[A-AS03] Extreme Events and Mesoscale Weather: Observations and Modeling

[AAS03-P05] Extreme Precipitation Estimation Using Stochastic Storm Transposition with Large Ensemble: A Case Study in Japan.

*Wasitha Randeepa Ranga Munipurage¹, Yusuke Hiraga¹, Jose Angelo Hokson¹ (1.Tohoku University, Japan)

Keywords:d4PDF, extreme rainfall estimation, flood risk management , probable maximum precipitation, stochastic storm transposition

Accurate estimation of Probable Maximum Precipitation (PMP) is crucial for the design and management of critical water infrastructure, particularly in regions prone to extreme rainfall events. Traditional methods of PMP estimation often rely on limited historical data and struggle to account for the potential impacts of climate change. Stochastic Storm Transposition (SST) has emerged as a powerful alternative, offering a more comprehensive and flexible approach to extreme rainfall estimation. SST is a storm-based technique that leverages observed rainfall patterns to generate synthetic extreme events. By applying temporal resampling and spatial transposition to historical storms, SST can create a vast array of realistic extreme rainfall scenarios, effectively extending the record of extreme events beyond the limitations of observe d data. This method is particularly valuable for estimating long return period events, such as the 10,000-year return period rainfall, which are critical for robust infrastructure design and risk assessment. This study aims to advance the application of SST by incorporating large ensemble climate data, specifically the d4PDF 5km resolution rainfall dataset with 12 ensemble members, to estimate PMP for the Akagawa and Arakawa watersheds in Western Tohoku and Hokuriku regions of Japan. The d4PDF dataset, covering the period from 1951 to 2011, provides more comprehensive representation of climate variability and potential extreme events. By coupling this dataset with SST, we aim to enhance the accuracy and reliability of PMP estimates while minimizing the uncertainty associated with estimating extreme precipitation for longer return periods such as 10,000 and 100,000 years. The large ensemble approach allows for a more thorough exploration of climate variability and its impact on extreme rainfall, potentially capturing a wider range of possible storm scenarios and their associated probabilities. In this study, we employ the SST methodology, facilitating efficient processing of the large ensemble data and generation of synthetic storm events. By leveraging the spatial and temporal diversity present in the d4PDF ensemble, the SST approach can provide a more robust assessment of potential extreme rainfall, accounting for both observed climate variability and possible future changes.