as well as the model parameters are simultaneously estimated. In this case, in addition to comparing the performances of the various state estimation techniques, the effect of number of estimated model parameters on the accuracy and convergence of these techniques are also assessed. The results of both comparative studies show that the PF provides a higher accuracy than the EKF, which is due to the limited ability of the EKF to handle highly nonlinear processes. The results also show that the VF provides a significant improvement over the PF because, unlike the PF which depends on the choice of sampling distribution used to estimate the posterior distribution, the VF yields an optimum choice of the sampling distribution, which also accounts for the observed data. The results of the second comparative study show that, for all techniques, estimating more model parameters affects the estimation accuracy as well as the convergence of the estimated states and parameters. However, the VF can still provide both convergence as well as accuracy related advantages over other estimation methods.",
keywords = "Extended Kalman filter, Leaf area index and soil moisture model, Nonlinear environmental system, Particle filter, State and parameter estimation, Variational filter",
author = "Majdi Mansouri and Benjamin Dumont and Destain, {Marie France}",
year = "2013",
doi = "10.1016/j.compag.2012.12.013",
language = "English",
volume = "92",
pages = "16--31",
journal = "Computers and Electronics in Agriculture",
issn = "0168-1699",
publisher = "Elsevier",

}

TY - JOUR

T1 - Modeling and prediction of nonlinear environmental system using Bayesian methods

AU - Mansouri, Majdi

AU - Dumont, Benjamin

AU - Destain, Marie France

PY - 2013

Y1 - 2013

N2 - An environmental dynamic system is usually modeled as a nonlinear system described by a set of nonlinear ODEs. A central challenge in computational modeling of environmental systems is the determination of the model parameters. In these cases, estimating these variables or parameters from other easily obtained measurements can be extremely useful. This work addresses the problem of monitoring and modeling a leaf area index and soil moisture model (LSM) using state estimation. The performances of various conventional and state-of-the-art state estimation techniques are compared when they are utilized to achieve this objective. These techniques include the extended Kalman filter (EKF), particle filter (PF), and the more recently developed technique variational filter (VF). Specifically, two comparative studies are performed. In the first comparative study, the state variables (the leaf-area index LAI , the volumetric water content of the soil layer 1, HUR1 and the volumetric water content of the soil layer 2, HUR2) are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error (RMSE) with respect to the noise-free data. In the second comparative study, the state variables as well as the model parameters are simultaneously estimated. In this case, in addition to comparing the performances of the various state estimation techniques, the effect of number of estimated model parameters on the accuracy and convergence of these techniques are also assessed. The results of both comparative studies show that the PF provides a higher accuracy than the EKF, which is due to the limited ability of the EKF to handle highly nonlinear processes. The results also show that the VF provides a significant improvement over the PF because, unlike the PF which depends on the choice of sampling distribution used to estimate the posterior distribution, the VF yields an optimum choice of the sampling distribution, which also accounts for the observed data. The results of the second comparative study show that, for all techniques, estimating more model parameters affects the estimation accuracy as well as the convergence of the estimated states and parameters. However, the VF can still provide both convergence as well as accuracy related advantages over other estimation methods.

AB - An environmental dynamic system is usually modeled as a nonlinear system described by a set of nonlinear ODEs. A central challenge in computational modeling of environmental systems is the determination of the model parameters. In these cases, estimating these variables or parameters from other easily obtained measurements can be extremely useful. This work addresses the problem of monitoring and modeling a leaf area index and soil moisture model (LSM) using state estimation. The performances of various conventional and state-of-the-art state estimation techniques are compared when they are utilized to achieve this objective. These techniques include the extended Kalman filter (EKF), particle filter (PF), and the more recently developed technique variational filter (VF). Specifically, two comparative studies are performed. In the first comparative study, the state variables (the leaf-area index LAI , the volumetric water content of the soil layer 1, HUR1 and the volumetric water content of the soil layer 2, HUR2) are estimated from noisy measurements of these variables, and the various estimation techniques are compared by computing the estimation root mean square error (RMSE) with respect to the noise-free data. In the second comparative study, the state variables as well as the model parameters are simultaneously estimated. In this case, in addition to comparing the performances of the various state estimation techniques, the effect of number of estimated model parameters on the accuracy and convergence of these techniques are also assessed. The results of both comparative studies show that the PF provides a higher accuracy than the EKF, which is due to the limited ability of the EKF to handle highly nonlinear processes. The results also show that the VF provides a significant improvement over the PF because, unlike the PF which depends on the choice of sampling distribution used to estimate the posterior distribution, the VF yields an optimum choice of the sampling distribution, which also accounts for the observed data. The results of the second comparative study show that, for all techniques, estimating more model parameters affects the estimation accuracy as well as the convergence of the estimated states and parameters. However, the VF can still provide both convergence as well as accuracy related advantages over other estimation methods.

KW - Extended Kalman filter

KW - Leaf area index and soil moisture model

KW - Nonlinear environmental system

KW - Particle filter

KW - State and parameter estimation

KW - Variational filter

UR - http://www.scopus.com/inward/record.url?scp=84873248164&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84873248164&partnerID=8YFLogxK

U2 - 10.1016/j.compag.2012.12.013

DO - 10.1016/j.compag.2012.12.013

M3 - Article

VL - 92

SP - 16

EP - 31

JO - Computers and Electronics in Agriculture

JF - Computers and Electronics in Agriculture

SN - 0168-1699

ER -