基于多重分形的降雨时空解集研究  

Study on Rainfall Space-time Downscaling Based on Multifractal Analysis

作　　者：周玉良[1,2] 陆桂华[2] 吴志勇[2] 金菊良[1] 

机构地区：[1]合肥工业大学土木与水利工程学院,安徽合肥230009 [2]河海大学水文水资源与水利工程科学国家重点实验室,江苏南京210098

出　　处：《四川大学学报（工程科学版）》2010年第2期26-33,共8页Journal of Sichuan University (Engineering Science Edition)

基　　金：国家科技支撑计划课题(2006BAC05B02;2007BAC03A060301);水利部公益性行业科研专项经费项目(200701039;20080100105);水利部"948"计划资助项目(200758);国家自然科学基金资助项目(70771035)

摘　　要：为建立不同时空尺度降雨间联系,分析"淮河流域水分和能量循环试验与研究"中Doppler雷达测雨数据的时空多重分形特性,估算降雨场的对流速度,实现了降雨场的时空转换。采用2参数对数泊松分布作为随机级联生成子,建立基于随机级联的降雨时空多重分形解集模型,研究对数泊松分布的参数与大尺度雨强及降雨场离差系数间关系,比较解集模型模拟和实测雷达降雨场的统计特征。结果表明,降雨近似为自相似多重分形过程,建立的降雨时空解集模型能重现雷达观测降雨场的标度特性和保持实测雷达降雨场的统计特征。In order to establish the relations between different scale rainfall fields,a space-time multifractal analysis on Doppler radar rainfall from HUBEX was proposed and scaling anisotropy parameters were estimated by spectral analysis.The results indicated that space-time rainfall can be considered to be a self-similar multifractal process approximately.The advection velocities of rainfall fields estimated through tracking techniques were adopted to rescale the time dimension to space dimension.The log-Poisson distribution with two parameters was used as stochastic cascade generator,and the multifractal rainfall space-time downscaling model was developed based on stochastic cascade.The relations between parameters of the log-Poisson distribution and rainfall intensity on larger scale and the coefficient of variation of rainfall fields was studied,and the characteristics of simulated rainfall fields through downscaling model was compared with the radar observed rainfall fields.The results showed that the simulated rainfall fields match scaling behaviors and preserve characteristic properties very well with those for the observed rainfall fields.

关 键 词：降雨 解集 尺度转换 多重分形 随机级联 

分 类 号：P33[天文地球—水文科学] P333.6[水利工程—水文学及水资源]