Singular Spectrum Analysis With Conditional Predictions for Real-Time State Estimation and Forecasting

H. Reed Ogrosky, Samuel N. Stechmann, Nan Chen, Andrew Majda

Research output: Contribution to journalArticle

Abstract

Singular spectrum analysis (SSA) or extended empirical orthogonal function methods are powerful, commonly used data-driven techniques to identify modes of variability in time series and space-time data sets. Due to the time-lagged embedding, these methods can provide inaccurate reconstructions of leading modes near the endpoints, which can hinder the use of these methods in real time. A modified version of the traditional SSA algorithm, referred to as SSA with conditional predictions (SSA-CP), is presented to address these issues. It is tested on low-dimensional, approximately Gaussian data, high-dimensional non-Gaussian data, and partially observed data from a multiscale model. In each case, SSA-CP provides a more accurate real-time estimate of the leading modes of variability than the traditional reconstruction. SSA-CP also provides predictions of the leading modes and is easy to implement. SSA-CP is optimal in the case of Gaussian data, and the uncertainty in real-time estimates of leading modes is easily quantified.

Original languageEnglish (US)
Pages (from-to)1851-1860
Number of pages10
JournalGeophysical Research Letters
Volume46
Issue number3
DOIs
StatePublished - Feb 16 2019

Fingerprint

state estimation
forecasting
spectrum analysis
prediction
predictions
multiscale models
orthogonal functions
estimates
embedding
analysis
time series
method

ASJC Scopus subject areas

  • Geophysics
  • Earth and Planetary Sciences(all)

Cite this

Singular Spectrum Analysis With Conditional Predictions for Real-Time State Estimation and Forecasting. / Ogrosky, H. Reed; Stechmann, Samuel N.; Chen, Nan; Majda, Andrew.

In: Geophysical Research Letters, Vol. 46, No. 3, 16.02.2019, p. 1851-1860.

Research output: Contribution to journalArticle

Ogrosky, H. Reed ; Stechmann, Samuel N. ; Chen, Nan ; Majda, Andrew. / Singular Spectrum Analysis With Conditional Predictions for Real-Time State Estimation and Forecasting. In: Geophysical Research Letters. 2019 ; Vol. 46, No. 3. pp. 1851-1860.
@article{078c77b0ed8a459989b0af3a00aa022f,
title = "Singular Spectrum Analysis With Conditional Predictions for Real-Time State Estimation and Forecasting",
abstract = "Singular spectrum analysis (SSA) or extended empirical orthogonal function methods are powerful, commonly used data-driven techniques to identify modes of variability in time series and space-time data sets. Due to the time-lagged embedding, these methods can provide inaccurate reconstructions of leading modes near the endpoints, which can hinder the use of these methods in real time. A modified version of the traditional SSA algorithm, referred to as SSA with conditional predictions (SSA-CP), is presented to address these issues. It is tested on low-dimensional, approximately Gaussian data, high-dimensional non-Gaussian data, and partially observed data from a multiscale model. In each case, SSA-CP provides a more accurate real-time estimate of the leading modes of variability than the traditional reconstruction. SSA-CP also provides predictions of the leading modes and is easy to implement. SSA-CP is optimal in the case of Gaussian data, and the uncertainty in real-time estimates of leading modes is easily quantified.",
author = "Ogrosky, {H. Reed} and Stechmann, {Samuel N.} and Nan Chen and Andrew Majda",
year = "2019",
month = "2",
day = "16",
doi = "10.1029/2018GL081100",
language = "English (US)",
volume = "46",
pages = "1851--1860",
journal = "Geophysical Research Letters",
issn = "0094-8276",
publisher = "American Geophysical Union",
number = "3",

}

TY - JOUR

T1 - Singular Spectrum Analysis With Conditional Predictions for Real-Time State Estimation and Forecasting

AU - Ogrosky, H. Reed

AU - Stechmann, Samuel N.

AU - Chen, Nan

AU - Majda, Andrew

PY - 2019/2/16

Y1 - 2019/2/16

N2 - Singular spectrum analysis (SSA) or extended empirical orthogonal function methods are powerful, commonly used data-driven techniques to identify modes of variability in time series and space-time data sets. Due to the time-lagged embedding, these methods can provide inaccurate reconstructions of leading modes near the endpoints, which can hinder the use of these methods in real time. A modified version of the traditional SSA algorithm, referred to as SSA with conditional predictions (SSA-CP), is presented to address these issues. It is tested on low-dimensional, approximately Gaussian data, high-dimensional non-Gaussian data, and partially observed data from a multiscale model. In each case, SSA-CP provides a more accurate real-time estimate of the leading modes of variability than the traditional reconstruction. SSA-CP also provides predictions of the leading modes and is easy to implement. SSA-CP is optimal in the case of Gaussian data, and the uncertainty in real-time estimates of leading modes is easily quantified.

AB - Singular spectrum analysis (SSA) or extended empirical orthogonal function methods are powerful, commonly used data-driven techniques to identify modes of variability in time series and space-time data sets. Due to the time-lagged embedding, these methods can provide inaccurate reconstructions of leading modes near the endpoints, which can hinder the use of these methods in real time. A modified version of the traditional SSA algorithm, referred to as SSA with conditional predictions (SSA-CP), is presented to address these issues. It is tested on low-dimensional, approximately Gaussian data, high-dimensional non-Gaussian data, and partially observed data from a multiscale model. In each case, SSA-CP provides a more accurate real-time estimate of the leading modes of variability than the traditional reconstruction. SSA-CP also provides predictions of the leading modes and is easy to implement. SSA-CP is optimal in the case of Gaussian data, and the uncertainty in real-time estimates of leading modes is easily quantified.

UR - http://www.scopus.com/inward/record.url?scp=85061671512&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85061671512&partnerID=8YFLogxK

U2 - 10.1029/2018GL081100

DO - 10.1029/2018GL081100

M3 - Article

AN - SCOPUS:85061671512

VL - 46

SP - 1851

EP - 1860

JO - Geophysical Research Letters

JF - Geophysical Research Letters

SN - 0094-8276

IS - 3

ER -