Climate scientists routinely rely on averaging over time or space to simplify complex information and to concisely communicate findings. Currently, no consistent definitions of 'warm' or 'cool' seasons for southern Australia exist, making comparisons across studies difficult. Similarly, numerous climate studies in Australia use either arbitrarily defined areas or the Natural Resource Management (NRM) clusters to perform spatial averaging. While the NRM regions were informed by temperature and rainfall information, they remain somewhat arbitrary. Here we use weather type influence on rainfall and clustering methods to quantitatively define climatic regions and seasons over southern Australia. Three methods are explored: k-means clustering and two agglomerative clustering methods, Ward linkage and average linkage. K-means was found to be preferred in temporal clustering, while the average linkage method was preferred for spatial clustering. For southern Australia as a whole, we define the cool season as April-September and warm season as October-March, though we note that a three-season split may provide more nuanced climate analysis. We also show that different regions across southern Australia experience different seasons and demonstrate the changing spatial influence of weather types with the seasons, which may aid regionally or seasonally specific climate analysis. Division of southern Australia into 15 climatic regions shows localised agreement with the NRM clusters where distinct differences in rainfall amounts exist. However, the climate regions defined here better represent the importance of topographical aspect on weather type influence and the inland extent of particular weather types. We suggest that the use of these regions would provide consistent climate analysis across studies if widely adopted. A key requirement for climate scientists is the simplification of data sets into both seasonally or regionally averaged subsets. This simplification, by grouping like regions or seasons, is done for a number of reasons both scientific and practical, including to help understand patterns of variability, underlying drivers and trends in climate and weather, to communicate large amounts of data concisely, to reduce the amount of data required for processing (which becomes increasingly important with higher resolution climate model output), or to more simply draw a physical boundary between regions for other purposes, such as flora and fauna habitat analysis, appropriate agricultural practices or water management.

Disastrous bushfires during the last months of 2019 and January 2020 affected Australia, raising the question to what extent the risk of these fires was exacerbated by anthropogenic climate change. To answer the question for southeastern Australia, where fires were particularly severe, affecting people and ecosystems, we use a physically based index of fire weather, the Fire Weather Index; long-Term observations of heat and drought; and 11 large ensembles of state-of-The-Art climate models. We find large trends in the Fire Weather Index in the fifth-generation European Centre for Medium-Range Weather Forecasts (ECMWF) Atmospheric Reanalysis (ERA5) since 1979 and a smaller but significant increase by at least 30 % in the models. Therefore, we find that climate change has induced a higher weather-induced risk of such an extreme fire season. This trend is mainly driven by the increase of temperature extremes. In agreement with previous analyses we find that heat extremes have become more likely by at least a factor of 2 due to the long-Term warming trend. However, current climate models overestimate variability and tend to underestimate the long-Term trend in these extremes, so the true change in the likelihood of extreme heat could be larger, suggesting that the attribution of the increased fire weather risk is a conservative estimate. We do not find an attributable trend in either extreme annual drought or the driest month of the fire season, September-February. The observations, however, show a weak drying trend in the annual mean. For the 2019/20 season more than half of the July-December drought was driven by record excursions of the Indian Ocean Dipole and Southern Annular Mode, factors which are included in the analysis here. The study reveals the complexity of the 2019/20 bushfire event, with some but not all drivers showing an imprint of anthropogenic climate change. Finally, the study concludes with a qualitative review of various vulnerability and<span idCombining double low line"page942"/> exposure factors that each play a role, along with the hazard in increasing or decreasing the overall impact of the bushfires.

Wind observations collected at citizen weather stations (CWSs) could be an invaluable resource in climate and meteorology studies, yet these observations are underutilised because scientists do not have confidence in their quality. These wind speed observations have systematic biases, likely caused by improper instrumentation and station sitings. Such systematic biases introduce spatial inconsistencies that prevent comparison of these stations spatially and limit the possible usage of the data. In this paper, we address these issues by improving and developing new methods for identifying suspect observations and adjusting systematic biases. Our complete quality control and bias adjustment procedure consists of four steps: (a) performing within-station quality control tests to check the plausible range and the temporal consistency of observations, (b) adjusting the systematic bias using empirical quantile mapping, (c) implementing between-station quality control to compare observations from neighbouring stations to identify spatially inconsistent observations, and (d) providing estimates of the true wind when CWSs falsely report zero wind speeds, as a complement to the bias adjustment. We apply these methods to CWSs from the Weather Observation Website (WOW) in the Netherlands, comparing the crowdsourced data with official data, and statistically assessing the improvements in data quality after each step. The results demonstrate that the crowdsourced wind speed data are more comparable with official data after quality control checks and bias adjustment steps. Our quality assessment methods therefore give confidence in CWSs, converting their observations into a usable data product and an invaluable resource for applications in need of additional wind observations.

To mitigate the risk posed by extreme rainfall events, we require statistical models that reliably capture extremes in continuous space with dependence. However, assuming a stationary dependence structure in such models is often erroneous, particularly over large geographical domains. Furthermore, there are limitations on the ability to fit existing models, such as max-stable processes, to a large number of locations. To address these modelling challenges, we present a regionalisation method that partitions stations into regions of similar extremal dependence using clustering. To demonstrate our regionalisation approach, we consider a study region of Australia and discuss the results with respect to known climate and topographic features. To visualise and evaluate the effectiveness of the partitioning, we fit max-stable models to each of the regions. This work serves as a prelude to how one might consider undertaking a project where spatial dependence is non-stationary and is modelled on a large geographical scale.