Optimisation of competition indices using simulated annealing and artificial neural networks

Mark Lee Andrew Richards, Allan James Stuart McDonald, Matthew Aitkenhead

Research output: Contribution to journalArticle

21 Citations (Scopus)

Abstract

In this study we explored the effects of numerical optimisation on the ability of three competition indices (Lorimer, Heygi and Pukkala-Kolstrom) to predict the annual radial growth of Scots pine (Pinus sylvestris L.) in Glen Affric, Scotland. We present methods that combined neural networks and simulated annealing to determine the optimal specification for each index. Simulating annealing was used to discover the optimal specification of a series of competition indices, while a neural network was used to discover the best-fit model with a given competition index. Other inputs to the neural network model included tree DBH (diameter at breast height), environmental (exposure, soil wetness) and topographic (aspect, altitude) variables. The neural network prediction accuracy was fed back to the optimisation algorithm, which altered the specification of the competition index based on changes in the prediction accuracy. The neural network design itself was also subjected to simulated annealing to obtain an optimised predictive model. Validation of the neural network model was carried out using data not used to train the model. The optimisation approach decreased the error rate for the neural network models. It was also shown (a) that the Pukkala-Kolstrom competition index model gave a markedly better prediction of tree growth rates, and (b) that the angle of inclusion of neighbouring trees in calculating competition indices should be lower than is often suggested. (c) 2008 Elsevier B.V. All rights reserved.

Original languageEnglish
Pages (from-to)375-384
Number of pages10
JournalEcological Modelling
Volume214
Issue number2-4
Early online date15 Apr 2008
DOIs
Publication statusPublished - 24 Jun 2008

Keywords

  • Scots pine
  • competition index
  • neural network
  • simulated annealing
  • tree growth model
  • basal area growth
  • tree growth
  • models
  • variables
  • Britain
  • systems
  • height

Cite this

Optimisation of competition indices using simulated annealing and artificial neural networks. / Richards, Mark Lee Andrew; McDonald, Allan James Stuart; Aitkenhead, Matthew.

In: Ecological Modelling, Vol. 214, No. 2-4, 24.06.2008, p. 375-384.

Research output: Contribution to journalArticle

Richards, Mark Lee Andrew ; McDonald, Allan James Stuart ; Aitkenhead, Matthew. / Optimisation of competition indices using simulated annealing and artificial neural networks. In: Ecological Modelling. 2008 ; Vol. 214, No. 2-4. pp. 375-384.
@article{79f8817ccd904523972f75d96dba2072,
title = "Optimisation of competition indices using simulated annealing and artificial neural networks",
abstract = "In this study we explored the effects of numerical optimisation on the ability of three competition indices (Lorimer, Heygi and Pukkala-Kolstrom) to predict the annual radial growth of Scots pine (Pinus sylvestris L.) in Glen Affric, Scotland. We present methods that combined neural networks and simulated annealing to determine the optimal specification for each index. Simulating annealing was used to discover the optimal specification of a series of competition indices, while a neural network was used to discover the best-fit model with a given competition index. Other inputs to the neural network model included tree DBH (diameter at breast height), environmental (exposure, soil wetness) and topographic (aspect, altitude) variables. The neural network prediction accuracy was fed back to the optimisation algorithm, which altered the specification of the competition index based on changes in the prediction accuracy. The neural network design itself was also subjected to simulated annealing to obtain an optimised predictive model. Validation of the neural network model was carried out using data not used to train the model. The optimisation approach decreased the error rate for the neural network models. It was also shown (a) that the Pukkala-Kolstrom competition index model gave a markedly better prediction of tree growth rates, and (b) that the angle of inclusion of neighbouring trees in calculating competition indices should be lower than is often suggested. (c) 2008 Elsevier B.V. All rights reserved.",
keywords = "Scots pine, competition index, neural network, simulated annealing, tree growth model, basal area growth, tree growth, models, variables, Britain, systems, height",
author = "Richards, {Mark Lee Andrew} and McDonald, {Allan James Stuart} and Matthew Aitkenhead",
year = "2008",
month = "6",
day = "24",
doi = "10.1016/j.ecolmodel.2008.03.008",
language = "English",
volume = "214",
pages = "375--384",
journal = "Ecological Modelling",
issn = "0304-3800",
publisher = "Elsevier Science B. V.",
number = "2-4",

}

TY - JOUR

T1 - Optimisation of competition indices using simulated annealing and artificial neural networks

AU - Richards, Mark Lee Andrew

AU - McDonald, Allan James Stuart

AU - Aitkenhead, Matthew

PY - 2008/6/24

Y1 - 2008/6/24

N2 - In this study we explored the effects of numerical optimisation on the ability of three competition indices (Lorimer, Heygi and Pukkala-Kolstrom) to predict the annual radial growth of Scots pine (Pinus sylvestris L.) in Glen Affric, Scotland. We present methods that combined neural networks and simulated annealing to determine the optimal specification for each index. Simulating annealing was used to discover the optimal specification of a series of competition indices, while a neural network was used to discover the best-fit model with a given competition index. Other inputs to the neural network model included tree DBH (diameter at breast height), environmental (exposure, soil wetness) and topographic (aspect, altitude) variables. The neural network prediction accuracy was fed back to the optimisation algorithm, which altered the specification of the competition index based on changes in the prediction accuracy. The neural network design itself was also subjected to simulated annealing to obtain an optimised predictive model. Validation of the neural network model was carried out using data not used to train the model. The optimisation approach decreased the error rate for the neural network models. It was also shown (a) that the Pukkala-Kolstrom competition index model gave a markedly better prediction of tree growth rates, and (b) that the angle of inclusion of neighbouring trees in calculating competition indices should be lower than is often suggested. (c) 2008 Elsevier B.V. All rights reserved.

AB - In this study we explored the effects of numerical optimisation on the ability of three competition indices (Lorimer, Heygi and Pukkala-Kolstrom) to predict the annual radial growth of Scots pine (Pinus sylvestris L.) in Glen Affric, Scotland. We present methods that combined neural networks and simulated annealing to determine the optimal specification for each index. Simulating annealing was used to discover the optimal specification of a series of competition indices, while a neural network was used to discover the best-fit model with a given competition index. Other inputs to the neural network model included tree DBH (diameter at breast height), environmental (exposure, soil wetness) and topographic (aspect, altitude) variables. The neural network prediction accuracy was fed back to the optimisation algorithm, which altered the specification of the competition index based on changes in the prediction accuracy. The neural network design itself was also subjected to simulated annealing to obtain an optimised predictive model. Validation of the neural network model was carried out using data not used to train the model. The optimisation approach decreased the error rate for the neural network models. It was also shown (a) that the Pukkala-Kolstrom competition index model gave a markedly better prediction of tree growth rates, and (b) that the angle of inclusion of neighbouring trees in calculating competition indices should be lower than is often suggested. (c) 2008 Elsevier B.V. All rights reserved.

KW - Scots pine

KW - competition index

KW - neural network

KW - simulated annealing

KW - tree growth model

KW - basal area growth

KW - tree growth

KW - models

KW - variables

KW - Britain

KW - systems

KW - height

U2 - 10.1016/j.ecolmodel.2008.03.008

DO - 10.1016/j.ecolmodel.2008.03.008

M3 - Article

VL - 214

SP - 375

EP - 384

JO - Ecological Modelling

JF - Ecological Modelling

SN - 0304-3800

IS - 2-4

ER -