Investigating variability in the Australian monsoon and rainfall with cluster analysis

 
Poster PDF

Authors

Stuart Evans — University at Buffalo
Roger Marchand — University of Washington
Thomas P. Ackerman — University of Washington

Category

Atmospheric State & Surface

Description

Northern Australia experiences a strong seasonal cycle: wet monsoon flow during austral summer and dry continental flow during winter. During the monsoon season the weather can be divided into active and break periods. The timing of the onset and the retreat of the monsoon as well as the intensity of the monsoon are known to have strong relationships with the El Nino-Southern Oscillation (ENSO). Well within the tropics, northern Australia also feels the effects of the Madden-Julian Oscillation (MJO), which has been shown to affect rainfall. Over the past several decades Darwin, Australia has experienced an increase in rainfall, though with large interannual variability. Here, we use an atmospheric classification for the Darwin region to investigate the sources of the trends and variability in the monsoon and precipitation.

We have previously developed an atmospheric classification technique that applies a clustering algorithm to reanalysis data to define a set of atmospheric states. We performed a classification for a region surrounding Darwin, Australia, that defines eight atmospheric states. We use this set of states, especially two that correspond to the active and break periods of the monsoon, to classify the state of the atmosphere at Darwin for the 32-year period of the ERA-Interim project.

We use the time series of state to precisely identify the onset (first occurrence of one of the monsoon states) and retreat (last occurrence) of the monsoon each season, to identify when the monsoon switches between active and break periods within each season, and to define additional metrics such as monsoon duration and strength. On interannual timescales, our metrics capture the known relationships between the Australian monsoon and ENSO and demonstrate that the monsoon has become more active over the past 32 years. On intraseasonal timescales, we show that the timing of the transitions between active and break periods are influenced by the passage of the MJO. We also use the atmospheric classification to explain trends and variability in rainfall at Darwin. We find that while year-to-year variability in rainfall is best explained by changes in the precipitation associated with each atmospheric state, the long-term trend toward more rain at Darwin is due to the increasingly active monsoon season.