The local characteristics of multi-dimensional random variables are seldom considered in the general modeling method of multivariate copula. A new modeling method, called pair-copula construction, is introduced to remedy this defect. Different types of copula distribution functions are allowed to be introduced in this method. Correspondingly, the related characteristics of complex multivariate can be described by a cascade of pair-copula acting on two variables at a time. In the analysis of asynchronism-synchronism of regional precipitation in WED interbasin water transfer areas, the pair-copula construction method is compared with the general modeling method of multivariate copula. The results show that the local dependence structure would exist among hydrologic variables even in three-dimensional cases. In this situation, the general modeling method of multivariate copula would face difficulties in fitting distribution. However, the pair-copula construction method could capture the local information of hydrologic variables efficiently by introducing different types of copula distribution functions. Moreover, the compensation capacity of water resources is strong in different hydrological areas of WED water transfer project. The asynchronous frequency of wetness and dryness is 69.64% and the favorable frequency for water transfer is 46.15%.