A projected process kriging algorithm for sensor networks with heterogeneous error characteristics

Benjamin R. Ingram, Dan Cornford, Lehel Csató

Research output: Contribution to conferencePaper

Abstract

Large monitoring networks are becoming increasingly common and can generate large datasets from thousands to millions of observations in size, often with high temporal resolution. Processing large datasets using traditional geostatistical methods is prohibitively slow and in real world applications different types of sensor can be found across a monitoring network. Heterogeneities in the error characteristics of different sensors, both in terms of distribution and magnitude, presents problems for generating coherent maps. An assumption in traditional geostatistics is that observations are made directly of the underlying process being studied and that the observations are contaminated with Gaussian errors. Under this assumption, sub–optimal predictions will be obtained if the error characteristics of the sensor are effectively non–Gaussian. One method, model based geostatistics, assumes that a Gaussian process prior is imposed over the (latent) process being studied and that the sensor model forms part of the likelihood term. One problem with this type of approach is that the corresponding posterior distribution will be non–Gaussian and computationally demanding as Monte Carlo methods have to be used. An extension of a sequential, approximate Bayesian inference method enables observations with arbitrary likelihoods to be treated, in a projected process kriging framework which is less computationally intensive. The approach is illustrated using a simulated dataset with a range of sensor models and error characteristics.
Original languageEnglish
Publication statusUnpublished - 2008
Event8th International Geostatistics Congress - Santiago , Chile
Duration: 1 Dec 20085 Dec 2008

Conference

Conference8th International Geostatistics Congress
Abbreviated titleGeostats 2008
CountryChile
CitySantiago
Period1/12/085/12/08

Fingerprint

Sensor networks
Sensors
Monitoring
Monte Carlo methods
Processing

Bibliographical note

Proceedings: Ortiz, Julián M.; Emery, Xavier (eds). GEOSTATS 2008 – VIII International Geostatistics Congress, 1-5 December, Santiago, Chile. Mining Engineering Department, University of Chile. Two volumes, 1188 pages.

Keywords

  • large monitoring networks
  • model based geostatistics
  • Gaussian process prior
  • Monte Carlo methods
  • approximate Bayesian inference method

Cite this

Ingram, B. R., Cornford, D., & Csató, L. (2008). A projected process kriging algorithm for sensor networks with heterogeneous error characteristics. Paper presented at 8th International Geostatistics Congress, Santiago , Chile.
Ingram, Benjamin R. ; Cornford, Dan ; Csató, Lehel. / A projected process kriging algorithm for sensor networks with heterogeneous error characteristics. Paper presented at 8th International Geostatistics Congress, Santiago , Chile.
@conference{3a32acc1c87d495db1ed01df73e3a84b,
title = "A projected process kriging algorithm for sensor networks with heterogeneous error characteristics",
abstract = "Large monitoring networks are becoming increasingly common and can generate large datasets from thousands to millions of observations in size, often with high temporal resolution. Processing large datasets using traditional geostatistical methods is prohibitively slow and in real world applications different types of sensor can be found across a monitoring network. Heterogeneities in the error characteristics of different sensors, both in terms of distribution and magnitude, presents problems for generating coherent maps. An assumption in traditional geostatistics is that observations are made directly of the underlying process being studied and that the observations are contaminated with Gaussian errors. Under this assumption, sub–optimal predictions will be obtained if the error characteristics of the sensor are effectively non–Gaussian. One method, model based geostatistics, assumes that a Gaussian process prior is imposed over the (latent) process being studied and that the sensor model forms part of the likelihood term. One problem with this type of approach is that the corresponding posterior distribution will be non–Gaussian and computationally demanding as Monte Carlo methods have to be used. An extension of a sequential, approximate Bayesian inference method enables observations with arbitrary likelihoods to be treated, in a projected process kriging framework which is less computationally intensive. The approach is illustrated using a simulated dataset with a range of sensor models and error characteristics.",
keywords = "large monitoring networks, model based geostatistics, Gaussian process prior, Monte Carlo methods, approximate Bayesian inference method",
author = "Ingram, {Benjamin R.} and Dan Cornford and Lehel Csat{\'o}",
note = "Proceedings: Ortiz, Juli{\'a}n M.; Emery, Xavier (eds). GEOSTATS 2008 – VIII International Geostatistics Congress, 1-5 December, Santiago, Chile. Mining Engineering Department, University of Chile. Two volumes, 1188 pages.; 8th International Geostatistics Congress, Geostats 2008 ; Conference date: 01-12-2008 Through 05-12-2008",
year = "2008",
language = "English",

}

Ingram, BR, Cornford, D & Csató, L 2008, 'A projected process kriging algorithm for sensor networks with heterogeneous error characteristics' Paper presented at 8th International Geostatistics Congress, Santiago , Chile, 1/12/08 - 5/12/08, .

A projected process kriging algorithm for sensor networks with heterogeneous error characteristics. / Ingram, Benjamin R.; Cornford, Dan; Csató, Lehel.

2008. Paper presented at 8th International Geostatistics Congress, Santiago , Chile.

Research output: Contribution to conferencePaper

TY - CONF

T1 - A projected process kriging algorithm for sensor networks with heterogeneous error characteristics

AU - Ingram, Benjamin R.

AU - Cornford, Dan

AU - Csató, Lehel

N1 - Proceedings: Ortiz, Julián M.; Emery, Xavier (eds). GEOSTATS 2008 – VIII International Geostatistics Congress, 1-5 December, Santiago, Chile. Mining Engineering Department, University of Chile. Two volumes, 1188 pages.

PY - 2008

Y1 - 2008

N2 - Large monitoring networks are becoming increasingly common and can generate large datasets from thousands to millions of observations in size, often with high temporal resolution. Processing large datasets using traditional geostatistical methods is prohibitively slow and in real world applications different types of sensor can be found across a monitoring network. Heterogeneities in the error characteristics of different sensors, both in terms of distribution and magnitude, presents problems for generating coherent maps. An assumption in traditional geostatistics is that observations are made directly of the underlying process being studied and that the observations are contaminated with Gaussian errors. Under this assumption, sub–optimal predictions will be obtained if the error characteristics of the sensor are effectively non–Gaussian. One method, model based geostatistics, assumes that a Gaussian process prior is imposed over the (latent) process being studied and that the sensor model forms part of the likelihood term. One problem with this type of approach is that the corresponding posterior distribution will be non–Gaussian and computationally demanding as Monte Carlo methods have to be used. An extension of a sequential, approximate Bayesian inference method enables observations with arbitrary likelihoods to be treated, in a projected process kriging framework which is less computationally intensive. The approach is illustrated using a simulated dataset with a range of sensor models and error characteristics.

AB - Large monitoring networks are becoming increasingly common and can generate large datasets from thousands to millions of observations in size, often with high temporal resolution. Processing large datasets using traditional geostatistical methods is prohibitively slow and in real world applications different types of sensor can be found across a monitoring network. Heterogeneities in the error characteristics of different sensors, both in terms of distribution and magnitude, presents problems for generating coherent maps. An assumption in traditional geostatistics is that observations are made directly of the underlying process being studied and that the observations are contaminated with Gaussian errors. Under this assumption, sub–optimal predictions will be obtained if the error characteristics of the sensor are effectively non–Gaussian. One method, model based geostatistics, assumes that a Gaussian process prior is imposed over the (latent) process being studied and that the sensor model forms part of the likelihood term. One problem with this type of approach is that the corresponding posterior distribution will be non–Gaussian and computationally demanding as Monte Carlo methods have to be used. An extension of a sequential, approximate Bayesian inference method enables observations with arbitrary likelihoods to be treated, in a projected process kriging framework which is less computationally intensive. The approach is illustrated using a simulated dataset with a range of sensor models and error characteristics.

KW - large monitoring networks

KW - model based geostatistics

KW - Gaussian process prior

KW - Monte Carlo methods

KW - approximate Bayesian inference method

M3 - Paper

ER -

Ingram BR, Cornford D, Csató L. A projected process kriging algorithm for sensor networks with heterogeneous error characteristics. 2008. Paper presented at 8th International Geostatistics Congress, Santiago , Chile.