Precipitation is the main water resource for agronomy. The frequency is one of the most significant characteristics of precipitation. The objectives of this study are (1) to identify the homogenous regions, (2) to determine the probability distribution that fits precipitation the best for each region, (3) to perform regional precipitation-frequency analysis by using L-moment methods, and (4) to derive the return levels for precipitation. In this study, the precipitation data collected from 24 meteorological stations for the period 1965–2013 over Serbia were analyzed. Three distributions [generalized extreme value (GEV), generalized Pareto (GPD), and generalized logistic (GLO)] based on three parameters (scale, shape, and location) are investigated. The L-moment method is applied to determine the parameters. Three independent precipitation regions (R1, R2, and R3) were studied. The homogeneity test indicates that the identified regions are homogenous. To confirm the goodness-of-fit for the selected three probability distributions, the Z-statistics was applied. Based on the obtained results, the GEV distribution best fits the precipitation data in the regions R1 and R2, while the GPD was selected for the region R3.

降水是农学的主要水资源。频率是降水的最重要特征之一。这项研究的目的是（1）确定同质区域；（2）确定每个区域最适合降水的概率分布；（3）使用L矩方法进行区域降水频率分析；以及（4）得出降水的返回水平。在这项研究中，分析了从1965年至2013年期间塞尔维亚24个气象站收集的降水数据。研究了基于三个参数（比例，形状和位置）的三种分布[广义极值（GEV），广义Pareto（GPD）和广义Logistic（GLO）]。L矩法用于确定参数。研究了三个独立的降水区域（R1，R2和R3）。均质性测试表明所识别的区域是均质的。为了确认所选三个概率分布的拟合优度，应用了Z统计量。根据获得的结果，GEV分布最适合区域R1和R2中的降水数据，而为区域R3选择了GPD。