Hartigan, J, MacNamara, S & Leslie, LM 2020, 'Application of machine learning to attribution and prediction of seasonal precipitation and temperature trends in Canberra, Australia', Climate, vol. 8, no. 6.View/Download from: Publisher's site
© 2020 by the authors. Southeast Australia is frequently impacted by drought, requiring monitoring of how the various factors influencing drought change over time. Precipitation and temperature trends were analysed for Canberra, Australia, revealing decreasing autumn precipitation. However, annual precipitation remains stable as summer precipitation increased and the other seasons show no trend. Further, mean temperature increases in all seasons. These results suggest that Canberra is increasingly vulnerable to drought. Wavelet analysis suggests that the El-Nino Southern Oscillation (ENSO) influences precipitation and temperature in Canberra, although its impact on precipitation has decreased since the 2000s. Linear regression (LR) and support vector regression (SVR) were applied to attribute climate drivers of annual precipitation and mean maximum temperature (TMax). Important attributes of precipitation include ENSO, the southern annular mode (SAM), Indian Ocean Dipole (DMI) and Tasman Sea SST anomalies. Drivers of TMax included DMI and global warming attributes. The SVR models achieved high correlations of 0.737 and 0.531 on prediction of precipitation and TMax, respectively, outperforming the LR models which obtained correlations of 0.516 and 0.415 for prediction of precipitation and TMax on the testing data. This highlights the importance of continued research utilising machine learning methods for prediction of atmospheric variables and weather pattens on multiple time scales.
Hartigan, J, MacNamara, S & Leslie, LM 2018, 'Comparing precipitation and temperature trends between inland and coastal locations', ANZIAM Journal : Electronic Supplement, vol. 60, pp. C109-C126.View/Download from: Publisher's site
Motivated by the Millennium Drought and the current drought over much of southern and eastern Australia, this detailed statistical study compares trends in annual wet season precipitation and temperature between a coastal site (Newcastle) and an inland site (Scone). Bootstrap permutation tests reveal Scone precipitation has decreased significantly over the past 40 years (p-value=0.070) whereas Newcastle has recorded little to no change (p-value=0.800). Mean maximum and minimum temperatures for Newcastle have increased over the past 40 years (p-values of 0.002 and 0.015, respectively) while the mean maximum temperature for Scone has increased (p-value = 0.058) and the mean minimum temperature has remained stable. This suggests mean temperatures during the wet season for both locations are increasing. Considering these trends along with those for precipitation, water resources in the Hunter region will be increasingly strained as a result of increased evaporation with either similar or less precipitation falling in the region. Wavelet analysis reveals that both sites have similar power spectra for precipitation and mean maximum temperature with a statistically significant signal in the two to seven year period, typically indicative of the El-Nino Southern Oscillation climate driver. The El-Nino Southern Oscillation also drives the Newcastle mean minimum temperature, whereas the Scone power spectra has no indication of a definitive driver for mean minimum temperature.
References R. A., R. L. Kitching, F. Chiew, L. Hughes, P. C. D. Newton, S. S. Schuster, A. Tait, and P. Whetton. Climate change 2014: Impacts, adaptation, and vulnerability. Part B: Regional aspects. Contribution of Working Group II to the Fifth Assessment of the Intergovernmental Panel on Climate Change. Technical report, Intergovernmental Panel on Climate Change, 2014. URL https://www.ipcc.ch/report/ar5/wg2/. Bureau of Meteorology. Climate Glossary-Dro...
Hartigan, J, MacNamara, S & Leslie, L 2019, 'Trends in precipitation and temperature in Canberra', 23rd International Congress on Modelling and Simulation, International Congress on Modelling and Simulation, Canberra, pp. 986-992.View/Download from: Publisher's site
The current drought over much of southern and eastern Australia began not long after the Millennium Drought. While the Millennium Drought motivated Canberra to introduce measures to improve water availability, Canberra's population is increasing, placing greater strain on water resources. Further, its latitude is similar to other areas in the world in which drought frequency is increasing. Analysis of precipitation trends is required to assess how vulnerable an area might be to drought. In addition, if the mean temperature of a location is increasing, then the region might become more vulnerable to drought due to increased potential evaporation.
This statistical study utilises resampling methods to analyse trends in precipitation and mean temperature over Canberra. These resampling methods highlight the non-stationary nature of both precipitation and temperature time series. Minimal trends in precipitation were found, however there was an increasing trend in both mean maximum temperature (p-value = 0.0028) and mean minimum temperature (p-value = 0.0358) suggesting an increased vulnerability to drought for the region.
Numerous large-scale influences of climate such as the El-Nino Southern Oscillation (ENSO) have ˜ been used in the seasonal prediction of various atmospheric conditions. For example, an El-Nino ˜ event is typically associated with warmer and drier than average conditions over eastern Australia. Wavelet analysis provides a more thorough understanding of high- and low-frequency signals driving a non-stationary time series, and is applied here to identify potential drivers of the climate. Wavelet power spectra for precipitation reveal a statistically significant signal in the 2–7-yr range, which is typically indicative of influence by ENSO. Wavelet power spectra for mean maximum temperature reveal an increasing influence by ENSO, while mean minimum temperature reveals a decreasing influence. These findings exemplify how the influence of climate drivers ...