摘要
文中建立了基于多状态一阶Markov链的逐日降水量随机模式并结合广义帕雷托分布(GPD)产生夏季逐日极端降水量的模拟资料,结果所显示的各种气候特征表明,绝大多数站点(尤其是中国东部多雨地区)都达到较高的精度。分析表明,该模式对中国东部极端降水特征的模拟能力在某些方面优于两状态一阶Markov链模式。对东部6个代表站模拟试验结果表明,月降水均方差、日降水极大值、月平均降水日数、日降水均方差、日平均降水量等指标与实况比较,均证明该模式对逐日降水量的模拟效果较好,基本模拟出降水量的各种特征。对中国东部78个代表站采用的两种模式模拟结果对比发现,除日平均降水量以外,月平均降水日数、日降水平均极大值都与实际观测结果较为一致,总体上优于两状态模式,说明用该模式在全国范围内模拟逐日降水特征尤其是极端降水特征有较高的可行性。例如,由其中6个代表站模拟资料所拟合的极端降水GPD模式具有较高的拟合优度。无论从门限值或重现期值来看都可发现模拟与实测结果有较好的相似性,且两者门限值的误差越小,重现期极值的差距也越小。证明Markov链模式对极端降水的模拟有广泛的适用性。
A multi-status Markov chain model is proposed to simulate daily rainfaIls records and the extreme rainfalls in summer are fitted from the simulated daily rainfall records by using a Generalization Pareto Distribution （GPD） model. The research results show that this statistical simulation method display most good dependability and the statistic climatic features from the simulated daily rainfall records have reached more high precision for most stations, specially the pluvial region over east area of China. The analysis comparatively shows that the multi-status Markov chain model is excellent with the two status Markov chain model for its simulation ability, specially, the simulated climatic features of extreme rainfalls in east region of China have reached most high precision. The experiment results for the selected six stations demonstrate the excellent simulation effects of above statistical model, for example, the following features： statandard deviation of monthly rainfall, maxmum daily rainfall, mean number of rain days during a month spell, statandard deviation of daily rainfall, mean value of daily rainfall amount etc. are consistent with the real obvervation values. Both the multi-status Markov chain model and the two status Markov chain model.
出处
《气象学报》
CAS
CSCD
北大核心
2009年第1期20-27,共8页
Acta Meteorologica Sinica
基金
国家自然科学基金项目(40675043)。
