Joint Hurricane Testbed
JHT Home
Terms of Reference (PDF)
Steering Committee
Main Activities
Current Projects
Past Projects
Administrative Presentations and Information

Goals 2003-2005

Upgrades to the operational GFDL hurricane prediction system: ($90,000 per year)
  • Test GFS surface physics package
  • Modify downdraft formulation in SAS scheme.
  • Improve momentum mixing in SAS scheme.
  • Include effect of evaporation in large-scale condensation.
  • Modify land-sea mask for consistency between atmospheric and ocean models.
  • Import and test latest version of Lin microphysics.

Improving the GFDL/URI coupled hurricane-ocean model: ($135,700 - year 1; $142,900 - year 2)
  • Improve ocean component of the coupled model in the Atlantic basin (include explicit init of Loop Current using altimeter data).
  • Implement ocean coupling in eastern Pacific basin (reduce positive intensity forecast biases).
  • Evaluate and transfer to operations a new high-resolution version of the coupled model (5 nm spacing inner mesh).
  • Test and implement new air-sea flux parameterizations.

Hurricane model transitions to operations at NCEP/EMC: ($165,000 per year)
  • Test GFS shallow convection and surface parameterizations in GFDL model.
  • Test bulk microphysics packages.
  • Run parallel version with NOAH LSM in 2003 season.
  • Test WRF physics packages.
  • Test movable nest HWRF.
  • Evaluate Hurricane WRF prototype, compare to GFDL w/WRF physics packages, run the prototype operationally in 2005?

Evaluation of upper ocean mixing parameterizations: ($93,000 - year 1; $96,900 - year 2)
  • Use primitive equation ocean model to simulate upper ocean response to three TCs (Gilbert, Isidore, Lili) using realistic forcing and several ocean vertical mixing models.
  • Compare simulated upper ocean thermal and current structure to data from AXBTs, conductivity, temperature and depth probes (AXCTDs), and current profilers (AXCPs).
  • Identify mixing models that produce most realistic simulations; provide best vertical mixing parameterization to NCEP for use in coupled models.

Real-time dissemination of hurricane wind fields determined from airborne doppler: ($82,300 - year 1; $100,900 - year 2)
  • Implement existing Doppler radar wind analysis scheme onboard NOAA recon aircraft.
  • Transmit analysis from aircraft to TPC and into H*Wind.
  • Preliminary development and testing of superobs on the aircraft for eventual assimilation into models.

Targeting strategies to improve hurricane track forecasts: ($122,600 - year 1; $127,400 - year 2)
  • Develop objective strategy: Ensemble Transform Kalman Filter (ETKF), used in winter storms, using operational ensemble forecasts.
  • Use present subjective targeting method operationally.
  • Identify cases in which ETKF selects targets similar to subjective method; re-run models to simulate ETKF being used operationally to produce sensitivity maps.
  • Refine ETKF software (incl. merge with flight track drawing under development) and procedures for transition to operations.
  • Operationally produce ETKF products that select best feasible flight tracks.

Objective and automated assessment of operational global forecast model predictions of tropical cyclone formation and life cycle: ($58,200 - year 1; $62,200 - year 2)
  • Provide a tool for development of probabilistic measures of guidance accuracy among global model products related to TC formation and life cycle.
  • Discriminate between circulations in models that may develop as predicted and those that are false alarms.
  • Automatically and objectively identify and evaluate TC precursor circulations, and their physical characteristics, in global analysis and model forecast fields.
  • Summarize and catalogue model performance on a circulation for comparison with future circulations.
  • Provide web interface for operational assessment of models relevant to TC formation.

An updated baseline for track forecast skill through five days for the Atlantic and Northeastern and Northwestern Pacific basins: ($19,000)
  • Update CLIPER track models.
  • Incorporate recent best track data used to derive the models.
  • Improve calculation of current storm motion.

Quantifying tropical cyclone track forecast uncertainty and improving extended-range tropical cyclone track forecasts using an ensemble of dynamical models: ($60,000)
  • Apply “CONU” approach to Atlantic and eastern Pacific basins to potentially include GFDN (FNMOC), COAMPS (FNMOC), and MM5 (AFWA) along with GUNA components.
  • Determine optimum mix of models at different lead times.
  • Perform statistical evaluations of consensus forecast errors, stratified by initial quantities known at time of forecast (e.g., spread, number of models available to consensus, and TC initial intensity, forward motion, and location).
  • Operationally provide tabular and graphical displays of uncertainty in each consensus forecast.
  • Validate uncertainty graphic using cross-validation techniques on 2000-02/3 database on a storm-by-storm basis.
  • Provide statistical evaluation software to be used to update uncertainty database at the end of each season, to incorporate changes to individual models.

Transition of revised dynamical model track prediction evaluation expert system (DYMES): ($80,000)
  • Create revised DYMES that implements the Atlantic error mechanisms determined with previous funding.
  • Demonstrate operational utility during real-time tests.

Implementation of the Advanced Objective Dvorak Technique (AODT) and Tropical Cyclone Intensity Estimation (TIE) algorithms at TPC: ($95,400 - year 1; $99,700 - year 2)
  • Implement latest version of AODT in NAWIPS, applicable to all stages of TC, non-McIDAS, can be automated.
  • Implement rudimentary version of TIE model in NAWIPS: fully objective regression between recon pressure and IR predictors, SST, and latitude of the TC; additional predictors to be added (e.g., synoptic environment, microwave imagery).
  • Integrate AODT and TIE into single code package.

Improvements in Deterministic and Probabilistic Tropical Cyclone Surface Wind Predictions: ($95,000 per year)
  • Develop and evaluate new component to SHIPS model utilizing recon and GOES IR data to better determine inner core structure.
  • Test neural network prediction method in SHIPS to account for nonlinear interactions between predictors.
  • Provide fields of the probability of surface wind speeds exceeding specified thresholds over certain time intervals.

Development of a rapid intensification index for the eastern Pacific basin: ($54,200 - year 1; $64,100 - year 2)
  • Develop RI index for the eastern Pacific using methodology similar to that for existing Atlantic index.
  • Output RI probability estimate with each operational eastern Pacific SHIPS forecast.
  • Modify existing code and provide training so that it can be used by TPC to re-derive the eastern Pacific index for future seasons.

Developing an inner-core SST cooling predictor for use in SHIPS: ($79,500 - year 1; $71,300 - year 2)
  • Test algorithm for use in SHIPS that relates the estimated amount of SST cooling directly under the storm to subsequent TC intensity change.
  • Extend algorithm, based on a developmental 23-hurricane sample (1975-2002) of observations (AXBTs, buoys), to weaker systems (by 2005).
  • To run operationally, predict (rather than observe) inner-core SST cooling using regression with ambient SST and storm location/speed/intensity.
  • Improve upon weekly Reynolds SSTs that provide initialization; higher spatial and temporal resolution.
  • Test additional predictors related to subsurface upper ocean parameters.

Improving the validation and prediction of tropical cyclone rainfall: ($120,000 - year 1; $127,500 - year 2)
  • Develop new rainfall validation schemes that provide a baseline of comparison for different forecast systems.
  • Provide new forecasting tool, based on the benchmark R-CLIPER, that incorporates information related to vertical shear and storm track, and can be run operationally.
  • Generate error statistics for rainfall forecasts using traditional and new validation techniques for the operational GFDL, Eta, and GFS models, and the R-CLIPER and modified R-CLIPER.

Return to NHC HomeContact NHC