In the present study, classical and proposed methods were used to investigate the monthly precipitation characteristics of 30 stations in the southeastern United States during 1968–2018. Maximal overlap discrete wavelet transform (MODWT) as preprocessing method and K-means clustering method were used. First, the monthly precipitation time series of stations were decomposed into several subseries using MODWT and considering db as the mother wavelet. Then, the energy values of theses subseries were calculated and used as inputs in K-means and radial basis functions (RBF) methods. The optimum number of clusters obtained for the considered stations in both classical and proposed methods was five clusters. In order to use the data as the input of the RBF method, the data correlation was evaluated by variogram. Based on the results of clustering and in accordance with the latitude and longitude variations of the stations, it was found that with increasing the energy of the clusters, the amount of precipitation in the stations decreased and vice versa. The silhouette coefficient of clustering for the classical method obtained was 0.3 and for the proposed method it was 0.8, which indicates better clustering of the selected area using the proposed method.

在本研究中，采用经典方法和拟议方法调查了1968-2018年美国东南部30个气象站的月降水特征。采用最大重叠离散小波变换（MODWT）作为预处理方法和K-均值聚类方法。首先，使用MODWT并将db作为母子波，将月降水量的时间序列分解为几个子序列。然后，计算这些子系列的能量值，并将其用作K均值和径向基函数（RBF）方法的输入。在经典方法和建议方法中，为考虑的站点获得的最佳聚类数量为五个聚类。为了将数据用作RBF方法的输入，通过方差图评估了数据相关性。根据聚类的结果并根据站点的纬度和经度变化，发现随着集群能量的增加，站点中的降水量减少，反之亦然。对于获得的经典方法，聚类的轮廓系数为0.3，对于所建议的方法，聚类的轮廓系数为0.8，这表明使用所提出的方法对所选区域进行了更好的聚类。