Hurricane forecasters use several different forecast models as guidance for predicting the track and strength of a tropical cyclone. One of the infamously large spreads in the hurricane computer models forecast track occurred with Hurricane Sandy in October 2012. The majority of the models below forecasted Sandy to move from the Bahamas northeastward out into the Atlantic Ocean not directly effecting the United States. However, the European Centre for Medium-Range Weather Forecasts (ECMWF), Consensus (TVCN), and United Kingdom Meteorological (UKX) models forecasted the storm to move northwestward into the Mid-Atlantic United States. Ultimately, these models were correct as Sandy made landfall in New Jersey and brought hurricane-force winds to a large part of the northeastern U.S. Slight differences in initial conditions with a complex system can play a huge role in the track and intensity of the system.
Here is my ranking of the forecast models from best to worst based on the accuracy of each individual model over the past 5 hurricane seasons. Note: The ECMWF European computer model, which is not free for public viewing, is the 2nd most accurate hurricane forecast model only after the TVCN in my opinion.
TVCN (Consensus of the GFS, HWRF, GFDL, EURO, and several other models).
This consensus model is typically the most accurate model as it includes a combination of several useful models. The National Hurricane Center (NHC) usually follows this model closely.
UKX (UKMET aviation tracker).
This computer model has a tendency to slightly under-strengthen tropical systems, but it usually is more accurate with forecasting the track. This model is run by the United Kingdom Meteorological Centre. http://www.metoffice.gov.uk/
AVNO (Aviation Ontime, also known as GFS or Global Forecast System).
This model is typically one of the more accurate hurricane forecast models. The GFS model is usually better at predicting the track of a system compared to its intensity. This model is run by National Centers for Environmental Prediction (NCEP). http://mag.ncep.noaa.gov/
FIM9 (Flow-following finite-volume Icosahedral Model).
This experimental computer model is expected to replace the GFS model in a few years. The FIM9 is a new model, but has proved to be a fairly accurate hurricane forecast model so far. This model is run by the Earth System Research Laboratory (ESRL). http://esrl.noaa.gov/
AEMN (Automated Environmental Monitoring Network, or GFS Ensemble Mean).
This model’s forecast is based on an ensemble forecast, which involves making slight changes to the model’s initial conditions and running the model again. Ensemble forecasts are used to help illustrate forecast uncertainty. The closer the ensemble forecasted tracks are, the higher the forecast confidence. This model is run by National Centers for Environmental Prediction (NCEP). http://mag.ncep.noaa.gov/
HWRF (Hurricane Weather Research and Forecast model).
This hurricane computer model uses GFS model fields for boundary, or initial conditions. This model is scheduled to eventually replace the GFDL model and has performed well lately. This model is run by National Centers for Environmental Prediction (NCEP). http://mag.ncep.noaa.gov/
GFDL (Geophysical Fluid Dynamics Laboratory model).
This hurricane computer model uses GFS model fields for boundary, or initial conditions.
This model is run by National Centers for Environmental Prediction (NCEP). http://mag.ncep.noaa.gov/
CMC (Canadian Meteorological Center model).
This computer model has a tendency to over-strengthen tropical systems. Forecasters use this model to spot areas of potential developing tropical cyclones. This model is run by the Government of Canada. http://weather.gc.ca/model_forecast/global_e.html
BAMM (Beta and Advection Model Medium).
This computer model forecasts the track of a tropical system as if it were a medium (moderate strength) system. This model typically performs better if the system is forecasted to remain moderately strong.
BAMS (Beta and Advection Model Shallow).
This computer model forecasts the track of a tropical system as if it were a shallow (weak) system. This model typically performs better if the system is forecasted to remain weak.
BAMD (Beta and Advection Model Deep).
This computer model forecasts the track of a tropical system as if it were a deep (intense) system. This model typically performs better if the system is forecasted to remain intense.
CLP5 (Climatology-Persistence Model 5-day).
This model’s forecast is based on the persistence of the current motion of the system and also integrates climatological track information. This model has been used since 1972, but does not take into account the actual storm environment.
This computer model uses the storm’s motion over the previous 12 hours to extrapolate or predict its future motion.
Our forecast models are colored so that the brighter color forecast tracks are from the more accurate models. The more the models diverge, the lower the forecast confidence.
Forecast Models Report Card