Stormy Subtropics and Stratiform South: A Radar-Based Classification of Australian Rainfall Events

Weather radar is an essential data source that allows for rainfall to be observed at a high temporal and spatial resolution. Further, radar data is useful for thunderstorm monitoring, flood nowcasting, and in the case of well-established archives, creation of rainfall climatologies and comparison of heavy and non-heavy rainfall events. Despite this and the presence of a large operational network spanning a variety of different climatic regions, Australian weather radar data is not commonly used to study rainfall. From both an Australian and international perspective, there is minimal focus in the literature on using radar data to identify and characterize heavy rainfall events and to identify what distinguishes them from less heavy events. For this work, processed radar data from the newly developed Australian Unified Radar Archive (AURA) is used in an innovative threshold-based approach to identify rainfall events at a variety of locations across Australia. Radar-derived areal rainfall characteristics such as duration, intensity, areal coverage, accumulated rainfall, and convective contributions are obtained and used to divide rainfall events into distinct classes using k-means clustering. Two different types of heavy rainfall events are extracted and investigated: high accumulation events, which are the 100 events with the highest total rainfall accumulations for each location, and high intensity events, the 100 events with the highest average intensities for each location. It is found that there are strong regional differences in the seasonality and character of rainfall events in Australia. Events in the tropics/subtropics are more convective in nature and generally more common in the warm season, while events in the midlatitudes have more stratiform attributes and tend to be more frequent in winter. Significant differences are also evident in the nature of high accumulation, high intensity, and non-heavy (neither high accumulation nor high intensity) rainfall events, with high accumulation events being long-lived, spatially extensive, and relatively intense, and high intensity events being localized, extremely intense, and strongly convective. In inland regions the rainfall events that produce the largest rainfall accumulations are more likely to be those with the highest intensities, while in coastal regions, there is a greater separation between the events producing the most accumulated rainfall and the events producing the most intense rainfall. Overall, radar data displays great utility in studying the characteristics and morphology of rainfall events and heavy rainfall events. But improvements in data quality, record lengths, and coverage of remote areas are needed in order for this utility to be fully realized for its potential use in operational forecasting in Australia. In particular, an ongoing challenge at present is screening out ground clutter and anomalous propagation in radar data, although many cases are able to be identified and omitted from analysis by searching for specific combinations of rainfall event characteristics. For future work, more detailed classification of rainfall event morphology is planned by developing an automated approach, as the majority of past work relies on manual classification of radar images into morphological archetypes.