DARMA模型在中国日降水量随机模拟中的适用性研究

为提高日降水量随机模拟精度，研究了DARMA模型模拟中国日降水量的普适性。采用中国大陆811个气象站点的日降水系列，建立DARMA（1，1）模型以模拟日降水事件的发生，进行自相关系数、干湿游程概率分布检验，根据Gamma函数随机生成不同湿游程序列日降水量，与DAR（1）模型对比。研究结果表明：中国超过50%的降水为2日及以上的多日降水事件，自相关系数衰减缓慢，符合DARMA模型特征；模拟产生的干湿游程概率分布精度优于DAR模型，优势随着降水量增加更显著，模拟序列统计特征值除均值外均优于DAR模型。DARMA模型适用于中国大陆地区日降水量的随机模拟，是一种可行的日降水量随机模拟模型。

Applicability of DARMA model in stochastic simulation of daily precipitation in China

The universality of DARMA model in simulating daily precipitation in China is studied to improve the random simulation accuracy of daily precipitation.In this paper,the DARMA (1,1) model is established to simulate the occurrence of daily precipitation events by using data series selected from 811 meteorological stations in Chinese mainland.The autocorrelation coefficient and the dry and wet run probability distribution are tested.The daily precipitation of different wet and dry run is generated randomly according to the Gamma function,and compared with the DAR (1) model.The results show that more than 50% of the precipitation in China is multi-day precipitation events of 2 days and above with slow decay autocorrelation coefficient,conforming to the characteristics of DARMA model.The accuracy of dry wet run probability distribution generated by simulation is better than that of DAR model,and the advantage is more significant with the increase of precipitation.Except for the mean value,the statistical eigenvalues of simulation series are better than DAR model.The DARMA model is applicable to the stochastic simulation of daily precipitation in Chinese mainland,and is a feasible stochastic simulation model for daily precipitation.