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 120

reliable results when values of the environmental indices are outside of the
range of historical records.
The availability of satellite imagery has significantly improved our
knowledge of TCs, with satellite remote sensing being vital for accurate
estimates of parameters such as TC position (e.g., the location of minimum
atmospheric pressure) and TC intensity; however with the latter to a lesser
degree of confidence compared to estimating TC position (Kuleshov et al.,
2010). Satellite images are used by forecasters for preparing operational (real-
time) and best-track data, and a complete digital Geostationary Meteorological
Satellite (GMS) archive for the Southern Hemisphere has been prepared at the
Australian Bureau of Meteorology for use in TC reanalysis (Broomhall et al . ,
2010). Thus, TC historical records for the Southern Hemisphere, at least in
terms of the annual number of cyclone occurrences, are of high quality for the
“satellite era” - that is from early 1970s (Holland, 1984, Kuleshov et al., 2008,
2010).
Utilising historical data for the Australian region, Nicholls (1979) examined
interannual variability in TC activity, and demonstrated a link between ENSO
and inter-seasonal variations in TC numbers. The TC-ENSO relationship was
used in developing statistical methodology for forecasting seasonal TC activity
in the Australian and some other regions in subsequent studies (e.g., Nicholls,
1992; Kuleshov et al., 2009; Liu and Chan, 2011). In general, such a
methodology of statistical seasonal forecasting of seasonal cyclone numbers
ahead of the season (November to April in the Southern Hemisphere) employs
ENSO indices (e.g., the SOI which describes the state of the atmospheric
circulations, or the NIÑO4 and NIÑO3.4 SST anomaly indices) for months
which precede the TC season (e.g., a three-month average for August, September
and October). These models performed reasonably well over past years;
however, during the 2010-11 TC season, which corresponded to a very strong
La Niña event, the statistical models significantly over-predicted the number
of TCs in the Australian region.
The 2010-11 Australian region cyclone season was actually a near-average
tropical cyclone season, with eleven tropical cyclones forming compared to an
average of 12. However, the seasonal forecast issued by the Bureau of
Meteorology's National Climate Centre (NCC) ahead of the season for the
Australian region (the area south of the equator, 90°E to 160°E) predicted on
the basis of strong La Niña conditions that the basin could turn into the most
active season since 1983-84, with 20-22 tropical cyclones developing in or
moving into the region (NCC, 2010). Similarly, the Guy Carpenter Asia-Pacific
Climate Impact Centre (GCACIC) at the City University of Hong Kong has
issued a forecast that predicted that 19 TCs would either develop within or
move into the basin (GCACIC, 2010).
Thus, a motivation for this study was to investigate prospects for improving
the skill of operational seasonal prediction of TC activity in the regions of the
Southern Hemisphere using statistical model-based approaches. In respect of
this, the new best track TC database for the Southern Hemisphere described in