Tropical cyclone activity in the western North Pacific has a large social and economical impact in the countries of the region. The probability of landfall depends on a cyclone's trajectory. While it is well known that these trajectories vary strongly with season, and tend to be affected by ENSO, the probabilistic behavior of tropical cyclone trajectories needs to be better understood in order to isolate potentially predictable aspects. In this analysis, we apply a new clustering technique to a large observational dataset of cyclone tracks, to identify different types of track, their seasonality, and their relationship to the large-scale circulation and ENSO.

The clustering technique consists of building a mixture of polinomial regression models (i.e. curves), which are used to fit the geographical "shape" of the trajectories. The finite mixture model allows the clustering to be posed in a rigorous probabilistic framework, and to easily accommodate tropical cyclone tracks of different lengths, giving advantages over the K-means method used in previous studies.

The new cluster analysis is applied to the best track dataset for the period 1950-2002. The dependence of the track clusters on season and ENSO phases are analysed. Depending on the ENSO phase and season, we find that different trajectory types have a higher incidence, and that these are associated with different preferential regions of landfall. These preferential trajectories will be discussed. In most cases the ideal number of clusters seems to be around 4 or 5. The two main trajectory-types identified correspond to straight-movers and recurvers, with the additional clusters corresponding to more detailed differences among these two main types, based on the location and trajectories of the tracks. The characteristics of the tropical cyclones in each cluster type are studied, including first position, mean track, landfall, intensity, lifetime and speed. Large-scale circulation anomalies composites for different track clusters will also be analyzed.