An Area-Orientated Analysis of the Temporal Variation of Extreme Daily Rainfall in Great Britain and Australia

This paper presents an analysis of the temporary variation of the area-orientated annual maximum daily rainfall (AMDR) with respect to the three spatial properties: location, size and shape of the region-of-interest (ROI) in Great Britain and Australia using two century-long datasets. The Maximum Likelihood and Bayesian Markov-Chain-Monte-Carlo methods are employed to quantify the time-varying frequency of AMDR, where a large proportion of the ROIs shows a non-decreasing level of most frequent AMDR. While the most frequent AMDR values generally decrease with larger-sized ROIs, their temporal variation that can be attributed to the climate change impact does not show the same dependency on the size. Climate change impact on ROI-orientated extreme rainfall is seen higher for rounded shapes although the ROI shape is not as significant as the other two spatial properties. Comparison of the AMDR at different return levels shows an underestimation by conventionally used stationary models in regions where a nonstationary (i.e., time-varying) model is preferred. The findings suggest an overhaul of the current storm design procedure in view of the impact of not only climate change but also spatial variation in natural processes.

Automated summary

This paper analyzes the temporal variation of annual maximum daily rainfall (AMDR) in Great Britain and Australia, considering the spatial properties of location, size, and shape. The study finds that most regions show an increasing trend in AMDR over time, and larger-sized regions tend to have lower AMDR values. However, the temporal variation of AMDR does not have a consistent relationship with the size of the region. The impact of climate change on extreme rainfall is higher for rounded shapes, although the shape of the region is not as significant as the other two spatial properties. Additionally, conventional stationary models underestimate AMDR in regions where a nonstationary (i.e., time-varying) model is preferred. The findings suggest the need for a comprehensive overhaul of storm design procedures to account for the impact of climate change and spatial variation in natural processes.