Summary

Model type: Probabilistic hazard
Model timeliness: Late model
Model status: Experimental (to be provided to NHC and JTWC)
Basins run in: All basins
ATCF TECH identifiers: Not yet included: Ensemble mean: FHLO
Individual ensemble members: FNNN, where NNN = “000” through “999"
Quantile forecasts: FHnn, where nn = (01,05,10,25,50,75,90,95,99) are the quantiles of the intensity forecast
Forecast period: 0 to 168 hours
Included in TCGP: Included graphical products: wind exceedance plots for 34-, 50-, 64-, 83-, and 96-kt thresholds
Not yet included: Diagnostic plots for key factors affecting the intensity forecast, ensemble and density plots for track and intensity
Intensity model component: FAST intensity emulator (coupled 1d ODE)
Wind radii input: RVCN (courtesy of the Joint Typhoon Warning Center)
Wind model component: Chavas wind radii model
Domain: N/A 
Vertical coordinate: N/A 
Grid: N/A 
Cumulus parameterization: N/A 
Microphysics parameterization: N/A 
Boundary layer parameterization: N/A 
Radiation parameterization: N/A 
Ocean coupling: 0D (mixing only)
Initialization method: Forcing term to azimuthal wind speed and inner-core moisture are adjusted to match observed storm intensity from 48 hours before initialization time up until the initialization time
Initial and boundary conditions: Ensemble NWP fields (ECMWF ensemble, GEFS, etc.); AVHRR SST daily analysis; mixed layer depth and sub-mixed layer thermal stratification from HYCOM + NCODA daily analysis
Output frequency: Every 1 hour
Primary contact: Jonathan Lin, Massachusetts Institute of Technology (MIT)
Model website: https://tcs.mit.edu
Full documentation: Lin et al (2020)

Brief Technical Description

The Forecasts of Hurricanes using Large-Ensemble Outputs (FHLO) model* is a probabilistic tropical cyclone (TC) forecast framework that quantifies the forecast uncertainty of a TC. This is achieved by generating probabilistic forecasts of track, intensity, and wind speed that incorporate the state-dependent uncertainty in the large-scale field. The main goal of FHLO is to provide useful probabilistic forecasts of wind at fixed points in space, but these require large-ensembles (O(1000)) to flesh out the tails of the distributions. FHLO accomplishes this by using a computationally inexpensive framework, which consists of three components: (1) a track model that generates synthetic tracks from the TC tracks of an ensemble numerical weather prediction (NWP) model, (2) a simplified intensity model (FAST - see Emanuel (2017)) that predicts the intensity along each synthetic track, and (3) a TC wind field model that estimates the time-varying two-dimensional surface wind field. The intensity and wind field of a TC evolve as though the TC were embedded in a time-evolving environmental field, which is derived from the forecast fields of ensemble NWP models. See Lin et. al. (2020) for more details.

*The data incorporated herein is generated from the use of the Massachusetts Institute of Technology (MIT)’s Forecasts of Hurricanes Using Large-ensemble Outputs (FHLO) version 1.50, © MIT, used with permission. All Rights Reserved.

Initialization Method

Each ensemble member is initialized from 48 hours prior to initialization time, with (1) the inner-core moisture and (2) a forcing term to the azimuthal wind speed adjusted to keep the intensity as close as possible to the observed value up until the beginning of the forecast. The observed intensities are perturbed using estimated observational uncertainties to generate spread in the initialization period. The potential intensity, vertical wind shear, and mid-level entropy deficit along the observed track are estimated from analysis fields of the operational models over the initialization period.  

FHLO Ensemble Members

FHLO is currently run as a 1000-member ensemble. The thermodynamic and dynamic fields of an ensemble NWP system are used to drive the uncertainty in FHLO. The following list outlines the largest sources of uncertainty treated in FHLO:

  1. Spatial variability from track uncertainty
  2. Internal variability in large-scale fields
    1. mid-level entropy deficit
    2. vertical wind shear
  3. Initial condition/observational uncertainty

FHLO Products Featured on TCGP

TCGP features FHLO graphical outputs for the spatial probability of wind exceedance for a number of different wind speed thresholds. To learn more about the graphical products from FHLO that TCGP provides, please visit TCGP's Guide to the Probabilistic Wind Hazard Plots.

Acknowledgments

FHLO was conceived by Professor Kerry Emanuel. The framework has developed over the past decade at the Massachusetts Institute of Technology by Prof. Emanuel, his graduate students, and collaborators. By 2016, Prof. Emanuel developed the FAST intensity module that is instrumental to helping FHLO run quickly. In 2017, Prof. Emanuel's graduate student, Mr. Jonathan Lin, took up the FHLO mantle as part of his doctoral studies. Mr. Lin developed the code infrastructure, data ingest and processing, numerical techniques, and product generation that have brought FHLO to its current capabilities. Mr. Lin also has undertook comprehensive testing and evaluation to validate and optimize FHLO's accuracy.

FHLO has been funded by various grants over the years. Most recently, FHLO's development has been funded through support from the National Oceanic and Atmospheric Administration (NOAA) grant NA18NWS4680058 of the proposal entitled "New Frameworks for Predicting Extreme Rapid Intensification", $339,571, 01 September 2018 - 31 August 2021.

In 2019, Mr. Lin worked with Dr. Jonathan Vigh to implement real-time runs of FHLO at NCAR during a summer visit to NCAR funded by the Advanced Study Program (ASP) Graduate Visitor Program.

In August 2021, FHLO's probabilistic wind exceedance plots were integrated into the Tropical Cyclone Guidance Project by Dr. Vigh through funding support from the National Center for Atmospheric Research (NCAR) Early Career Faculty Innovators Program (NCAR staff support for the award to Fernando Tormos-Aponte for the proposal entitled "Energy Inequality in the Wake of Disasters: Building Optimal Disaster Resource Allocation Approaches through Assessments of Social Vulnerability", 01 June 2021 - 31 August 2023.

For the Western Pacific, North Indian Ocean, and Southern Hemisphere basins, FHLO uses wind radii forecasts (ATCF TECH id: RVCN) provided by the Joint Typhoon Warning Center (JTWC) through the JTWC Collaboration Site.

References

  1. Emanuel, K., 2017: A fast intensity simulator for tropical cyclone risk analysis. Nat. Hazards, 88, 779-796, https://doi.org/10.1007/s11069-017-2890-7.
  2. Lin, J., K. Emanuel, and J. L. Vigh, 2020: Forecasts of hurricanes using large-ensemble outputs. Wea. Forecast., 35(5), 1713-1731, https://doi.org/10.1175/WAF-D-19-0255.1.

This page was last updated 23 August 2021 by Jonathan Vigh (NCAR/RAL). It has been reviewed by Jonathan Lin.