几种模拟逐日降水的分布函数比较分析  

Comparative Analysis on Three Distribution Functions Simulating Daily Rainfall

作　　者：王斌[1,2] 付强[1,2] 王敏[1,2] 雷呈瑞 

机构地区：[1]东北农业大学水利与建筑学院,黑龙江哈尔滨150030 [2]黑龙江省高校节水农业重点实验室,黑龙江哈尔滨150030 [3]哈尔滨市气象台,黑龙江哈尔滨150080

出　　处：《数学的实践与认识》2011年第9期128-133,共6页Mathematics in Practice and Theory

基　　金：国家自然科学基金(50309003);黑龙江省博士后资助经费(LRB10-173);东北农业大学博士科研启动基金

摘　　要：在模拟逐日降水方面,将马尔科夫链和某种分布函数相结合建立随机模型的方法在国外很多地区被证明是有效的,但该方法在我国的适用性研究并不多见.本文应用哈尔滨48年的实测降水资料,基于一阶马尔科夫两状态转移概率,分别采用皮尔逊Ⅲ型分布、伽玛分布和偏正态分布函数模拟了哈尔滨的1000年逐日降水过程.对比分析结果表明,伽玛分布更适合模拟哈尔滨的逐日降水过程,基于伽玛分布函数模拟的月份降水量和降水天数与实测降水数据符合较好.In simulation daily rainfall, a combination of Markov chain and some distribution function has been recognized as a popular method and demonstrated to be effective for large range of environment abroad.However,this method and its application are rarely seen in China. According to the same Markov transitional probabilities, P-III distribution, gamma distribution and skewed normal distribution were used to simulate the daily rainfall process for 1000 years based on 48-year daily rainfall data of Harbin. The results show that gamma distribution can be used to generate daily rainfall data in Harbin and its accuracy is high, while P-III distribution and skewed normal distribution are not applicable in Harbin area.

关 键 词：降水模拟 转移概率 P—III型分布 伽玛分布 偏正态分布 

分 类 号：P426.613[天文地球—大气科学及气象学]