Performance comparison of GPM IMERG V07 with its predecessor V06 and its application in extreme precipitation clustering over Türkiye

Defining regions with similar characteristics for extreme precipitation is crucial for understanding the impacts of climate change, planning and managing water resources, and designing hydraulic structures. However, studies on the regionalization of extreme precipitation for T & uuml;rkiye are limited, and regional extreme precipitation characteristics are not well defined. In this study, motivated by the need to contribute to this field, homogenous regions for extreme precipitation across T & uuml;rkiye were determined using the latest version (V07) of Integrated Multi-satellitE Retrievals for GPM (IMERG). We initially validated IMERG V07 estimates using data from 214 ground-based stations and compared the results with its predecessor V06. The results revealed that IMERG showed some notable improvements from V06 to V07 for all seasons, especially in winter. During this season, the correlation coefficient increased from 0.57 to 0.64, the mean absolute bias decreased from 78.22 % to 69.27 %, and the RMSE decreased from 11.10 mm/day to 9.70 mm/day. In V07, while the trend of decreasing accuracy with increasing elevation observed in V06 continues, it has been shown that some notable improvements were achieved in continuous and categorical metrics. We then applied widely used non-hierarchical (K-means) and hierarchical (Ward's method) clustering techniques. To perform this, we first applied Principal Component Analysis (PCA) to reduce the number of variables related to extreme precipitation (e.g. amount, frequency, standard deviation, and seasonality) and geographic characteristics to identify the most significant variables for analysis. The K-means method delineated T & uuml;rkiye into eight extreme precipitation regions, while the Ward's method resulted in six distinct extreme precipitation regions. We evaluated the results based on the existing extreme precipitation climatology literature for T & uuml;rkiye and by associating them to known precipitation dynamics, and as a result, we recommended eight precipitation regions determined by the K-means. The identified precipitation regions are expected to contribute to future studies analyzing the effects of climate change and to regional studies on natural disasters resulting from extreme precipitation.