Improving tropical cyclone intensity forecasts with PRIME

Kieran T. Bhatia, David S Nolan, Andrea B. Schumacher, Mark DeMaria

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

The Prediction of Intensity Model Error (PRIME) forecasting scheme uses various large-scale meteorological parameters as well as proxies for initial condition uncertainty and atmospheric flow stability to provide operational forecasts of tropical cyclone intensity forecast error. PRIME forecasts of bias and absolute error are developed for the Logistic Growth Equation Model (LGEM), Decay Statistical Hurricane Intensity Prediction Scheme (DSHP), Hurricane Weather Research and Forecasting Interpolated Model (HWFI), and Geophysical Fluid Dynamics Laboratory Interpolated Hurricane Model (GHMI). These forecasts are evaluated in the Atlantic and east Pacific basins for the 2011-15 hurricane seasons. PRIME is also trained with retrospective forecasts (R-PRIME) from the 2015 version of each model. PRIME error forecasts are significantly better than forecasts that use error climatology for a majority of forecast hours, which raises the question of whether PRIME could provide more than error guidance. PRIME bias forecasts for each model are used to modify intensity forecasts, and the corrected forecasts are compared with the original intensity forecasts. For almost all basins, forecast intervals, and versions of PRIME, the bias-corrected forecasts achieve significantly lower errors than the original intensity forecasts. PRIME absolute error and bias forecasts are also used to create unique ensembles of the four models. These PRIME-modified ensembles are found to frequently outperform the intensity consensus (ICON), the equally weighted ensemble of DSHP, LGEM, GHMI, and HWFI.

Original languageEnglish (US)
Pages (from-to)1353-1377
Number of pages25
JournalWeather and Forecasting
Volume32
Issue number4
DOIs
StatePublished - Aug 1 2017

Fingerprint

tropical cyclone
prediction
hurricane
forecast
logistics
flow stability
weather
fluid dynamics
basin
climatology

Keywords

  • Error analysis
  • Forecast verification/skill
  • Forecasting techniques
  • Hurricanes/typhoons
  • Statistical forecasting

ASJC Scopus subject areas

  • Atmospheric Science

Cite this

Improving tropical cyclone intensity forecasts with PRIME. / Bhatia, Kieran T.; Nolan, David S; Schumacher, Andrea B.; DeMaria, Mark.

In: Weather and Forecasting, Vol. 32, No. 4, 01.08.2017, p. 1353-1377.

Research output: Contribution to journalArticle

Bhatia, Kieran T. ; Nolan, David S ; Schumacher, Andrea B. ; DeMaria, Mark. / Improving tropical cyclone intensity forecasts with PRIME. In: Weather and Forecasting. 2017 ; Vol. 32, No. 4. pp. 1353-1377.
@article{9c7c961c08eb44bcb8cc393ecef3fa27,
title = "Improving tropical cyclone intensity forecasts with PRIME",
abstract = "The Prediction of Intensity Model Error (PRIME) forecasting scheme uses various large-scale meteorological parameters as well as proxies for initial condition uncertainty and atmospheric flow stability to provide operational forecasts of tropical cyclone intensity forecast error. PRIME forecasts of bias and absolute error are developed for the Logistic Growth Equation Model (LGEM), Decay Statistical Hurricane Intensity Prediction Scheme (DSHP), Hurricane Weather Research and Forecasting Interpolated Model (HWFI), and Geophysical Fluid Dynamics Laboratory Interpolated Hurricane Model (GHMI). These forecasts are evaluated in the Atlantic and east Pacific basins for the 2011-15 hurricane seasons. PRIME is also trained with retrospective forecasts (R-PRIME) from the 2015 version of each model. PRIME error forecasts are significantly better than forecasts that use error climatology for a majority of forecast hours, which raises the question of whether PRIME could provide more than error guidance. PRIME bias forecasts for each model are used to modify intensity forecasts, and the corrected forecasts are compared with the original intensity forecasts. For almost all basins, forecast intervals, and versions of PRIME, the bias-corrected forecasts achieve significantly lower errors than the original intensity forecasts. PRIME absolute error and bias forecasts are also used to create unique ensembles of the four models. These PRIME-modified ensembles are found to frequently outperform the intensity consensus (ICON), the equally weighted ensemble of DSHP, LGEM, GHMI, and HWFI.",
keywords = "Error analysis, Forecast verification/skill, Forecasting techniques, Hurricanes/typhoons, Statistical forecasting",
author = "Bhatia, {Kieran T.} and Nolan, {David S} and Schumacher, {Andrea B.} and Mark DeMaria",
year = "2017",
month = "8",
day = "1",
doi = "10.1175/WAF-D-17-0009.1",
language = "English (US)",
volume = "32",
pages = "1353--1377",
journal = "Weather and Forecasting",
issn = "0882-8156",
publisher = "American Meteorological Society",
number = "4",

}

TY - JOUR

T1 - Improving tropical cyclone intensity forecasts with PRIME

AU - Bhatia, Kieran T.

AU - Nolan, David S

AU - Schumacher, Andrea B.

AU - DeMaria, Mark

PY - 2017/8/1

Y1 - 2017/8/1

N2 - The Prediction of Intensity Model Error (PRIME) forecasting scheme uses various large-scale meteorological parameters as well as proxies for initial condition uncertainty and atmospheric flow stability to provide operational forecasts of tropical cyclone intensity forecast error. PRIME forecasts of bias and absolute error are developed for the Logistic Growth Equation Model (LGEM), Decay Statistical Hurricane Intensity Prediction Scheme (DSHP), Hurricane Weather Research and Forecasting Interpolated Model (HWFI), and Geophysical Fluid Dynamics Laboratory Interpolated Hurricane Model (GHMI). These forecasts are evaluated in the Atlantic and east Pacific basins for the 2011-15 hurricane seasons. PRIME is also trained with retrospective forecasts (R-PRIME) from the 2015 version of each model. PRIME error forecasts are significantly better than forecasts that use error climatology for a majority of forecast hours, which raises the question of whether PRIME could provide more than error guidance. PRIME bias forecasts for each model are used to modify intensity forecasts, and the corrected forecasts are compared with the original intensity forecasts. For almost all basins, forecast intervals, and versions of PRIME, the bias-corrected forecasts achieve significantly lower errors than the original intensity forecasts. PRIME absolute error and bias forecasts are also used to create unique ensembles of the four models. These PRIME-modified ensembles are found to frequently outperform the intensity consensus (ICON), the equally weighted ensemble of DSHP, LGEM, GHMI, and HWFI.

AB - The Prediction of Intensity Model Error (PRIME) forecasting scheme uses various large-scale meteorological parameters as well as proxies for initial condition uncertainty and atmospheric flow stability to provide operational forecasts of tropical cyclone intensity forecast error. PRIME forecasts of bias and absolute error are developed for the Logistic Growth Equation Model (LGEM), Decay Statistical Hurricane Intensity Prediction Scheme (DSHP), Hurricane Weather Research and Forecasting Interpolated Model (HWFI), and Geophysical Fluid Dynamics Laboratory Interpolated Hurricane Model (GHMI). These forecasts are evaluated in the Atlantic and east Pacific basins for the 2011-15 hurricane seasons. PRIME is also trained with retrospective forecasts (R-PRIME) from the 2015 version of each model. PRIME error forecasts are significantly better than forecasts that use error climatology for a majority of forecast hours, which raises the question of whether PRIME could provide more than error guidance. PRIME bias forecasts for each model are used to modify intensity forecasts, and the corrected forecasts are compared with the original intensity forecasts. For almost all basins, forecast intervals, and versions of PRIME, the bias-corrected forecasts achieve significantly lower errors than the original intensity forecasts. PRIME absolute error and bias forecasts are also used to create unique ensembles of the four models. These PRIME-modified ensembles are found to frequently outperform the intensity consensus (ICON), the equally weighted ensemble of DSHP, LGEM, GHMI, and HWFI.

KW - Error analysis

KW - Forecast verification/skill

KW - Forecasting techniques

KW - Hurricanes/typhoons

KW - Statistical forecasting

UR - http://www.scopus.com/inward/record.url?scp=85028995262&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85028995262&partnerID=8YFLogxK

U2 - 10.1175/WAF-D-17-0009.1

DO - 10.1175/WAF-D-17-0009.1

M3 - Article

VL - 32

SP - 1353

EP - 1377

JO - Weather and Forecasting

JF - Weather and Forecasting

SN - 0882-8156

IS - 4

ER -