影响我国的热带气旋年频数预测

The Annual Frequency Prediction of Tropical Cyclones Affecting China

利用中国气象局《热带气旋年鉴》的热带气旋大风和降水资料集,确定了明显影响我国及华东和华南地区的热带气旋个例,并研制热带气旋年频数的预测方案,使得频数预测对防灾减灾更为实用。预测因子采用相关普查的方法,从1961-2000年前期的海表温度、海平面气压及200,500 hPa和850 hPa位势高度和风场中选出,所用的资料为NOAAER SST和NCEP/NCAR再分析资料。在相关分析的基础上,构建因子时兼顾了因子的系统性的空间结构和时间的变化,并用主成分分析方法去除因子的多重共线性;在最优子集回归建模的基础上进一步对模型进行检验和优化。模型检验和2001—2008年回报试验说明各模型均对各自热带气旋频数(TCF)具有较好的预测能力。

Seasonal prediction schemes are developed for the annual frequency of tropical cyclones（TCs） affecting China,which are more practicable than predicting the genesis frequency for disaster mitigation.The frequencies of TCs affecting the whole China,East China and South China are identified by using the China Meteorological Administration TC-induced wind and precipitation data under one of the three criteria：The storm precipitation heavier than 50 mm has been observed at more than one station in the region;the sustained wind severer than Beaufort scale 7 or wind gusts larger than Beaufort scale 8 has been observed at more than one station in the region;the storm precipitation heavier than 30 mm,and either the sustained wind severer than Beaufort scale 6 or wind gusts larger than Beaufort scale 7 has been observed at more than one station in the region. Seasonal prediction schemes are then developed for these TC frequencies（TCFs） according to their lag correlations with the sea surface temperature（SST） and atmospheric variables during the period of 1961—2000.The NOAA ER SST and the NCEP/NCAR reanalysis data,including sea level pressure,geopotential height at 200,500 hPa and 850 hPa,and both the zonal and meridional components of wind vectors at 200,500 hPa and 850 hPa,are used to derive the predictors.For better representing the variation of the circulation systems in three dimensions,the predictor series are constructed by averaging data within those adjoining significant areas of correlation at various levels in each month.For the frequencies of TCs affecting the South and East China,respectively,analyses on the predictors suggest that their predictors of previous autumn and winter are quite consistent with each other;however,their predictors of previous spring show more differences.For each model,the colinearity among the predictors,including data since 2001,is reduced by applying the Principal Component Analysis approach,and the optimal subset regression model is then developed based on those derived independent predictors.All prediction schemes for TCFs are validated using the data of 2001-2008 and the results indicate that all schemes show skills in predicting frequencies of TCs affecting China though they still can be further improved.