Large-scale ocean-atmospheric processes and seasonal rainfall variability in South Australia: potential for improving seasonal hydroclimatic forecasts

• Main Authors: Tozer, C. R., Kiem, AS
• Other Authors: The University of Newcastle. Faculty of Science, School of Environmental and Life Sciences
• Format: Article in Journal/Newspaper
• Language: English
• Published: John Wiley & Sons 2017
• Subjects: rainfall variability; seasonal forecasting; Partial Mutual Information; ENSO; IOD; subtropical ridge; non-linear; blocking; Indian; Pacific; Southern Ocean
• Online Access: http://hdl.handle.net/1959.13/1352385

Seasonal rainfall forecasts are an important tool for risk management across many sectors. However, significant challenges arise in the development of skilful and practically useful seasonal forecasts for regions where the temporal and spatial variability of rainfall is large and/or knowledge about what causes this variability is in its infancy. This is evident in the state of South Australia (SA), where seasonal rainfall currently has low predictive skill. The key climate processes have yet to be fully identified in SA and therefore may not be adequately represented in forecast models. The aim of this paper is to identify and quantify relationships between large-scale ocean-atmospheric processes and seasonal rainfall variability across SA. We identify two distinct climate zones: (1) the arid northern region, where rainfall is mostly influenced by climate processes stemming from the tropical Indian and/or Pacific Oceans and (2) southern SA, which is dominated by Southern Ocean processes. The average percent of variability of SA rainfall accounted for by any single large-scale climate process (i.e. linear regression using a single predictor) is 8% in summer, 19% in autumn, 33% in winter and 24% in spring. However, when two or more processes are considered in combination (through multiple linear regression), this rises to 13, 26, 46, and 33%, respectively, highlighting the importance of capturing the interaction among multiple climate processes. Importantly, the findings from this study provide a set of metrics against which existing statistical and dynamical forecasting schemes can be tested and highlight processes that should be focused on in order to improve (or develop new) forecasting schemes. The study also recommends the need for further investigations into non-linear relationships between rainfall and large-scale ocean-atmospheric processes and the development of more objective methods for determining which climate process, or combination of processes, are most important for a certain season or location.