Nonlinear stability of the ensemble kalman filter with adaptive covariance inflation

Xin T. Tong, Andrew J. Majda, David Kelly

Research output: Contribution to journalArticle

Abstract

The ensemble Kalman filter and ensemble square root filters are data assimilation methods used to combine high dimensional nonlinear models with observed data. These methods have proved to be indispensable tools in science and engineering as they allow computationally cheap, low dimensional ensemble state approximation for extremely high dimensional turbulent forecast models. From a theoretical perspective, these methods are poorly understood, with the exception of a recently established but still incomplete nonlinear stability theory. Moreover, recent numerical and theoretical studies of catastrophic filter divergence have indicated that stability is a genuine mathematical concern and can not be taken for granted in implementation. In this article we propose a simple modification of ensemble based methods which resolves these stability issues entirely. The method involves a new type of adaptive covariance inflation, which comes with minimal additional cost. We develop a complete nonlinear stability theory for the adaptive method, yielding Lyapunov functions and geometric ergodicity under weak assumptions. We present numerical evidence which suggests the adaptive methods have improved accuracy over standard methods and completely eliminate catastrophic filter divergence. This enhanced stability allows for the use of extremely cheap, unstable forecast integrators, which would otherwise lead to widespread filter malfunction.

Original languageEnglish (US)
Pages (from-to)1283-1313
Number of pages31
JournalCommunications in Mathematical Sciences
Volume14
Issue number5
DOIs
StatePublished - 2016

Fingerprint

Ensemble Kalman Filter
Nonlinear Stability
Kalman filters
Inflation
Filter
Ensemble
Stability Theory
Adaptive Method
Forecast
Divergence
High-dimensional
Geometric Ergodicity
Lyapunov functions
Data Assimilation
Square root
Lyapunov Function
Exception
Nonlinear Model
Resolve
Eliminate

Keywords

  • Data assimilation
  • Ensemble kalman filter
  • Ergodicity of markov chains
  • Nonlinear stability

ASJC Scopus subject areas

  • Mathematics(all)
  • Applied Mathematics

Cite this

Nonlinear stability of the ensemble kalman filter with adaptive covariance inflation. / Tong, Xin T.; Majda, Andrew J.; Kelly, David.

In: Communications in Mathematical Sciences, Vol. 14, No. 5, 2016, p. 1283-1313.

Research output: Contribution to journalArticle

Tong, Xin T. ; Majda, Andrew J. ; Kelly, David. / Nonlinear stability of the ensemble kalman filter with adaptive covariance inflation. In: Communications in Mathematical Sciences. 2016 ; Vol. 14, No. 5. pp. 1283-1313.
@article{47c5c78ebb8c44ef9d8d5e0d23e23c13,
title = "Nonlinear stability of the ensemble kalman filter with adaptive covariance inflation",
abstract = "The ensemble Kalman filter and ensemble square root filters are data assimilation methods used to combine high dimensional nonlinear models with observed data. These methods have proved to be indispensable tools in science and engineering as they allow computationally cheap, low dimensional ensemble state approximation for extremely high dimensional turbulent forecast models. From a theoretical perspective, these methods are poorly understood, with the exception of a recently established but still incomplete nonlinear stability theory. Moreover, recent numerical and theoretical studies of catastrophic filter divergence have indicated that stability is a genuine mathematical concern and can not be taken for granted in implementation. In this article we propose a simple modification of ensemble based methods which resolves these stability issues entirely. The method involves a new type of adaptive covariance inflation, which comes with minimal additional cost. We develop a complete nonlinear stability theory for the adaptive method, yielding Lyapunov functions and geometric ergodicity under weak assumptions. We present numerical evidence which suggests the adaptive methods have improved accuracy over standard methods and completely eliminate catastrophic filter divergence. This enhanced stability allows for the use of extremely cheap, unstable forecast integrators, which would otherwise lead to widespread filter malfunction.",
keywords = "Data assimilation, Ensemble kalman filter, Ergodicity of markov chains, Nonlinear stability",
author = "Tong, {Xin T.} and Majda, {Andrew J.} and David Kelly",
year = "2016",
doi = "10.4310/CMS.2016.v14.n5.a5",
language = "English (US)",
volume = "14",
pages = "1283--1313",
journal = "Communications in Mathematical Sciences",
issn = "1539-6746",
publisher = "International Press of Boston, Inc.",
number = "5",

}

TY - JOUR

T1 - Nonlinear stability of the ensemble kalman filter with adaptive covariance inflation

AU - Tong, Xin T.

AU - Majda, Andrew J.

AU - Kelly, David

PY - 2016

Y1 - 2016

N2 - The ensemble Kalman filter and ensemble square root filters are data assimilation methods used to combine high dimensional nonlinear models with observed data. These methods have proved to be indispensable tools in science and engineering as they allow computationally cheap, low dimensional ensemble state approximation for extremely high dimensional turbulent forecast models. From a theoretical perspective, these methods are poorly understood, with the exception of a recently established but still incomplete nonlinear stability theory. Moreover, recent numerical and theoretical studies of catastrophic filter divergence have indicated that stability is a genuine mathematical concern and can not be taken for granted in implementation. In this article we propose a simple modification of ensemble based methods which resolves these stability issues entirely. The method involves a new type of adaptive covariance inflation, which comes with minimal additional cost. We develop a complete nonlinear stability theory for the adaptive method, yielding Lyapunov functions and geometric ergodicity under weak assumptions. We present numerical evidence which suggests the adaptive methods have improved accuracy over standard methods and completely eliminate catastrophic filter divergence. This enhanced stability allows for the use of extremely cheap, unstable forecast integrators, which would otherwise lead to widespread filter malfunction.

AB - The ensemble Kalman filter and ensemble square root filters are data assimilation methods used to combine high dimensional nonlinear models with observed data. These methods have proved to be indispensable tools in science and engineering as they allow computationally cheap, low dimensional ensemble state approximation for extremely high dimensional turbulent forecast models. From a theoretical perspective, these methods are poorly understood, with the exception of a recently established but still incomplete nonlinear stability theory. Moreover, recent numerical and theoretical studies of catastrophic filter divergence have indicated that stability is a genuine mathematical concern and can not be taken for granted in implementation. In this article we propose a simple modification of ensemble based methods which resolves these stability issues entirely. The method involves a new type of adaptive covariance inflation, which comes with minimal additional cost. We develop a complete nonlinear stability theory for the adaptive method, yielding Lyapunov functions and geometric ergodicity under weak assumptions. We present numerical evidence which suggests the adaptive methods have improved accuracy over standard methods and completely eliminate catastrophic filter divergence. This enhanced stability allows for the use of extremely cheap, unstable forecast integrators, which would otherwise lead to widespread filter malfunction.

KW - Data assimilation

KW - Ensemble kalman filter

KW - Ergodicity of markov chains

KW - Nonlinear stability

UR - http://www.scopus.com/inward/record.url?scp=84970965492&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84970965492&partnerID=8YFLogxK

U2 - 10.4310/CMS.2016.v14.n5.a5

DO - 10.4310/CMS.2016.v14.n5.a5

M3 - Article

VL - 14

SP - 1283

EP - 1313

JO - Communications in Mathematical Sciences

JF - Communications in Mathematical Sciences

SN - 1539-6746

IS - 5

ER -