PREDICTION OF ANNUAL TROPICAL CYCLONE ACTIVITY OVER THE WESTERN NORTH PACIFIC AND THE SOUTH CHINA SEA

This study investigates the relationship between the spatial distribution of sea-surface temperature (SST) and the annual tropical cyclone (TC) activity over the western North Pacific and the South China Sea. Monthly distributions of SST are represented by a set of empirical orthogonal functions (EOFs). The principal components (PCs) of the monthly SST associated with these EOFs are then correlated statistically with the annual number of TCs occurring over (i) the western North Pacific and (ii) the South China Sea using a forward stepwise linear multiple regression scheme. It is found that four PCs of the monthly SST distribution together explain 89 per cent of the total variance of the annual number of TCs over the western North Pacific. Predictions of this number from both the dependent and independent (simulated using the jack-knife method) samples have errors of less than half a standard deviation of the actual number. Two of these PCs are from the September SST, whereas the other two are from May and June SSTs. If only the latter two PCs are used, the accuracies of the prediction are slightly reduced. Therefore, a preliminary prediction can be made using these two PCs and a modified prediction can be performed at the end of September. Two PCs are found to correlate significantly with annual TC activity over the South China Sea, representing the SST distribution in May and July. Predictions using these two PCs also give very reasonable results for both the dependent and independent samples.