Abstract
The interannual variability of tropical cyclone (TC) movement has been found to be rather significant in certain regions of the western North Pacific. Westward and northwestward moving TCs are found to occur mostly at low latitudes and have the largest interannual variations in the region just east of the Philippines and over the northern part of the South China Sea. The area east of the Ryukyu Islands and south of Japan is identified as having the largest interannual variation of northward moving TCs.
Correlations are made between the annual number of occurrences of TCs in these prescribed regions and the principal components of the monthly mean 850‐ and 500‐hPa zonal wind patterns over the western North Pacific (for the months of November of the previous year to April of the current year). Westward‐moving cyclones are found to correlate well with the 850‐hPa zonal wind patterns in January and March. Using the principal components associated with these patterns, prediction equations are then developed using the total (dependent) sample and the jackknife method (simulating an independent sample). The predictions made with both the dependent and “independent” samples are found to be very good.
For north‐westward moving cyclones, no prediction equation is developed because principal components of the 850‐ and 500‐hPa zonal winds found to be significant can explain less than 30 per cent only of the total variance. However, the February 500‐hPa zonal winds correlate well with northward‐moving cyclones in the region east of the Ryukyu Islands and south of Japan. Predictions made using the principal components associated with this flow pattern for both the dependent and “independent” samples give rather good results.
Because all the predictors are between January and March, they can be used operationally to predict the annual number of occurrences of TCs in the predefined regions. As TCs in these regions are likely either to move into the South China Sea or affect Japan, these results should prove useful in the seasonal prediction of TC occurrence in these latter areas.