Bayesian inference for wind field retrieval

Dan Cornford, Ian T. Nabney, Christopher K. I. Williams

Research output: Working paperTechnical report

Abstract

In many problems in spatial statistics it is necessary to infer a global problem solution by combining local models. A principled approach to this problem is to develop a global probabilistic model for the relationships between local variables and to use this as the prior in a Bayesian inference procedure. We show how a Gaussian process with hyper-parameters estimated from Numerical Weather Prediction Models yields meteorologically convincing wind fields. We use neural networks to make local estimates of wind vector probabilities. The resulting inference problem cannot be solved analytically, but Markov Chain Monte Carlo methods allow us to retrieve accurate wind fields.
Original languageEnglish
Place of PublicationBirmingham
PublisherAston University
Number of pages8
ISBN (Print)NCRG/98/023
Publication statusPublished - 22 Oct 1998

Fingerprint

wind field
Markov chain
weather
prediction
method
statistics
parameter
global model

Keywords

  • Bayesian inference
  • surface winds
  • spatial priors
  • Gaussian processes

Cite this

Cornford, D., Nabney, I. T., & Williams, C. K. I. (1998). Bayesian inference for wind field retrieval. Birmingham: Aston University.
Cornford, Dan ; Nabney, Ian T. ; Williams, Christopher K. I. / Bayesian inference for wind field retrieval. Birmingham : Aston University, 1998.
@techreport{c9f3e0d51f674a5bb8b2002a1700a663,
title = "Bayesian inference for wind field retrieval",
abstract = "In many problems in spatial statistics it is necessary to infer a global problem solution by combining local models. A principled approach to this problem is to develop a global probabilistic model for the relationships between local variables and to use this as the prior in a Bayesian inference procedure. We show how a Gaussian process with hyper-parameters estimated from Numerical Weather Prediction Models yields meteorologically convincing wind fields. We use neural networks to make local estimates of wind vector probabilities. The resulting inference problem cannot be solved analytically, but Markov Chain Monte Carlo methods allow us to retrieve accurate wind fields.",
keywords = "Bayesian inference, surface winds, spatial priors, Gaussian processes",
author = "Dan Cornford and Nabney, {Ian T.} and Williams, {Christopher K. I.}",
year = "1998",
month = "10",
day = "22",
language = "English",
isbn = "NCRG/98/023",
publisher = "Aston University",
type = "WorkingPaper",
institution = "Aston University",

}

Cornford, D, Nabney, IT & Williams, CKI 1998 'Bayesian inference for wind field retrieval' Aston University, Birmingham.

Bayesian inference for wind field retrieval. / Cornford, Dan; Nabney, Ian T.; Williams, Christopher K. I.

Birmingham : Aston University, 1998.

Research output: Working paperTechnical report

TY - UNPB

T1 - Bayesian inference for wind field retrieval

AU - Cornford, Dan

AU - Nabney, Ian T.

AU - Williams, Christopher K. I.

PY - 1998/10/22

Y1 - 1998/10/22

N2 - In many problems in spatial statistics it is necessary to infer a global problem solution by combining local models. A principled approach to this problem is to develop a global probabilistic model for the relationships between local variables and to use this as the prior in a Bayesian inference procedure. We show how a Gaussian process with hyper-parameters estimated from Numerical Weather Prediction Models yields meteorologically convincing wind fields. We use neural networks to make local estimates of wind vector probabilities. The resulting inference problem cannot be solved analytically, but Markov Chain Monte Carlo methods allow us to retrieve accurate wind fields.

AB - In many problems in spatial statistics it is necessary to infer a global problem solution by combining local models. A principled approach to this problem is to develop a global probabilistic model for the relationships between local variables and to use this as the prior in a Bayesian inference procedure. We show how a Gaussian process with hyper-parameters estimated from Numerical Weather Prediction Models yields meteorologically convincing wind fields. We use neural networks to make local estimates of wind vector probabilities. The resulting inference problem cannot be solved analytically, but Markov Chain Monte Carlo methods allow us to retrieve accurate wind fields.

KW - Bayesian inference

KW - surface winds

KW - spatial priors

KW - Gaussian processes

M3 - Technical report

SN - NCRG/98/023

BT - Bayesian inference for wind field retrieval

PB - Aston University

CY - Birmingham

ER -

Cornford D, Nabney IT, Williams CKI. Bayesian inference for wind field retrieval. Birmingham: Aston University. 1998 Oct 22.