A time-dependent non-asymptotic statistical analysis of extreme precipitation events

Time-dependent precipitation frequency analyses were often hampered by the availability of relatively short data records, which result in large uncertainty in the estimation of extremes. The recently developed non-asymptotic statistical methods, based on fitting ordinary events rather than extreme events only, represent a potential solution to the problem of data scarcity and are finding wide application in literature under assumptions of stationarity. Recent studies investigated the use of non-asymptotic methods under non-stationary conditions (e.g., Vidrio-Sahagún and He, 2022) and advocated their use over other methods for non-stationary frequency analysis of extreme precipitation. In this study we formalize a non-stationary time-dependent approach for the statistical analysis of multi-duration precipitation extremes using simplified metastatistical extreme value (SMEV) approach. The study focuses on a catchment in the Eastern Italian Alps, where trends in extreme precipitation where reported (Dallan et al., 2022) and which was impacted by the exceptional Vaia event in 2018. We provide an estimation of extreme return levels of precipitation in six stations in the neighborhood of the catchment and compare them with precipitation maxima observed during Vaia storm. The results show that using a non-stationary left-censored Weibull distribution, with both scale and shape parameters linearly dependent on time, allows to properly describe the observed trends of intense precipitation for different durations. Our results suggest that the probability of observing events like Vaia increased over the past decades, leading to the need for updating local adaptation measures.

 

References:

Dallan, E., Borga, M., Zaramella, M., & Marra, F. (2022). Enhanced summer convection explains observed trends in extreme subdaily precipitation in the eastern Italian Alps. Geophysical Research Letters, 49(5), e2021GL096727.

Vidrio-Sahagún, C. T., & He, J. (2022). Hydrological frequency analysis under nonstationarity using the Metastatistical approach and its simplified version. Advances in Water Resources, 166, 104244.