Extreme rainfall archetypes for Australia

The concept of an archetype appears in areas relating to behavior, historical psychology, and literary analysis. Cutler and Breiman (1994) introduced archetypal analysis (AA), a data clustering method use to identify and differentiate between extreme configurations in point sets. Within the National Environmental Science Program (NESP), The Earth Systems and Climate Change Hub’s project, entitled Extreme Climate: Dry, Wet, Hot-and-Dry, we apply the AA methodology for the first time to Australia’s notoriously variable rainfall – allowing us to capture extreme rainfall patterns across the continent as a whole, together with assessments of when these patterns occur and how they relate to one another. This information serves as a new starting point for understanding key wet and dry events in Australia at the continent scale. This document should be used in conjunction with Extreme rainfall archetypes explainer currently in preparation and to be made available on the NESP The Earth Systems and Climate Change Hub website upon completion.\nLineage: The archetypal analysis of Australian rainfall relies solely on Australian Gridded Climate Data monthly precipitation (Evans et al., 2020) in NetCDF format hosted by National Computing Infrastructure’s (NCI) on its data services portal. The following data source summary describing the dataset is available from each and every NetCDF file’s global attribute and reads "The monthly rainfall data represents the amount of precipitation of any type (including the liquid equivalent of frozen hydrometeors such as hail and snow) observed by means of rain gauges stationed across Australia. These gauges measure millimetres of liquid water depth over a 24-hour period running from 9am\nlocal clock time the previous day to 9am on the observed day. Monthly rainfall totals are calculated from the accumulation of the contributing daily rainfall totals at stations, and may incorporate multi-day rainfall totals if observations weremissing on any one day. Rainfall data from up to 8000 sites across the country\nare analysed, with limited quality control, onto 0.01x0.01 degree grids using the statistical interpolation technique described in Evans et al. (2020). Multi-month\nrainfall totals are calculated from the sum of monthly grids"