An Extended-range Forecast Study Of Tropical Cyclone Generation And Track In The Northwest Pacific

Tropical cyclones(TCs)pose a major threat to life and property in the coastal countries and regions.Forewarning of TC occurrence by developing an effective and seamless forecast,which is the major content and difficulty of current TC operational forecast,can be of great benefit for emergency preparedness and loss minimization.Sub-seasonal prediction bridges the gap between the short/medium range and seasonal predictions,but the current forecast skill of TC activity at the sub-seasonal time scale is still limited.Studying the statistical and physical relationships between TC activity and factors are beneficial to deepen the understanding of TC activity and improve the limited forecasting skill at sub-seasonal time.By analyzing the combined effect of physical factors on TC activity at interannual and sub-seasonal time scales,and the climatological seasonal cycle of TC activity,a statistical model was developed in this paper to predict TC genesis and occurrence over the western North Pacific(WNP)based on logistic regression.The skills of statistical model to predict the TC genesis and track over the WNP were evaluated,which was compared to dynamical model.The performance of dynamical model to simulate the TC activity at the climatic,interannual and sub-seasonal time scales on prediction skills was analyzed,and then revealed the main factors that affecting the performance of dynamic model on TC activity prediction at sub-seasonal.The main results are as follows:(1)Reproducing the climatological seasonal cycle of TC activity is the key to implement a successful sub-seasonal forecast.The statistical model has a better performance than reference forecast on TC prediction,which is ascribe to the constructed factors can well represent the climatology of TC activity.(2)The effect of ENSO and intra-seasonal oscillations on TC genesis and track forecast skills over the WNP are regionally dependent.On the regional scale,ENSO is the important predictor over the eastern WNP and the western WNP,and then MJO and QBWO are the important predictors over the western WNP and the South China Sea.(3)A statistical model,which can provide more skillful forecasts than the current dynamic models,was developed to predict the probability of TC genesis and track for the WNP at sub-seasonal time scale.In 2-7(2-5)week forecast lead,the statistical model skill in predicting TC genesis(track)is higher than dynamic model.(4)The main factors affecting the performance of dynamic model to predict TC genesis and track at sub-seasonal were clarified.Most of dynamic models have a better skill in TC track prediction than cyclogenesis at sub-seasonal time scale,which reflects the underprediction of TC intensity in models.The lower skill of cyclogenesis restricts the improvement of track prediction.The better simulation of climatic and interannual cyclogenesis may be conducive to the improvement in track prediction of dynamic models.