Development and Testing of Statistical Tropical Cyclone Forecasting Techniques for the Southern Hemisphere

Abstract

The Fleet Numerical Oceanography Center (FNOC) Numerical Variational Analysis (NVA) has been used to develop statistical–synoptic tropical cyclone forecast schemes for the Southern Hemisphere. Results, from a limited sample, showed that the best objective techniques were at least as good as the official Joint Typhoon Warning Center (JTWC) forecasts. Development and testing of the aids also indicated that approaches used in the Northern Hemisphere did not necessarily lead to better forecasts. Geographically orientated, rather than track-orientated coordinate systems, were found to have lower errors, especially with operational data. Stratification of the datasets by geographical location did not result in any significant improvements in the forecasting accuracy. Similarly, use of prediction equations based on storm position relative to the subtropical ridge did not result in the improvement found for the Northern Hemisphere. A comparison of techniques using height versus wind field data for the synoptic forcing component of the equations showed lowest errors with the mass field predictors. However, the differences were not statistically significant and this finding may have been a result of data quality rather than the field type.