Neural network-based wind vector retrieval from satellite scatterometer data

Dan Cornford, Ian T. Nabney, Christopher M. Bishop

Research output: Contribution to journalArticle

Abstract

Obtaining wind vectors over the ocean is important for weather forecasting and ocean modelling. Several satellite systems used operationally by meteorological agencies utilise scatterometers to infer wind vectors over the oceans. In this paper we present the results of using novel neural network based techniques to estimate wind vectors from such data. The problem is partitioned into estimating wind speed and wind direction. Wind speed is modelled using a multi-layer perceptron (MLP) and a sum of squares error function. Wind direction is a periodic variable and a multi-valued function for a given set of inputs; a conventional MLP fails at this task, and so we model the full periodic probability density of direction conditioned on the satellite derived inputs using a Mixture Density Network (MDN) with periodic kernel functions. A committee of the resulting MDNs is shown to improve the results.
Original languageEnglish
Pages (from-to)206-217
Number of pages12
JournalNeural Computing and Applications
Volume8
Issue number3
DOIs
Publication statusPublished - Aug 1999

Fingerprint

Satellites
Neural networks
Multilayer neural networks
Weather forecasting

Keywords

  • conditional probability density estimation
  • mixture density network
  • multi-layer perceptron
  • periodic variables
  • scatterometer
  • wind vectors

Cite this

Cornford, Dan ; Nabney, Ian T. ; Bishop, Christopher M. / Neural network-based wind vector retrieval from satellite scatterometer data. In: Neural Computing and Applications. 1999 ; Vol. 8, No. 3. pp. 206-217.
@article{20ce96e06dd24db9b6e2a71c5f7fa52b,
title = "Neural network-based wind vector retrieval from satellite scatterometer data",
abstract = "Obtaining wind vectors over the ocean is important for weather forecasting and ocean modelling. Several satellite systems used operationally by meteorological agencies utilise scatterometers to infer wind vectors over the oceans. In this paper we present the results of using novel neural network based techniques to estimate wind vectors from such data. The problem is partitioned into estimating wind speed and wind direction. Wind speed is modelled using a multi-layer perceptron (MLP) and a sum of squares error function. Wind direction is a periodic variable and a multi-valued function for a given set of inputs; a conventional MLP fails at this task, and so we model the full periodic probability density of direction conditioned on the satellite derived inputs using a Mixture Density Network (MDN) with periodic kernel functions. A committee of the resulting MDNs is shown to improve the results.",
keywords = "conditional probability density estimation, mixture density network, multi-layer perceptron, periodic variables, scatterometer, wind vectors",
author = "Dan Cornford and Nabney, {Ian T.} and Bishop, {Christopher M.}",
year = "1999",
month = "8",
doi = "10.1007/s005210050023",
language = "English",
volume = "8",
pages = "206--217",
journal = "Neural Computing and Applications",
issn = "0941-0643",
publisher = "Springer",
number = "3",

}

Neural network-based wind vector retrieval from satellite scatterometer data. / Cornford, Dan; Nabney, Ian T.; Bishop, Christopher M.

In: Neural Computing and Applications, Vol. 8, No. 3, 08.1999, p. 206-217.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Neural network-based wind vector retrieval from satellite scatterometer data

AU - Cornford, Dan

AU - Nabney, Ian T.

AU - Bishop, Christopher M.

PY - 1999/8

Y1 - 1999/8

N2 - Obtaining wind vectors over the ocean is important for weather forecasting and ocean modelling. Several satellite systems used operationally by meteorological agencies utilise scatterometers to infer wind vectors over the oceans. In this paper we present the results of using novel neural network based techniques to estimate wind vectors from such data. The problem is partitioned into estimating wind speed and wind direction. Wind speed is modelled using a multi-layer perceptron (MLP) and a sum of squares error function. Wind direction is a periodic variable and a multi-valued function for a given set of inputs; a conventional MLP fails at this task, and so we model the full periodic probability density of direction conditioned on the satellite derived inputs using a Mixture Density Network (MDN) with periodic kernel functions. A committee of the resulting MDNs is shown to improve the results.

AB - Obtaining wind vectors over the ocean is important for weather forecasting and ocean modelling. Several satellite systems used operationally by meteorological agencies utilise scatterometers to infer wind vectors over the oceans. In this paper we present the results of using novel neural network based techniques to estimate wind vectors from such data. The problem is partitioned into estimating wind speed and wind direction. Wind speed is modelled using a multi-layer perceptron (MLP) and a sum of squares error function. Wind direction is a periodic variable and a multi-valued function for a given set of inputs; a conventional MLP fails at this task, and so we model the full periodic probability density of direction conditioned on the satellite derived inputs using a Mixture Density Network (MDN) with periodic kernel functions. A committee of the resulting MDNs is shown to improve the results.

KW - conditional probability density estimation

KW - mixture density network

KW - multi-layer perceptron

KW - periodic variables

KW - scatterometer

KW - wind vectors

UR - http://www.springerlink.com/content/03fqqkfnhpugv9dr/

UR - http://www.scopus.com/inward/record.url?scp=0033240927&partnerID=8YFLogxK

U2 - 10.1007/s005210050023

DO - 10.1007/s005210050023

M3 - Article

VL - 8

SP - 206

EP - 217

JO - Neural Computing and Applications

JF - Neural Computing and Applications

SN - 0941-0643

IS - 3

ER -