skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Cluster Sampling Filters for Non-Gaussian Data Assimilation

Abstract

This paper presents a fully non-Gaussian filter for sequential data assimilation. The filter is named the “cluster sampling filter”, and works by directly sampling the posterior distribution following a Markov Chain Monte-Carlo (MCMC) approach, while the prior distribution is approximated using a Gaussian Mixture Model (GMM). Specifically, a clustering step is introduced after the forecast phase of the filter, and the prior density function is estimated by fitting a GMM to the prior ensemble. Using the data likelihood function, the posterior density is then formulated as a mixture density, and is sampled following an MCMC approach. Four versions of the proposed filter, namely C ℓ MCMC , C ℓ HMC , MC- C ℓ HMC , and MC- C ℓ HMC are presented. C ℓ MCMC uses a Gaussian proposal density to sample the posterior, and C ℓ HMC is an extension to the Hamiltonian Monte-Carlo (HMC) sampling filter. MC- C ℓ MCMC and MC- C ℓ HMC are multi-chain versions of the cluster sampling filters C ℓ MCMC and C ℓ HMC respectively. The multi-chain versions are proposed to guarantee that samples are taken from the vicinities of all probability modes of the formulated posterior. The new methodologies aremore » tested using a simple one-dimensional example, and a quasi-geostrophic (QG) model with double-gyre wind forcing and bi-harmonic friction. Here, numerical results demonstrate the usefulness of using GMMs to relax the Gaussian prior assumption especially in the HMC filtering paradigm.« less

Authors:
 [1];  [2];  [2]
  1. Argonne National Lab. (ANL), Argonne, IL (United States)
  2. Virginia Polytechnic Inst. and State Univ. (Virginia Tech), Blacksburg, VA (United States)
Publication Date:
Research Org.:
Argonne National Lab. (ANL), Argonne, IL (United States)
Sponsoring Org.:
Air Force Research Laboratory (AFRL), Air Force Office of Scientific Research (AFOSR); National Science Foundation (NSF); Virginia Polytechnic Institute, Dept of Computer Science; USDOE
OSTI Identifier:
1463676
Grant/Contract Number:  
AC02-06CH11357
Resource Type:
Journal Article: Accepted Manuscript
Journal Name:
Atmosphere (Basel)
Additional Journal Information:
Journal Volume: 9; Journal Issue: 6; Journal ID: ISSN 2073-4433
Publisher:
MDPI
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING; data assimilation; ensemble filters; gaussian mixture models; hamiltonian monte-carlo; markov chain monte-carlo sampling

Citation Formats

Attia, Ahmed, Moosavi, Azam, and Sandu, Adrian. Cluster Sampling Filters for Non-Gaussian Data Assimilation. United States: N. p., 2018. Web. doi:10.3390/atmos9060213.
Attia, Ahmed, Moosavi, Azam, & Sandu, Adrian. Cluster Sampling Filters for Non-Gaussian Data Assimilation. United States. https://doi.org/10.3390/atmos9060213
Attia, Ahmed, Moosavi, Azam, and Sandu, Adrian. Thu . "Cluster Sampling Filters for Non-Gaussian Data Assimilation". United States. https://doi.org/10.3390/atmos9060213. https://www.osti.gov/servlets/purl/1463676.
@article{osti_1463676,
title = {Cluster Sampling Filters for Non-Gaussian Data Assimilation},
author = {Attia, Ahmed and Moosavi, Azam and Sandu, Adrian},
abstractNote = {This paper presents a fully non-Gaussian filter for sequential data assimilation. The filter is named the “cluster sampling filter”, and works by directly sampling the posterior distribution following a Markov Chain Monte-Carlo (MCMC) approach, while the prior distribution is approximated using a Gaussian Mixture Model (GMM). Specifically, a clustering step is introduced after the forecast phase of the filter, and the prior density function is estimated by fitting a GMM to the prior ensemble. Using the data likelihood function, the posterior density is then formulated as a mixture density, and is sampled following an MCMC approach. Four versions of the proposed filter, namely C ℓ MCMC , C ℓ HMC , MC- C ℓ HMC , and MC- C ℓ HMC are presented. C ℓ MCMC uses a Gaussian proposal density to sample the posterior, and C ℓ HMC is an extension to the Hamiltonian Monte-Carlo (HMC) sampling filter. MC- C ℓ MCMC and MC- C ℓ HMC are multi-chain versions of the cluster sampling filters C ℓ MCMC and C ℓ HMC respectively. The multi-chain versions are proposed to guarantee that samples are taken from the vicinities of all probability modes of the formulated posterior. The new methodologies are tested using a simple one-dimensional example, and a quasi-geostrophic (QG) model with double-gyre wind forcing and bi-harmonic friction. Here, numerical results demonstrate the usefulness of using GMMs to relax the Gaussian prior assumption especially in the HMC filtering paradigm.},
doi = {10.3390/atmos9060213},
url = {https://www.osti.gov/biblio/1463676}, journal = {Atmosphere (Basel)},
issn = {2073-4433},
number = 6,
volume = 9,
place = {United States},
year = {2018},
month = {5}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record

Citation Metrics:
Cited by: 1 work
Citation information provided by
Web of Science

Save / Share:

Works referenced in this record:

Estimating the Dimension of a Model
journal, March 1978


Hybrid Monte Carlo
journal, September 1987


The Maximum Likelihood Ensemble Filter as a non‐differentiable minimization algorithm
journal, April 2008

  • Zupanski, Milija; Navon, I. Michael; Zupanski, Dusanka
  • Quarterly Journal of the Royal Meteorological Society, Vol. 134, Issue 633
  • https://doi.org/10.1002/qj.251

An Iterative EnKF for Strongly Nonlinear Systems
journal, June 2012


Analysis methods for numerical weather prediction
journal, October 1986


Data Assimilation Using an Ensemble Kalman Filter Technique
journal, March 1998


Evaluation of probabilistic prediction systems for a scalar variable
journal, July 2005


Distance-Dependent Filtering of Background Error Covariance Estimates in an Ensemble Kalman Filter
journal, November 2001


A deterministic formulation of the ensemble Kalman filter: an alternative to ensemble square root filters
journal, January 2008


Cluster ensemble Kalman filter
journal, January 2007


A Sequential Ensemble Kalman Filter for Atmospheric Data Assimilation
journal, January 2001


An Iterative Ensemble Kalman Filter for Multiphase Fluid Flow Data Assimilation
journal, November 2007


Auxiliary Variable Methods for Markov Chain Monte Carlo with Applications
journal, June 1998


A Hybrid Monte Carlo Sampling Filter for Non-Gaussian Data Assimilation
journal, January 2015


Analysis Scheme in the Ensemble Kalman Filter
journal, June 1998


Construction of correlation functions in two and three dimensions
journal, January 1999


Covariance Tapering for Interpolation of Large Spatial Datasets
journal, September 2006


A New Approach to Linear Filtering and Prediction Problems
journal, March 1960


Ensemble Data Assimilation without Perturbed Observations
journal, July 2002


New Results in Linear Filtering and Prediction Theory
journal, March 1961


Maximum Likelihood Ensemble Filter: Theoretical Aspects
journal, June 2005


The reduced-order hybrid Monte Carlo sampling smoother: The reduced-order hybrid Monte Carlo sampling smoother
journal, June 2016


Riemann manifold Langevin and Hamiltonian Monte Carlo methods: Riemann Manifold Langevin and Hamiltonian Monte Carlo Methods
journal, March 2011


A Hybrid Monte-Carlo sampling smoother for four-dimensional data assimilation: HMC SAMPLING SMOOTHER FOR 4D-DA
journal, June 2016


Ensemble Square Root Filters*
journal, July 2003


Measures of multivariate skewness and kurtosis with applications
journal, January 1970