Arabian Peninsula Extreme Rainfall Characteristics Using Self-Organizing Maps

Precipitation climatology analyses in the Arabian Peninsula has been really scarce partly due to the enormous area to cover with limited observational data. Previous studies only attempted at observed mean rainfall trends. Extreme precipitation climatology specifically has not been in the spotlight until recently. When it is investigated, only case studies of recent well-known events are conducted using both reanalysis and modeling. Analyzing extreme precipitation characteristics for the Arabian Peninsula with longer time scale in the order of decades is therefore particularly important before any modeling work on the same topic. Studies such as sub-seasonal to seasonal forecasting of extreme precipitation will benefit greatly from knowledge of the characteristics of extreme precipitation climatology of the region.

In this study, an unsupervised learning artificial neural network called self-organizing maps has been used to characterize an in-house reanalysis of 5 km convective-permitting resolution precipitation. This is a unique state of the art data for the region with advanced data assimilation technique. The goal of the study is to relate the extreme rainfall of the high-resolution reanalysis to the large-scale circulations from the driving data. Result suggests that the method could be used to characterize extreme climatology for the Arabian Peninsula. This analysis successfully separates the extreme events from the crowd. The maps also capture patterns of what has been found as an Active Red Sea Trough rare weather phenomenon that framed to cause heavy rainfall for the region.