Extreme precipitation events are among the most critical climatic components, exerting a direct impact on flood occurrences, soil erosion, water resource management, and ecosystem stability. In semi-arid regions, changes in the frequency and spatial extent of heavy precipitation can lead to significant hydrological and environmental consequences. Therefore, investigating the spatiotemporal patterns of such precipitation events is of paramount importance. This study aimed to analyze the temporal variations of extreme precipitation in the southern region of Lake Urmia using percentile analysis and various statistical tests. To this end, daily precipitation data from eight meteorological stations over the period 2004–2025 (1382–1403 Solar Hijri) were examined. Prior to executing the primary tests, daily precipitation data from all stations were aggregated in a software environment, and Python programming was utilized to extract different percentiles—including the 75th, 90th, 95th, 99th, and 100th—as precipitation intensity indices. Subsequently, a set of non-parametric statistical methods was employed to investigate the statistical properties, temporal trends, and spatial dynamics of extreme precipitation events. These methods included the Mann-Kendall test for trend analysis, the Pettitt test for detecting abrupt change-points in time series, and the Generalized Extreme Value (GEV) distribution for estimating precipitation amounts corresponding to various return periods.
The trend evaluation based on non-parametric tests revealed that significant increasing or decreasing trends in intermediate percentiles (75th and 90th) were present in only a limited number of stations, whereas extreme percentiles (99th and 100th) exhibited heterogeneous behavior, showing decreasing trends in certain cases. Furthermore, results from the Pettitt step-change test confirmed the existence of regime shifts and change-points in the precipitation regime of several stations, predominantly occurring in the late decade. In this context, return period analyses based on probability distribution functions indicated substantial spatial discrepancies in extreme precipitation intensity between the western and eastern stations of the study area. Additionally, examining the temporal distribution of events showed that the highest frequency of moderate percentiles occurred in March and April, while highly extreme events demonstrated a greater propensity to occur during the colder months of the year. Overall, the findings of this research indicate that the pattern of extreme precipitation in the south of Lake Urmia has undergone structural changes, which carries vital implications for flood management and water resource planning in the region, particularly within the context of sustainable restoration efforts for Lake Urmia.