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 122

A linear regression model technique was used to model the relationship
between the number of cyclones in three regions of the Southern Hemisphere.
Studies by Ramsay et al. (2008) and Kuleshov et al. (2009) demonstrated a
strong correlation (about -0.7) between the annual number of TCs in the
Australian region and the August-September-October-averaged NIÑO4 and
NIÑO3.4 indices, with some other ENSO indices also showing high correlation.
For the eastern South Pacific Ocean, the NIÑO3.4, SOI and 5VAR indices
correlated with the TC number better than other ENSO indices (Kuleshov et
al., 2009).
The NIÑO3.4 and the SOI are the two most commonly used indices in
defining ENSO phases. The SOI data used in this study were obtained from
the Australian Bureau of Meteorology and are available on its website at
www.bom.gov.au/climate/current/soihtm1.shtml. Values for the NIÑO3.4 (SST
anomalies in Niño3.4 region, 3-month running mean) were obtained from the
Climate Prediction Center, NOAA (ftp.cpc.ncep.noaa.gov/wd52dg/data/indices/
sstoi.indices). A multivariate ENSO index, based on the first principal
component of monthly Darwin mean sea level pressure (MSLP), Tahiti MSLP,
and the NIÑO3, NIÑO3.4 and NIÑO4 SST indices, has been developed at the
NCC (see also Kuleshov et al., 2008, 2009). Its base period is 1950-1999.
Strength of a multivariate index is in integrating both atmospheric and oceanic
responses to changes in the ENSO phases in one index. We denote this
standardised monthly anomaly index as the 5VAR index. In this study, the
NIÑO3.4, the SOI and the 5VAR indices were used for further investigation of
the TC-ENSO relationship.
3. Results
3.1 Correlation between the Annual Number of
TCs and the ENSO Indices
The correlation coefficient was calculated between the annual number of TCs
and the three selected indices for each month of two consecutive years in which
the TC season is included. Thus, we measure a degree of correlation between
the number of cyclones in the TC season with the indices before, during and
after the TC season. For each index, there would be twenty-four correlation
coefficients for individual months denoted as January( t ), February( t ), …,
December( t ), January( t +1), … and December( t +1), where t is the year in which
the cyclone season starts. We also investigated the correlation by averaging
values of the ENSO indices for m neighbouring months, where m takes values
2 and 3. The higher the values of m , the less importance is given to individual
monthly values of the index as equal weights are given to each of the m months.
The correlation between annual number of TCs in the Australian region and
the monthly 5VAR, NIÑO3.4 and SOI indices is presented in Fig. 2 (for the
SOI, correlations with -SOI are plotted for consistency of sign with the other
indices).