A technique called ‘Dynamic Time Warping’ is applied to identify ensemble tracks that are similar to the analysis track. Beside small spatial displacements, this technique also permits small temporal shifts in the development. Such similar tracks are then used to calculate trajectories in cyclone phase space (CPS), providing insights into the cyclone’s thermal structure. A composite approach highlights the differences between the ensemble members that predicted the storm and those that did not. Further separating the developing storms into warm and cold terciles of the upper-level core, composites of environmental variables reveal under which scenario the storms actually undergo TT.
Results show that Chris’ occurrence is predicted by those members that also predict the merging of the two PV maxima, even at lead times of more than four days. The strongest increase in the number of similar tracks is found for the lead time when the model is able to accurately predict the cyclonic wrap-up of the PV streamer. The position of the storm relative to the PV streamer determines whether Chris follows the TT pathway. The transitioning storms are located inside a favorable pouch of high equivalent potential temperatures that result from the cyclonic wrap-up of the PV streamer. The tropical characteristics of the TT-storm cluster can be also inferred from the more symmetric, compact, and intense wind patterns. A systematic investigation of consecutive ensemble forecasts indicates that forecast improvements are linked to specific events, such as the PV merging or the first occurrence of the precursor cyclone in the initialization. Although the results from a single case study may not be generalized, the methods presented open a promising avenue to multi-scale predictablity studies of TTs in all tropical and subtropical oceans and seas.