Prospects for Improving the Operational Seasonal Prediction of Tropical Cyclone Activity in the Southern Hemisphere - Monitoring and Prediction of Tropical Cyclones in the Indian Ocean and Climate Change - page 123

Fig. 2: The correlation between annual number of TCs in the Australian region and
the monthly 5VAR, NIÑO3.4 and SOI indices.
In agreement with earlier studies, the best correlation is found for the months
from August, A( t ), to January of the next year, J( t +1). While the 5VAR and
NIÑO3.4 indices demonstrate high correlation in a range from -0.6 to -0.7 for
six months from A( t ) to J( t +1), the SOI has the strongest correlation in A( t )
(-0.62) which then decreases to less than -0.5 from October, O( t ), onwards. It
appears that the state of the atmosphere alone (as described by the SOI) is an
important contributor to the environment in which TCs form early in the season,
but not as important as ocean (or combined contribution of ocean and the
atmosphere as described for example by 5VAR) during the TC season.
The correlation coefficients were also computed for two-month and three-
month averages (Figs 3 and 4, respectively). Similar conclusions can be drawn
from the analysis of the correlation of the TC number with bi- and tri-monthly
averaged ENSO indices. The highest correlation was found for the averages
that include A( t ) up to those that include J( t +1) (although still with the exception
of the SOI).
A TC seasonal forecast for the regions of the Southern Hemisphere is
typically issued in October, prior to the beginning of the TC season. As monthly-
averaged values of the ENSO indices are usually available in the second week
of the following month, for the prediction model to be implemented
operationally the values for the months beyond September cannot be used.
Consequently, in Table 1, we present the highest correlation coefficient of the
total annual number of TCs in the Australian region with the three ENSO indices
examined and the corresponding months that have been averaged only for
months prior to October. Note that the highest correlations for the entire
24-month period examined are only marginally higher for the months beyond
September. For example, in the case of the 2-month average for the 5VAR
index, the strongest correlation of -0.678 arises from the average of the
September and October values, while the correlation arising from the average
of August and September values (presented in the Table 1) is -0.665.