This section focuses on tropical cyclone (TC) motion with an emphasis on the prediction aspects. Such an emphasis reflects the surge in the development of track prediction techniques in the last few years. In Topic 3.1, a summary of the current status of numerical and statistical model guidance to the prediction of TC motion is discussed with a proposal of how to increase the accuracy further. The idea of combining different numerical forecast guidance is described in detail in Topic 3.2, which has known as the consensus approach to track forecasting. The possibility of applying ensemble forecasting techniques in track prediction has received much attention in recent years and the progress of such an application is reviewed in Topic 3.3. Only a few papers have been published in the theory of TC motion during the past few years and the main results are discussed in section 3.4.

3.0.1. Numerical and statistical guidance

Because of the rapid increase in computer power, many new ideas in numerical prediction (data assimilation methods, vortex specification) can be incorporated in the numerical models. Together with an increase in model resolution and better physics, very significant progress has been made in the accuracy of track prediction. However, systematic errors can be identified in some models and statistical post-processing techniques have been proposed to correct for such errors. In addition, major forecast failures continue to occur especially when the track is erratic. Thus, detailed evaluations of model behaviour become an important issue to be addressed. Further research into theoretical understanding of the cause of erratic tracks, oscillatory motion and interaction with neighboring mesoscale and sub-synoptic-scale systems is also proposed. On the operational side, many suggestions and requests are also listed for consideration by major weather centres.

3.0.2. Consensus approach to track forecasting

Because every numerical model has its own merits and weaknesses, it has been proposed that all the model predictions be combined to give a “consensus” forecast, which should theoretically be better than any of the individual forecasts. For this reason, the Joint Typhoon Warning Center has developed such a system that has been in operation for three years and resulted in a decrease in the overall forecast error. Diagnoses of model behaviour have indicated that some model forecasts have large systematic biases under certain synoptic situations. If one can identify such “problems” so that a specific model forecast is removed before all the model forecasts are combined, the forecasts may be even better. However, such a “selective consensus” turned out to be of little additional value to the forecasters. In addition, even the simple consensus forecast may sometimes not be appropriate, e.g. when the tropical cyclone is not well analyzed in the model or when the model produces erroneous vortices or other systems in the vicinity of a tropical cyclone. Detailed documentation of such problems could lead to further improvements in such an approach.

3.0.3. Ensemble prediction methods for tropical cyclone forecasting

The consensus approach in a sense is similar to the idea of ensemble prediction. In the latter, the initial analyses are perturbed to simulate the error distribution associated with errors in the observations and analysis methods. The assumption is that if such a simulation is correct, the envelope of forecasts resulting from each of the perturbed analyses would encompass the true behaviour of the atmosphere. Even though such a concept has been applied in major weather forecast centres for daily weather forecasts, its use in the prediction of tropical cyclone motion only began a few years ago. Various perturbation techniques have been tested using both barotropic and baroclinic models, with varying results. A “super-ensemble” approach has also been proposed in which the model behaviour during a certain period is used as a basis for prediction in a latter period. Because ensemble prediction for tropical cyclones is still in its infancy, much more research is needed in this area in terms of perturbation methodologies, identification of model errors, development of skill evaluation techniques etc.

3.0.4. Theoretical advancements

The amount of research in the theory of TC motion has decreased sharply in the last few years and the major focus has been on applying the concept of potential vorticity tendency to explain the baroclinic behaviour of TC motion. Both numerical and observational studies have shown that a TC tends to move towards an area of maximum azimuthal wavenumber-1 potential vorticity tendency, which is mainly contributed by the horizontal advection of potential vorticity and (asymmetric) diabatic heating. This approach can apparently explain many types of track behaviour, including track oscillations and some erratic tracks. However, further verification of such a theory is necessary and research into the theory of motion when the TC is interacting with adjacent features (land, terrain, mesoscale convective systems, sea-surface temperature anomalies, etc) should also be carried out.