Identification of past extreme precipitation events and their connection to recorded impacts: a multi-data and multi-method assessment over the Central-Eastern Alps.

Extreme hydro-meteorological events are increasingly prevalent in southern Europe, particularly in the European Alps, posing significant threats to ecological and socio-economic systems. The accurate detection and analysis of these events require a nuanced definition of what constitutes "extreme." While statistical approaches typically define extremes based on the tails of probability distributions, it is essential to recognize that the severity of these events may not always align with statistical extremes. Impact-related thresholds can vary spatially and temporally, making a single absolute threshold inadequate for capturing extremes across different locations, time periods, and seasons.

In this study, we focus on identifying and characterizing extreme hydro-meteorological events in a transboundary Alpine region between Austria and Italy from 2003 to 2021. We employ various definitions of extreme events that consider spatiotemporal aspects and utilize multiple datasets. Daily accumulated precipitation serves as the primary parameter due to its widespread availability across datasets and its role as a triggering factor for various hazards like landslides, debris flows, and floods.

We employ three statistical methods to detect extreme events: regional-scale identification of highest daily precipitation amounts, local-scale detection of high-intensity daily precipitation values, and identification of exceptional daily precipitation records relative to average conditions for specific periods of the year. These methods are applied to four gridded precipitation datasets, including observation and reanalysis products, each with different technical specifications.

Subsequently, we compare the identified events from each method-dataset combination with existing records of gravitational mass movements and fluvial floods in the Austrian-Italian border region to assess their ability to detect actual impacts. Findings suggest that a majority of detected precipitation extremes (e.g., 74% for regional scale identification with reanalysis data) correlate with observed impacts. However, different method-dataset combinations exhibit varying strengths and weaknesses, reflecting the characteristics of the dataset and/or statistical method employed. Some combinations show lower performance in detecting impactful events due to conflicts between dataset resolution and statistical method requirements.