The disproportionate rates of change between extreme and mean temperatures over land

• Main Author: Gross, Mia
• Format: Doctoral or Postdoctoral Thesis
• Language: unknown
• Published: UNSW Sydney 2019
• Subjects: Climate models; Climate change; Temperature extremes; Observational datasets; Arctic
• Online Access: https://dx.doi.org/10.26190/unsworks/21291; http://hdl.handle.net/1959.4/62700

The world is warming, with both extreme and mean temperatures getting warmer. But how extremes are changing and will change relative to the mean remains less clear. Disproportionate rates of change between extreme and mean temperatures need to be better understood because this changes the shape of the distribution, affecting the probability of extreme events and thus their impacts. Therefore, the goal of this thesis is to understand if, when and where disproportionate rates of change occur in the past and future and it does this by using observations, reanalyses and climate models. I find that for past decades, cold extremes have been warming faster than the mean for much of the Northern Hemisphere extratropics, while warm extremes have been warming faster than the mean in some subtropical regions. Future changes are systematic and robust across a range of climate model simulations. The most striking disproportionate changes are in the Northern Hemisphere mid- to high-latitudes, where cold extremes are projected to warm substantially faster than mean temperatures in all seasons except boreal summer. Exploring conditions on or leading into the day of the projected cold extremes reveals that the disproportionate warming is driven by different mechanisms in different seasons. In boreal winter, reduced cold air advection is the dominant driver, circulating anomalously warm temperatures from the Arctic to lower latitudes. But during spring and autumn, it is mostly due to feedbacks related to decreases in snow cover. Analyses of temperature extremes have inherent uncertainties. I evaluate several commonly used reanalyses with a gridded in situ-based daily temperature dataset to assess sensitivities related to dataset choice. Trends in extremes and statistical moments, other than the mean, exhibit sensitivity. However, the conclusions drawn in this thesis remain robust irrespective of dataset choice, and regardless of methodological choice including choice of base period and how extremes are defined. Ultimately, this thesis provides a comprehensive understanding of changes in daily temperature extremes relative to the mean and their drivers. In turn, this provides essential information for decision-makers who can act to reduce the negative impacts stemming from extreme temperatures.