Using high-resolution simulations to quantify errors in radar estimates of tornado intensity

Nathan A. Dahl, David S Nolan

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Observation experiments are performed on a set of high-resolution large-eddy simulations of translating tornado-like vortices. Near-surface Doppler wind measurements are taken by emulating a mobile radar positioned from 1 to 10 km south of each vortex track and conducting single-level scans every 2 s. The departure of each observed gust (wind measurement averaged over two successive scans) from the corresponding true maximum 3-s gust at 10 m AGL ("S10-3s") is partitioned into error sources associated with resolution volume size, wind direction relative to the radar beam, beam elevation, and temporal sampling. The distributions of each error type are diagrammed as functions of range, observed wind speed, and predicted deviation between the wind direction and the radar beam. The results indicate that the deviation between the wind direction and the radar beam is the predominant source of error in these rapid scan scenarios, although range is also a substantial factor. The median total error is ~10% for small deviation at close range, but it approximately doubles if the range is increased from 1 to 10 km; a more pronounced increase in both the median value and the variance of the total error is seen as the deviation becomes large. Because of this, the underestimate of the global maximum S10-3s approaches 30-40 m s-1 at a longer range, although the global maximum of the time-averaged observed wind speed gives a reasonable approximation of the time-mean maximum S10-3s in many cases. Because of simplifying assumptions and the limited number of cases examined, these results are intended as a baseline for further research.

Original languageEnglish (US)
Pages (from-to)2271-2296
Number of pages26
JournalMonthly Weather Review
Volume146
Issue number7
DOIs
StatePublished - Jul 1 2018

Fingerprint

tornado
radar
wind direction
gust
wind measurement
simulation
vortex
wind velocity
large eddy simulation
sampling
experiment

Keywords

  • Data processing
  • Large eddy simulations
  • Radars/Radar observations
  • Tornadoes

ASJC Scopus subject areas

  • Atmospheric Science

Cite this

Using high-resolution simulations to quantify errors in radar estimates of tornado intensity. / Dahl, Nathan A.; Nolan, David S.

In: Monthly Weather Review, Vol. 146, No. 7, 01.07.2018, p. 2271-2296.

Research output: Contribution to journalArticle

@article{d51943db622d406cb21c601b1e3a71d9,
title = "Using high-resolution simulations to quantify errors in radar estimates of tornado intensity",
abstract = "Observation experiments are performed on a set of high-resolution large-eddy simulations of translating tornado-like vortices. Near-surface Doppler wind measurements are taken by emulating a mobile radar positioned from 1 to 10 km south of each vortex track and conducting single-level scans every 2 s. The departure of each observed gust (wind measurement averaged over two successive scans) from the corresponding true maximum 3-s gust at 10 m AGL ({"}S10-3s{"}) is partitioned into error sources associated with resolution volume size, wind direction relative to the radar beam, beam elevation, and temporal sampling. The distributions of each error type are diagrammed as functions of range, observed wind speed, and predicted deviation between the wind direction and the radar beam. The results indicate that the deviation between the wind direction and the radar beam is the predominant source of error in these rapid scan scenarios, although range is also a substantial factor. The median total error is ~10{\%} for small deviation at close range, but it approximately doubles if the range is increased from 1 to 10 km; a more pronounced increase in both the median value and the variance of the total error is seen as the deviation becomes large. Because of this, the underestimate of the global maximum S10-3s approaches 30-40 m s-1 at a longer range, although the global maximum of the time-averaged observed wind speed gives a reasonable approximation of the time-mean maximum S10-3s in many cases. Because of simplifying assumptions and the limited number of cases examined, these results are intended as a baseline for further research.",
keywords = "Data processing, Large eddy simulations, Radars/Radar observations, Tornadoes",
author = "Dahl, {Nathan A.} and Nolan, {David S}",
year = "2018",
month = "7",
day = "1",
doi = "10.1175/MWR-D-17-0333.1",
language = "English (US)",
volume = "146",
pages = "2271--2296",
journal = "Monthly Weather Review",
issn = "0027-0644",
publisher = "American Meteorological Society",
number = "7",

}

TY - JOUR

T1 - Using high-resolution simulations to quantify errors in radar estimates of tornado intensity

AU - Dahl, Nathan A.

AU - Nolan, David S

PY - 2018/7/1

Y1 - 2018/7/1

N2 - Observation experiments are performed on a set of high-resolution large-eddy simulations of translating tornado-like vortices. Near-surface Doppler wind measurements are taken by emulating a mobile radar positioned from 1 to 10 km south of each vortex track and conducting single-level scans every 2 s. The departure of each observed gust (wind measurement averaged over two successive scans) from the corresponding true maximum 3-s gust at 10 m AGL ("S10-3s") is partitioned into error sources associated with resolution volume size, wind direction relative to the radar beam, beam elevation, and temporal sampling. The distributions of each error type are diagrammed as functions of range, observed wind speed, and predicted deviation between the wind direction and the radar beam. The results indicate that the deviation between the wind direction and the radar beam is the predominant source of error in these rapid scan scenarios, although range is also a substantial factor. The median total error is ~10% for small deviation at close range, but it approximately doubles if the range is increased from 1 to 10 km; a more pronounced increase in both the median value and the variance of the total error is seen as the deviation becomes large. Because of this, the underestimate of the global maximum S10-3s approaches 30-40 m s-1 at a longer range, although the global maximum of the time-averaged observed wind speed gives a reasonable approximation of the time-mean maximum S10-3s in many cases. Because of simplifying assumptions and the limited number of cases examined, these results are intended as a baseline for further research.

AB - Observation experiments are performed on a set of high-resolution large-eddy simulations of translating tornado-like vortices. Near-surface Doppler wind measurements are taken by emulating a mobile radar positioned from 1 to 10 km south of each vortex track and conducting single-level scans every 2 s. The departure of each observed gust (wind measurement averaged over two successive scans) from the corresponding true maximum 3-s gust at 10 m AGL ("S10-3s") is partitioned into error sources associated with resolution volume size, wind direction relative to the radar beam, beam elevation, and temporal sampling. The distributions of each error type are diagrammed as functions of range, observed wind speed, and predicted deviation between the wind direction and the radar beam. The results indicate that the deviation between the wind direction and the radar beam is the predominant source of error in these rapid scan scenarios, although range is also a substantial factor. The median total error is ~10% for small deviation at close range, but it approximately doubles if the range is increased from 1 to 10 km; a more pronounced increase in both the median value and the variance of the total error is seen as the deviation becomes large. Because of this, the underestimate of the global maximum S10-3s approaches 30-40 m s-1 at a longer range, although the global maximum of the time-averaged observed wind speed gives a reasonable approximation of the time-mean maximum S10-3s in many cases. Because of simplifying assumptions and the limited number of cases examined, these results are intended as a baseline for further research.

KW - Data processing

KW - Large eddy simulations

KW - Radars/Radar observations

KW - Tornadoes

UR - http://www.scopus.com/inward/record.url?scp=85050147351&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85050147351&partnerID=8YFLogxK

U2 - 10.1175/MWR-D-17-0333.1

DO - 10.1175/MWR-D-17-0333.1

M3 - Article

AN - SCOPUS:85050147351

VL - 146

SP - 2271

EP - 2296

JO - Monthly Weather Review

JF - Monthly Weather Review

SN - 0027-0644

IS - 7

ER -