Data assimilation considerations for improved ocean predictability during the Gulf of Mexico Grand Lagrangian Deployment (GLAD)

Gregg A. Jacobs, Brent P. Bartels, Darek J. Bogucki, Francisco J Beron-Vera, Shuyi S Chen, Emanuel F. Coelho, Milan Curcic, Annalisa Griffa, Matthew Gough, Brian K Haus, Angelique C. Haza, Robert W. Helber, Patrick J. Hogan, Helga S. Huntley, Mohamed Iskandarani, Falko Judt, A. D. Kirwan, Nathan Laxague, Arnoldo Valle-Levinson, Bruce L. Lipphardt & 13 others Arthur J Mariano, Hans E. Ngodock, Guillaume Novelli, Maria J Olascoaga, Tamay M Ozgokmen, Andrew C. Poje, Ad J H M Reniers, Clark D. Rowley, Edward H. Ryan, Scott R. Smith, Peter L. Spence, Prasad G. Thoppil, Mozheng Wei

Research output: Contribution to journalArticle

22 Citations (Scopus)

Abstract

Ocean prediction systems rely on an array of assumptions to optimize their data assimilation schemes. Many of these remain untested, especially at smaller scales, because sufficiently dense observations are very rare. A set of 295 drifters deployed in July 2012 in the north-eastern Gulf of Mexico provides a unique opportunity to test these systems down to scales previously unobtainable. In this study, background error covariance assumptions in the 3DVar assimilation process are perturbed to understand the effect on the solution relative to the withheld dense drifter data. Results show that the amplitude of the background error covariance is an important factor as expected, and a proposed new formulation provides added skill. In addition, the background error covariance time correlation is important to allow satellite observations to affect the results over a period longer than one daily assimilation cycle. The results show the new background error covariance formulations provide more accurate placement of frontal positions, directions of currents and velocity magnitudes. These conclusions have implications for the implementation of 3DVar systems as well as the analysis interval of 4DVar systems.

Original languageEnglish (US)
Pages (from-to)98-117
Number of pages20
JournalOcean Modelling
Volume83
DOIs
StatePublished - Nov 1 2014

Fingerprint

drifter
data assimilation
ocean
prediction
Satellites
assimilation
gulf
observation satellite
analysis
effect
test

Keywords

  • Assimilation
  • Covariance
  • Drifter
  • Gulf of Mexico
  • Modeling
  • Ocean

ASJC Scopus subject areas

  • Atmospheric Science
  • Oceanography
  • Geotechnical Engineering and Engineering Geology
  • Computer Science (miscellaneous)

Cite this

Data assimilation considerations for improved ocean predictability during the Gulf of Mexico Grand Lagrangian Deployment (GLAD). / Jacobs, Gregg A.; Bartels, Brent P.; Bogucki, Darek J.; Beron-Vera, Francisco J; Chen, Shuyi S; Coelho, Emanuel F.; Curcic, Milan; Griffa, Annalisa; Gough, Matthew; Haus, Brian K; Haza, Angelique C.; Helber, Robert W.; Hogan, Patrick J.; Huntley, Helga S.; Iskandarani, Mohamed; Judt, Falko; Kirwan, A. D.; Laxague, Nathan; Valle-Levinson, Arnoldo; Lipphardt, Bruce L.; Mariano, Arthur J; Ngodock, Hans E.; Novelli, Guillaume; Olascoaga, Maria J; Ozgokmen, Tamay M; Poje, Andrew C.; Reniers, Ad J H M; Rowley, Clark D.; Ryan, Edward H.; Smith, Scott R.; Spence, Peter L.; Thoppil, Prasad G.; Wei, Mozheng.

In: Ocean Modelling, Vol. 83, 01.11.2014, p. 98-117.

Research output: Contribution to journalArticle

Jacobs, GA, Bartels, BP, Bogucki, DJ, Beron-Vera, FJ, Chen, SS, Coelho, EF, Curcic, M, Griffa, A, Gough, M, Haus, BK, Haza, AC, Helber, RW, Hogan, PJ, Huntley, HS, Iskandarani, M, Judt, F, Kirwan, AD, Laxague, N, Valle-Levinson, A, Lipphardt, BL, Mariano, AJ, Ngodock, HE, Novelli, G, Olascoaga, MJ, Ozgokmen, TM, Poje, AC, Reniers, AJHM, Rowley, CD, Ryan, EH, Smith, SR, Spence, PL, Thoppil, PG & Wei, M 2014, 'Data assimilation considerations for improved ocean predictability during the Gulf of Mexico Grand Lagrangian Deployment (GLAD)', Ocean Modelling, vol. 83, pp. 98-117. https://doi.org/10.1016/j.ocemod.2014.09.003
Jacobs, Gregg A. ; Bartels, Brent P. ; Bogucki, Darek J. ; Beron-Vera, Francisco J ; Chen, Shuyi S ; Coelho, Emanuel F. ; Curcic, Milan ; Griffa, Annalisa ; Gough, Matthew ; Haus, Brian K ; Haza, Angelique C. ; Helber, Robert W. ; Hogan, Patrick J. ; Huntley, Helga S. ; Iskandarani, Mohamed ; Judt, Falko ; Kirwan, A. D. ; Laxague, Nathan ; Valle-Levinson, Arnoldo ; Lipphardt, Bruce L. ; Mariano, Arthur J ; Ngodock, Hans E. ; Novelli, Guillaume ; Olascoaga, Maria J ; Ozgokmen, Tamay M ; Poje, Andrew C. ; Reniers, Ad J H M ; Rowley, Clark D. ; Ryan, Edward H. ; Smith, Scott R. ; Spence, Peter L. ; Thoppil, Prasad G. ; Wei, Mozheng. / Data assimilation considerations for improved ocean predictability during the Gulf of Mexico Grand Lagrangian Deployment (GLAD). In: Ocean Modelling. 2014 ; Vol. 83. pp. 98-117.
@article{82558dd0adfe4dbb9833b7d7c6118ca6,
title = "Data assimilation considerations for improved ocean predictability during the Gulf of Mexico Grand Lagrangian Deployment (GLAD)",
abstract = "Ocean prediction systems rely on an array of assumptions to optimize their data assimilation schemes. Many of these remain untested, especially at smaller scales, because sufficiently dense observations are very rare. A set of 295 drifters deployed in July 2012 in the north-eastern Gulf of Mexico provides a unique opportunity to test these systems down to scales previously unobtainable. In this study, background error covariance assumptions in the 3DVar assimilation process are perturbed to understand the effect on the solution relative to the withheld dense drifter data. Results show that the amplitude of the background error covariance is an important factor as expected, and a proposed new formulation provides added skill. In addition, the background error covariance time correlation is important to allow satellite observations to affect the results over a period longer than one daily assimilation cycle. The results show the new background error covariance formulations provide more accurate placement of frontal positions, directions of currents and velocity magnitudes. These conclusions have implications for the implementation of 3DVar systems as well as the analysis interval of 4DVar systems.",
keywords = "Assimilation, Covariance, Drifter, Gulf of Mexico, Modeling, Ocean",
author = "Jacobs, {Gregg A.} and Bartels, {Brent P.} and Bogucki, {Darek J.} and Beron-Vera, {Francisco J} and Chen, {Shuyi S} and Coelho, {Emanuel F.} and Milan Curcic and Annalisa Griffa and Matthew Gough and Haus, {Brian K} and Haza, {Angelique C.} and Helber, {Robert W.} and Hogan, {Patrick J.} and Huntley, {Helga S.} and Mohamed Iskandarani and Falko Judt and Kirwan, {A. D.} and Nathan Laxague and Arnoldo Valle-Levinson and Lipphardt, {Bruce L.} and Mariano, {Arthur J} and Ngodock, {Hans E.} and Guillaume Novelli and Olascoaga, {Maria J} and Ozgokmen, {Tamay M} and Poje, {Andrew C.} and Reniers, {Ad J H M} and Rowley, {Clark D.} and Ryan, {Edward H.} and Smith, {Scott R.} and Spence, {Peter L.} and Thoppil, {Prasad G.} and Mozheng Wei",
year = "2014",
month = "11",
day = "1",
doi = "10.1016/j.ocemod.2014.09.003",
language = "English (US)",
volume = "83",
pages = "98--117",
journal = "Ocean Modelling",
issn = "1463-5003",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Data assimilation considerations for improved ocean predictability during the Gulf of Mexico Grand Lagrangian Deployment (GLAD)

AU - Jacobs, Gregg A.

AU - Bartels, Brent P.

AU - Bogucki, Darek J.

AU - Beron-Vera, Francisco J

AU - Chen, Shuyi S

AU - Coelho, Emanuel F.

AU - Curcic, Milan

AU - Griffa, Annalisa

AU - Gough, Matthew

AU - Haus, Brian K

AU - Haza, Angelique C.

AU - Helber, Robert W.

AU - Hogan, Patrick J.

AU - Huntley, Helga S.

AU - Iskandarani, Mohamed

AU - Judt, Falko

AU - Kirwan, A. D.

AU - Laxague, Nathan

AU - Valle-Levinson, Arnoldo

AU - Lipphardt, Bruce L.

AU - Mariano, Arthur J

AU - Ngodock, Hans E.

AU - Novelli, Guillaume

AU - Olascoaga, Maria J

AU - Ozgokmen, Tamay M

AU - Poje, Andrew C.

AU - Reniers, Ad J H M

AU - Rowley, Clark D.

AU - Ryan, Edward H.

AU - Smith, Scott R.

AU - Spence, Peter L.

AU - Thoppil, Prasad G.

AU - Wei, Mozheng

PY - 2014/11/1

Y1 - 2014/11/1

N2 - Ocean prediction systems rely on an array of assumptions to optimize their data assimilation schemes. Many of these remain untested, especially at smaller scales, because sufficiently dense observations are very rare. A set of 295 drifters deployed in July 2012 in the north-eastern Gulf of Mexico provides a unique opportunity to test these systems down to scales previously unobtainable. In this study, background error covariance assumptions in the 3DVar assimilation process are perturbed to understand the effect on the solution relative to the withheld dense drifter data. Results show that the amplitude of the background error covariance is an important factor as expected, and a proposed new formulation provides added skill. In addition, the background error covariance time correlation is important to allow satellite observations to affect the results over a period longer than one daily assimilation cycle. The results show the new background error covariance formulations provide more accurate placement of frontal positions, directions of currents and velocity magnitudes. These conclusions have implications for the implementation of 3DVar systems as well as the analysis interval of 4DVar systems.

AB - Ocean prediction systems rely on an array of assumptions to optimize their data assimilation schemes. Many of these remain untested, especially at smaller scales, because sufficiently dense observations are very rare. A set of 295 drifters deployed in July 2012 in the north-eastern Gulf of Mexico provides a unique opportunity to test these systems down to scales previously unobtainable. In this study, background error covariance assumptions in the 3DVar assimilation process are perturbed to understand the effect on the solution relative to the withheld dense drifter data. Results show that the amplitude of the background error covariance is an important factor as expected, and a proposed new formulation provides added skill. In addition, the background error covariance time correlation is important to allow satellite observations to affect the results over a period longer than one daily assimilation cycle. The results show the new background error covariance formulations provide more accurate placement of frontal positions, directions of currents and velocity magnitudes. These conclusions have implications for the implementation of 3DVar systems as well as the analysis interval of 4DVar systems.

KW - Assimilation

KW - Covariance

KW - Drifter

KW - Gulf of Mexico

KW - Modeling

KW - Ocean

UR - http://www.scopus.com/inward/record.url?scp=84908017762&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84908017762&partnerID=8YFLogxK

U2 - 10.1016/j.ocemod.2014.09.003

DO - 10.1016/j.ocemod.2014.09.003

M3 - Article

VL - 83

SP - 98

EP - 117

JO - Ocean Modelling

JF - Ocean Modelling

SN - 1463-5003

ER -