Extreme precipitation has numerous impacts, such as flooding, damage to infrastructure, availability of food and water, and changes in insurance costs. Understanding the causes and behaviour of these extremes is important, as extreme precipitation is predicted to become both more intense and more frequent in the future. An atmospheric feature associated with extreme precipitation are extra-tropical cyclones. Cyclones are key components of the atmospheric general circulation due to their ability to transport large quantities of heat, moisture, and momentum. The baroclinic instability, which feeds cyclones, largely balances the planetary budgets of energy and moisture at mid-latitudes. In this thesis, precipitation extremes are examined within cyclones across the Southern Hemisphere, and the underlying processes driving extreme precipitation within cyclones are investigated.

ERA5 reanalysis output is used in a mean sea level pressure cyclone tracking scheme to identify cyclone centres and tracks. A cyclone compositing framework is used to provide a cyclone-centred perspective by examining data within 2000 km of each identified cyclone centre. Before using ERA5 to determine extreme precipitation and its drivers, we assess its suitability for cyclone compositing and investigate an appropriate threshold for extreme precipitation.

ERA5 is initially compared with the WindSat dataset to assess the representation of cyclones within ERA5. WindSat is not assimilated into ERA5 and provides an independent assessment of ERA5’s quality. Compared to WindSat, ERA5 shows comparable spatial structures but underestimates total column water vapour by up to 5% and cloud liquid water by up to 40%. ERA5 also underestimates precipitation in the warm sector by up to 15%, but overestimates precipitation in the cold sector by up to 60%. This potentially suggests that cloud parameterisations are biased within ERA5. Despite these biases ERA5 shows a strong correlation with WindSat across the cyclone lifecycle, indicating its suitability for use in further cyclone compositing analysis.

ERA5 is also assessed globally to determine a threshold for extreme precipitation to use within the cyclone composites. This threshold is required to change spatially depending on the latitude/longitude of the cyclone, as an extreme in one region might not correspond to an extreme in another. By grouping precipitation based on how often it rains, we develop a new framework for assessing global precipitation. This framework finds a strong relationship identified between the frequency of wet days and the intensity distribution of precipitation on those days, even though it connects spatially disparate regions of the globe. This key result is also observed in a number of gauge, satellite and reanalysis precipitation datasets, and is not just a product of how precipitation is represented within ERA5. This reinforces the value of ERA5 in precipitation related studies.

Next, we apply the wet-day frequency regions within the cyclone compositing framework to assess extreme precipitation. Precipitation intensity distributions are used to mask cyclone composites based on 90th and 98th percentile thresholds, which change depending on the wet-day frequency. This novel methodology provides a way of quantitatively assessing precipitation extremes within a cyclone compositing framework. When cyclones are in a phase of strengthening before peak intensity, extreme precipitation occurs more often. A higher proportion of precipitation is determined as extreme during this period. The spatial extent of extreme precipitation is also found to be constrained around the cyclone centre compared to non-extreme precipitation, though extreme precipitation still displays a good spatial correlation with non-extreme precipitation.

Finally, we investigate the underlying processes causing extreme precipitation within cyclones and how they relate to both large-scale and convective precipitation. Overall, large-scale precipitation predominates and is most commonly found in the comma cloud region. Convective precipitation is weaker overall but more frequent in drier regions of the cyclone. During periods of extreme precipitation events, large-scale precipitation becomes even more dominant across the cyclone composite, but weakens as the cyclone decays. Vertical velocity strongly influences large-scale precipitation regardless of cyclone strength, though correlation weakens after peak intensity is reached. For extreme precipitation this correlation is more moderate and absent during the decay phase, suggesting that a further driver such as water vapour is also important for extreme large-scale precipitation. Convective available potential energy and integrated water vapour flux do not exhibit a spatial correlation with convective precipitation. However, a strong correlation between convective available potential energy and convective precipitation is observed during extreme precipitation. This potentially suggests CAPE is a necessary condition for extreme precipitation but is not capable of producing the rainfall on its own.