What are the model designations?

Model designations are given by a 3 or 4 letter abbreviation in the legends of each figure (e.g., AVNO, ECMF, UKX, etc.). An acronym table can be found on the National Hurricane Center's Automated Tropical Cyclone Forecasting web directory for most of the models.

What are interpolated models?

Interpolated models have a model designation that ends in an "I" if it is a model run 6 hours previous to the analysis time on the figure. The old model track and/or intensity forecast is corrected to the new position and estimated intensity. A model designation that ends in a "2" is a model run 12 hours previous to the analysis time on the figure.

Why are there different models for different tropical cyclones?

Some models are run over specific regions or are based on the statistics over a particular ocean basin. Tropical cyclones in different parts of the world will correspondingly have different sets of guidance. Additionally, not all guidance is publicly available in basins outside of the North Atlantic, eastern North Pacific, and central North Pacific.

Why are some models not available after the analysis time?

There are a variety of models that comprise the guidance suite. Some models are simple statistical models that take a very short time to run. Other models are numerical weather prediction models that assimilate observations of the atmosphere and integrate complex equations to arrive at a forecast. Numerical weather prediction models take hours to run on supercomputers, and their track output is typically delayed anywhere from three to six hours after the analysis time, depending on the complexity of the model.

What are model ensembles?

Model ensembles are formed by taking slightly different initial conditions and/or model physics and integrating each of these ensemble members to generate a spread of possible outcomes. Since the atmosphere is a chaotic system, small initial differences can have profound impacts on the tracks and intensities of tropical cyclones at future times. Ensembles are used to estimate uncertainty in a forecast. Ensemble output is typically delayed six to twelve hours after the analysis time.

What do the ensemble graphics show?

The thick, black line is the deterministic track, the thick, red line is the ensemble mean, and the thin, pink lines show the individual ensemble tracks. Based on the ensemble tracks, there are two other pieces of information. First, the shading shows the probability of the tropical cyclone center passing within 150 km of a point over the forecast period. Second, the oval-like shapes give the likely location of the tropical cyclone center in 24-hour increments (24, 48, 72, 96, 120 h), going from the darkest to lightest red. Not all ovals may appear.

What do the trend graphics show?

The deterministic trend graphics show the last couple days of model runs up the most currently available model run (oldest = lightest blue to newest = darkest blue). A subset of model trends, for track and intensity, are shown to gain a sense of how forecasts have changed over the past couple of days.

The ensemble trend graphics are split up into 24-hour forecast increments. At each forecast hour, the ensemble mean track, up to the forecast time, and the ensemble ovals, at the forecast time, are shown. Oldest to newest ensemble runs are shown, going from the lightest to darkest colors. Typically, the ensemble ovals will shrink, going from older to newer ensemble runs, as uncertainty decreases in the forecast tropical cyclone location with decreasing lead time.

What do the verification graphics show?

For track, the bias shows the average forecast error in position for a subset of models at various lead times. For example, a northward bias indicates that (on average) the forecast location of the tropical cyclone has been too far northward at that forecast lead time. Points closer to the origin indicate less bias. The mean absolute error shows the average forecast error in distance, which is always a positive number. Lower values are better.

For intensity, the bias is the average forecast error in the intensity. For example, a negative bias indicates that (on average) the forecast intensity of the tropical cyclone has been too weak at that forecast lead time. A bias closer to zero is better. The mean absolute error is the average of the absolute value of the intensity forecast errors. Lower values are better. Keep in mind that biases and mean absolute errors vary from storm to storm, and depend on sample size (given by the small numbers above each bar in some of the verification graphics).

May I use the figures on this website?

Reproduction is allowed for research and educational purposes. Please give proper credit to Brian Tang (University at Albany).

What are the data sources for the guidance?

Data is provided by the National Oceanic and Atmospheric Administration, the United Kingdom Meterological Office, the European Centre for Medium-Range Weather Forecasts, and the National Center for Atmospheric Research Tropical Cyclone Guidance Project.

Who may I contact if I have questions or comments regarding these products?

This website is maintained by Brian Tang. Questions and comments are appreciated.