Tropical cyclones are observed to vary on intra-seasonal time scales (Fig. 5.8). There is potential predictability here, especially if the intra-seasonal variability is associated with low-frequency tropical circulation anomalies. A few Atlantic studies have been carried out by Ray (1935), Namias (1955), Ballensweig (1959), and more recently by Shapiro (1987). However, most of the predictive signals have not been verified over a long period record and it is premature to make any definitive statements on prediction methods.
Figure 5.8: Temporal clustering of tropical cyclone genesis in the Northern Hemisphere. Numbers in shaded (unshaded) areas give the tropical cyclone formation during active (inactive) periods. Numbers between arrows indicate cyclones before and after the main summer period (Gray, 1979).
Good intra-seasonal forecast prospects may develop as new data sets become available and as we learn more of the physical processes of the tropical cyclone and its environmental interactions. This appears to be a profitable area of future research, a few examples of potential intra-seasonal predictability signals are discussed in the following sections.
Figure 5.9: Temporal clustering of tropical cyclone genesis in the Southern Hemisphere. Numbers in shaded (unshaded) areas give the tropical cyclone formation during active (inactive) periods. Numbers between arrows indicate cyclones before and after the main summer period (Gray, 1979).
Tropical cyclones tend to cluster in time (Figs. 5.8 and 5.9). One can occasionally observe 10-20 cyclones about the globe within several weeks and such active periods typically are separated by an equivalent period of much reduced cyclone activity. An obvious alternation of active and inactive periods of 15-25 d duration is evident in all years in both Figs. 5.8 and 5.9. Figure 5.10 portrays the annual cycle of global tropical cyclone formation for a recent 7-year period (1978-1985). Also shown are the monthly number of named tropical cyclones occurring in active and inactive periods. The 3-5:1 ratio of cyclone numbers between active and inactive periods seen in the hemispheric observations is maintained throughout the year.
This obvious 30-50 d cycle in tropical cyclone activity implies that there may be some connection with the Madden-Julian Oscillation (MJO), especially since the variability named tropical cyclone activity is primarily confined to tropical cyclone activity at latitudes equatorwards of 20o lat. This relationship has not been properly investigated at this stage, however.
Figure 5.10: Annual variation in global tropical cyclone formation, together with the average numbers forming during active and inactive periods. The data used were all named tropical cyclones for the period 1978-1985.
Good forecast practice would be to keep a log of both tropical cyclone activity, on a global basis, and also to keep in mind the current phase of the MJO. This should provide qualitative indications of the potential degree of activity over the next few weeks, and such statistics can ultimately be used to develop objective forecast techniques.
Joseph (1989), and Joseph and Liebmann (1991) have noted an inverse association between Indian monsoon rainfall and western North Pacific tropical cyclone activity. When the Indian monsoon is delayed or is in a mid-summer monsoon break period, they have observed increased numbers of tropical cyclones in the western and northern portion of the North Pacific, and these cyclones tend to follow a more poleward track. Eastern North Pacific cyclone tracks are also affected similarly. In contrast, when India has an early monsoon onset or enhanced monsoon rainfall mounts western North Pacific tropical cyclone activity is reduced, the systems tend to track more westwards, and they are confined more to the western part of the basin. An awareness of this relationship can provide forecasters with a qualitative indication of potential future cyclone frequency and track characteristics.
Recent research by Harr and Elsberry (1991) and by Chan (1991) is showing the benefits of using EOF (Empirical Orthogonal Function) analysis of upper level winds at various pressure levels and of SST in specifying cyclone activity from 1-4 weeks into the future. Their techniques also provides an indication of the likely tracks of future cyclones. Large-scale flow patterns in the tropics often persist for periods of a week to a month. Such EOF analyses, which can be routinely made at numerical analyses centres, appear to offer promise for the development of new intra-seasonal forecast schemes. Additional implementation and testing is required, however.