加权马尔可夫链在铜川地区降水量预测中的应用  

Predicting Precipitation in Tongchuan Using Weighting Markov Chain Model

作　　者：李亚斌[1,2] 徐盼盼[1,2] 钱会[1,2] 王海科[1,2] 

机构地区：[1]长安大学环境科学与工程学院,西安710054 [2]长安大学旱区地下水文与生态效应教育部重点实验室,西安710054

出　　处：《灌溉排水学报》2017年第5期96-102,共7页Journal of Irrigation and Drainage

基　　金：水利部公益性行业科研专项经费项目(201301084)

摘　　要：为了全面分析铜川地区降水的时间分布特征,为该区域水资源预测提供依据,基于铜川市1960—2013年年降水量资料,采用样本均值-均方差分组方法,确定了其分级标准及状态,并根据马尔可夫理论,运用统计学原理,验证了该降水序列具有马尔可夫性,建立了适用于该地区的加权马尔可夫链对降水量的预测模型,并根据已有资料验证了预测结果的真实有效性;再结合模糊集理论的级别特征值,对降水量做了具体预测。结果表明,采用1~5阶时间步长预测时,相对误差较小,更符合要求;2014、2015年均为正常年,2014年的降水量为585.82 mm,2015年的降水量为649.21 mm;正常年和偏枯年出现的概率较大,枯水年出现的可能性最小。This paper analyzed the statistics of temporal variation of precipitation in Tongchuan of Shanxi Prov- ince using rainfall data measured from 1960--2013 and the Sample-mean method and the square-moment meth- od. We first demonstrated that the temporal series of the precipitation is statistically Markovian, and then devel- oped a weighting Markov Chain model to predict the precipitation and tested it against available data in this re- gion. The model was further combined with the fuzzy theory to predict rainfall. The results showed that the model is accurate and reduces error when using 1-5 backward precipitations. The model predicted that both 2014 and 2015 were weakly drought with an annual rainfall of 585.82 mm and 649.21 mm respectively, and that the possibility of occurrence of normaland weak drought yearswere high, while the occurrence of drought year was statistically low.

关 键 词：加权马尔可夫链 铜川地区 降水量预测 模糊集理论 遍历性与平稳分布 

分 类 号：P457.6[天文地球—大气科学及气象学]