The contamination of 'data impact' in global models by rapidly growing mesoscale instabilities

Daniel Hodyss, Sharanya J Majumdar

Research output: Contribution to journalArticle

19 Citations (Scopus)

Abstract

This paper illustrates a caveat in the 'data impact' method, in which the influence of assimilating a specific set of observations on a numerical weather forecast is evaluated. The 'signal' of data impact is defined as the difference between two forecasts, which are identical except that one forecast withholds the data in question from the assimilation. While it is anticipated that a coherent signal from observations in the midlatitude storm track is propagated dynamically by a forecast model from the vicinity of the observation locations, the reality is that the signal may become contaminated by initially small instabilities in dynamically unrelated locations. The initial signal may be non-zero in these remote locations due to small differences between the two analyses arising from truncation errors in the data assimilation scheme and/or the model's truncated spectral basis. The notion that the dynamical signal is contaminated is corroborated by assessing the data impact in the Northern Hemisphere of one rawinsonde released over Antarctica. After just a few hours, an amplifying signal manifests itself in convective areas in the Tropics, and even in locations along the midlatitude storm track where moist instabilities exist. Rapid growth and upscale evolution from the mesoscale to synoptic scales is evident. We find that the evaluation of the efficacy of a given set of observations on weather forecasts of more than two days may be compromised by initially small instabilities particularly for spectral models and data assimilation schemes. The effective time is expected to be shorter in the Tropics.

Original languageEnglish (US)
Pages (from-to)1865-1875
Number of pages11
JournalQuarterly Journal of the Royal Meteorological Society
Volume133
Issue number628
DOIs
StatePublished - Oct 2007

Fingerprint

storm track
data assimilation
weather
global model
contamination
Northern Hemisphere
forecast
tropics
assimilation
evaluation
method
Antarctica

Keywords

  • Mesoscale predictability
  • Signal propagation
  • THORPEX

ASJC Scopus subject areas

  • Atmospheric Science

Cite this

The contamination of 'data impact' in global models by rapidly growing mesoscale instabilities. / Hodyss, Daniel; Majumdar, Sharanya J.

In: Quarterly Journal of the Royal Meteorological Society, Vol. 133, No. 628, 10.2007, p. 1865-1875.

Research output: Contribution to journalArticle

@article{d1f8b5fa5b684415b93885160b1ffd5b,
title = "The contamination of 'data impact' in global models by rapidly growing mesoscale instabilities",
abstract = "This paper illustrates a caveat in the 'data impact' method, in which the influence of assimilating a specific set of observations on a numerical weather forecast is evaluated. The 'signal' of data impact is defined as the difference between two forecasts, which are identical except that one forecast withholds the data in question from the assimilation. While it is anticipated that a coherent signal from observations in the midlatitude storm track is propagated dynamically by a forecast model from the vicinity of the observation locations, the reality is that the signal may become contaminated by initially small instabilities in dynamically unrelated locations. The initial signal may be non-zero in these remote locations due to small differences between the two analyses arising from truncation errors in the data assimilation scheme and/or the model's truncated spectral basis. The notion that the dynamical signal is contaminated is corroborated by assessing the data impact in the Northern Hemisphere of one rawinsonde released over Antarctica. After just a few hours, an amplifying signal manifests itself in convective areas in the Tropics, and even in locations along the midlatitude storm track where moist instabilities exist. Rapid growth and upscale evolution from the mesoscale to synoptic scales is evident. We find that the evaluation of the efficacy of a given set of observations on weather forecasts of more than two days may be compromised by initially small instabilities particularly for spectral models and data assimilation schemes. The effective time is expected to be shorter in the Tropics.",
keywords = "Mesoscale predictability, Signal propagation, THORPEX",
author = "Daniel Hodyss and Majumdar, {Sharanya J}",
year = "2007",
month = "10",
doi = "10.1002/qj.157",
language = "English (US)",
volume = "133",
pages = "1865--1875",
journal = "Quarterly Journal of the Royal Meteorological Society",
issn = "0035-9009",
publisher = "John Wiley and Sons Ltd",
number = "628",

}

TY - JOUR

T1 - The contamination of 'data impact' in global models by rapidly growing mesoscale instabilities

AU - Hodyss, Daniel

AU - Majumdar, Sharanya J

PY - 2007/10

Y1 - 2007/10

N2 - This paper illustrates a caveat in the 'data impact' method, in which the influence of assimilating a specific set of observations on a numerical weather forecast is evaluated. The 'signal' of data impact is defined as the difference between two forecasts, which are identical except that one forecast withholds the data in question from the assimilation. While it is anticipated that a coherent signal from observations in the midlatitude storm track is propagated dynamically by a forecast model from the vicinity of the observation locations, the reality is that the signal may become contaminated by initially small instabilities in dynamically unrelated locations. The initial signal may be non-zero in these remote locations due to small differences between the two analyses arising from truncation errors in the data assimilation scheme and/or the model's truncated spectral basis. The notion that the dynamical signal is contaminated is corroborated by assessing the data impact in the Northern Hemisphere of one rawinsonde released over Antarctica. After just a few hours, an amplifying signal manifests itself in convective areas in the Tropics, and even in locations along the midlatitude storm track where moist instabilities exist. Rapid growth and upscale evolution from the mesoscale to synoptic scales is evident. We find that the evaluation of the efficacy of a given set of observations on weather forecasts of more than two days may be compromised by initially small instabilities particularly for spectral models and data assimilation schemes. The effective time is expected to be shorter in the Tropics.

AB - This paper illustrates a caveat in the 'data impact' method, in which the influence of assimilating a specific set of observations on a numerical weather forecast is evaluated. The 'signal' of data impact is defined as the difference between two forecasts, which are identical except that one forecast withholds the data in question from the assimilation. While it is anticipated that a coherent signal from observations in the midlatitude storm track is propagated dynamically by a forecast model from the vicinity of the observation locations, the reality is that the signal may become contaminated by initially small instabilities in dynamically unrelated locations. The initial signal may be non-zero in these remote locations due to small differences between the two analyses arising from truncation errors in the data assimilation scheme and/or the model's truncated spectral basis. The notion that the dynamical signal is contaminated is corroborated by assessing the data impact in the Northern Hemisphere of one rawinsonde released over Antarctica. After just a few hours, an amplifying signal manifests itself in convective areas in the Tropics, and even in locations along the midlatitude storm track where moist instabilities exist. Rapid growth and upscale evolution from the mesoscale to synoptic scales is evident. We find that the evaluation of the efficacy of a given set of observations on weather forecasts of more than two days may be compromised by initially small instabilities particularly for spectral models and data assimilation schemes. The effective time is expected to be shorter in the Tropics.

KW - Mesoscale predictability

KW - Signal propagation

KW - THORPEX

UR - http://www.scopus.com/inward/record.url?scp=36749015116&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=36749015116&partnerID=8YFLogxK

U2 - 10.1002/qj.157

DO - 10.1002/qj.157

M3 - Article

VL - 133

SP - 1865

EP - 1875

JO - Quarterly Journal of the Royal Meteorological Society

JF - Quarterly Journal of the Royal Meteorological Society

SN - 0035-9009

IS - 628

ER -