利用全局与局部相关函数分析流域降水空间变异性

Rainfall spatial variation models using global and local spatial correlation functions

为合理诊断流域降水的空间变异性,阐明了6种常见的全局和局部空间相关指标的联系与区别,并用空间相关函数进行统一表述,然后提出了同时从全局和局部2个方面对连续性与间断性并存的降水空间变异性进行综合分析的方法。采用全局半方差函数和局部Moran指数,对2003～2009年赣江流域降水空间变异规律进行了实例研究。结果表明:全局半方差函数便于描述流域降水的整体相关结构,而局部Moran指数可揭示降水的局部聚集形态和地带性规律,并能够诊断其中的奇异点。赣江流域年、月、日降水量在整体上呈显著的时变空间正相关性;在局部表现出较强的聚集性,并具有非平稳性和奇异性。同时,还初步指出了赣江流域降水空间变异性分析对于解释降水空间插值结果的作用。

Six global or local spatial correlation indices characterizing rainfall spatial variations were compared. Their similarities and differences were clarified. Then, they were uniformly named by spatial correlation function （SCF）. A global and local variation analysis methodology was developed for complex rainfall field both continuous and discontinuous. The precipitation from 2003 2009 over the Ganjiang River Basin was analyzed using the global semi variogram （SVF） and local Moran＇s index （LMI）. The results suggest that SVF is capable of describing the global rainfall distribution correlation and its decay while LMI can describe local clustering and singularities for yearly, monthly and daily time steps. For the Ganjiang River Basin, the global rainfall at various time steps all manifest positive temporal and spatial correlations. However, there are local clusters and some singular points. Also, the role of rainfall variation analysis for interpreting spatial interpolation results was preliminarily stated.