Event-based extreme precipitation variability analysis over a part of the Hindu Kush Himalayan region

Extreme precipitation is a primary driver of flood hazards causing damage to humans and infrastructure. Analysis of extreme precipitation without considering temporally compounding events underestimates the severity of the hazard. Especially mountainous regions such as the Hindu Kush Himalayan region are highly vulnerable to multiday extreme precipitation events. However, analysis of such events and their time distribution patterns are still unknown, and a comprehensive assessment is needed. To this, the concept of event-based extreme precipitation, which accounts for the preceding and succeeding precipitation, is employed for the Upper Indus basin using three reference gridded datasets, namely, Indian Meteorological Department (IMD), Multi-Source Weighted Ensemble Precipitation (MSWEP) and Himalayan Adaption, Water and Resilience (HI-AWARE). Four different types of time distribution patterns were identified based on the occurrence of extreme precipitation during an event. We identified that the time distribution pattern with the peak on the right side is predominant among the four types. Subsequently, trend analysis on the characteristics, namely amount, frequency, duration and concentration ratio for all four events, display negative trends with IMD and MSWEP datasets, whereas HI-AWARE displays positive trends in the northwestern part of the catchment. Further, the 100-year return level of the amount of multiday extreme precipitation is computed along with the traditional method of single-day extremes considering nonstationarity. The differences between the return values obtained using the traditional method and the event-based extreme precipitation concept were distinct and substantial (>50 mm for three datasets). The findings clearly show that the analysis of multiday events is much more essential than single-day extreme events, particularly for events like floods. Moreover, the results of the study were found to be conflicting among reference datasets, demonstrating the importance of identifying the suitability of reference datasets in extreme event analysis.