Abstract
The lack of data in rainfall stations of Iran is one of the main problems in design and management of hydrologic systems. Moreover, the density of these stations network is not sufficient for estimation of rainfall at ungauged regions. Therefore, regionalization can be an essential tool to be applied for clustering the rainfall and spatial pattern analysis of homogeneous regions to quantify regional rainfall patterns. Homogeneous regions are usually defined based on different methods and with consideration of a category of attributes. Selection of attributes as representatives of the study region is an important aspect in clustering of a region, as is the importance degree (or determined weight) that each of these attributes can allocate to themselves. Consequently, the aim of this study is to select a broad category of climatic, geographical, and statistical attributes of the maximum 24‐h rainfall of the Urmia Lake Basin for 63 selected stations for the period 1979–2008 and next to determine an appropriate weight for each of the attributes in each defined category. To investigate the weighting effect in regionalizing and to determine the appropriate weight for each defined attribute, respectively, Ward's clustering technique, principal component analysis, and correlation coefficients matrix methods were used. The homogeneity measure test showed that all identified clusters are homogeneous. The clustering results showed that based on the different attributes categories, different results can be presented in terms of the number of clusters, distribution of stations, and spatial pattern of clusters. Moreover, the performances of the proposed weighting approaches for spatial clustering analysis are better than no‐weight scenario in most modes according to the spatial patterns and homogeneity measurements.