To be notified of news about this storm and others through the season:

For computer model forecasts, visit our charts page.

What are the differences between tropical models?


Bryan, What are the differences between the tropical models? You seem to list a lot of them on your site. Could you please explain them?

Track Models:

The AVN is the Aviation model and is now a member of the GFS (Global Forecasting System). The AVN is a global baroclinic model and one of NCEP's (National Centers for Environmental Prediction) primary models of choice. The AVN is a spectral model which means it represents atmospheric flow as being a combination of waves (in this case, 126 of them). Included in the AVN are 28 vertical levels and parameterizations for convective, radiative, and boundary layer conditions. To adapt the AVN for tropical cyclones, the model data in initialized with about multi-level synthetic observations from about 50 locations located within approximate 200 nm of the circulation center. The AVN provides track and intensity estimates out to 72 hours in advance.

The NOGAPS model in the Navy's global spectral model. NOGAPS includes 159 waves but only has 18 vertical levels. In general, the NOGAPS operates in a very similar manner to the AVN, just at a slightly lower resolution.

The UKMET is another global baroclinic models and is operated by the UK Meteorological Office. The UKMET is very similar to the NOGAPS and AVN.

The GFDL is a forecast model developed by the Geophysical Fluid Dynamics Laboratory specifically to forecast hurricanes. The GFDL is a baroclinic model that is run over a limited area instead of globally. The model is separated into 18 vertical levels, and has shifting nested grids to follow the motion of the hurricane. The horizontal resolution of the inner grids is 1/6 of a degree. The data used to initialize the GFDL and control conditions on its boundaries are taken from the AVN. The GFDL uses physical parameterizations specialized for hurricane forecasting. As a result of its specialization, the GFDL also provides forecasts for tropical cyclone intensity.

The LBAR stands for limited area sine transform barotropic model. The LBAR is only a two dimensional model and uses shallow-water equations calculated with general tropospheric winds and heights from the AVN to forecast storm movement. One downfall of the LABAR is that it is run with initial and boundary conditions that are actually taken from a 6-hour old model run of the AVN.

The BAM model is known as the Beta and Advection model. It takes trajectory information from the AVN and uses it to provide a track forecast for advection. This advection track is then modified to account for the beta-drift effect that arises as a result of the storm's rotation. The BAM is run with winds averaged over three depths: shallow (850-700 mb), medium (850-400 mb), and deep (850-200 mb). These runs are respectively called the BAMS, BAMM, and BAMD.

The NHC90 and CLIPER are actually track forecast models based on statistics rather than governing dynamics. The CLIPER (Climatology and Persistence) generates a forecast based on the storm's initial position, storm motion speed and direction, the date, and initial storm intensity. The CLIPER forecasts are rarely used a defining predictor of storm motion, but rather is used to shift forecast tracks towards historically favored paths. The NHC90 was developed to modify CLIPER forecasts to account for current circulation patterns output by the AVN.

Intensity Models:

The SHIFOR model stands for Statistical Hurricane Intensity Forecast. As the name suggests, the SHIFOR is a model that simple provides statistical guidance based on climatology and persistence. As a result, the SHIFOR is just used as a reference to see what previous storms in similar conditions have done. One limiting factor to the SHIFOR is that it was developed using a data set that did not include any cases where storms made landfall. As a result, it is only applicable to storms out over the open ocean.

The SHIPS model stands for the Statistical Hurricane Intensity Prediction Scheme. Like the SHIFOR, SHIPS accounts for climatology and persistence, however, SHIPS has the added benefit of taking into account synoptic predictors. Among these predictors are the difference between the current intensity and maximum possible intensity, vertical wind shear, the previous 12-hr intensity change, outflow above the cyclone, and upper level wind and temperature patterns within 1000 km of the circulation center. The SHIPS model uses the forecast positions generated by the LBAR. The model also uses synoptic conditions forecasted by AVN out to 48 hours. Like the SHIFOR, the SHIPS model was developed using cases where the systems did not reach land. This leaves the same limitations experienced when applying the model to systems approaching land.


Posted by Bryan Woods | Permalink