权马尔可夫链在降水量预测中的应用  

Application of weighted Markov chain in precipitation prediction

作　　者：胡鑫[1] 车欣原 HU Xin;CHE Xinyuan(Qujing Branch of Yunnan Hydrological and Water Resources Bureau,Qujing 655000,China;School of Economics,Yunnan University,Kunming 650500,China)

机构地区：[1]云南省水文水资源局曲靖分局,云南曲靖655000 [2]云南大学经济学院,云南昆明650500

出　　处：《水利信息化》2020年第1期32-38,共7页Water Resources Informatization

摘　　要：选取1976-2017年曲靖市年降水量资料,采用均值-标准差分级法对其进行分级,以规范化的各阶自相关系数为权重,应用权马尔可夫链模型并结合模糊集理论中的级别特征值,分析预测2016-2018年的年降水量,并对模型进行验证。研究结果表明:预测值与实测值的相对误差为3.54%~8.38%,符合《水文情报预报规范》规定的误差值要求,将该方法应用于降水预测是可行、有效的。The annual precipitation data of Qujing City from 1976 to 2017 are selected and classified by means of mean-standard deviation classification method. The weights of the normalized autocorrelation coefficients of each order are used. The weighted Markov chain model is applied to analyze and predict the annual precipitation of Qujing City from 2016 to 2018. And the model is validated by combining the hierarchical eigenvalues in the theory of fuzzy sets. The results show that the relative error between the predicted and measured values is between 3.54% and 8.38%. It conforms to the error value requirements specified in the code for Hydrological Information and Prediction. It is feasible and effective to apply this method to precipitation prediction.

关 键 词：权马尔可夫链 降水量预测 级别特征值 曲靖市 

分 类 号：P457.6[天文地球—大气科学及气象学]