봄철 대기 변수들을 이용한 여름철 북서태평양 열대저기압 발생 빈도의 통계적 계절예측

Abstract

In this study, two statistical models were developed to predict the typhoon genesis frequency from May to December over the western North Pacific using lagged correlation analysis of the relationship between the typhoon genesis frequency and several environmental parameters for the study period, 1975-2010. Moreover spring atmospheric variables are also used for the model which is based on the least absolute deviation (LAD) regression method. Through lagged correlation analysis and stepwise regression, four parameters (200 hPa U wind component, 850 hPa V wind component, North Atlantic Oscillation index, Sea Level Pressure) are chosen as a final predictors. Results from cross-validation suggest that two statistical models are skillful in seasonal predicting typhoon genesis frequency with a correlation coefficient of 0.76 during study period. UV index are defined using two atmospheric variables (200 hPa U wind component, 850 hPa V component). This study shows that typhoon genesis frequency has positive relation with UV index. In a positive UV index phase, North Pacific anticyclone is weaken and moved to north-eastwards, as a result Pacific Ocean has negative geopotential height anomaly and Sea Surface Temperature and solar radiation flux of East Pacific has positive anomaly. Consequently, weaken of Walker Circulation due to the decrease of temperature gradient effects on typhoon genesis frequency.