Using Artificial Intelligence To Forecast Monthly Rainfall Under Present And Future Climates For The Bowen Basin, Queensland, Australia
Free (open access)
Volume 10 (2015), Issue 1
66 - 75
J. ABBOT & J. MAROHASY
There is a need for more skilful medium-term rainfall forecasts for the Bowen Basin, a key coal-mining region in Queensland, Australia. Prolonged heavy rainfall during the 2010–2011 summer was not forecasted and it severely affected industry operations. Official forecasts are currently based on general circulation models (GCMs) and indicate there will be change in the timing and strength of the rainfall in the Bowen Basin with climate change.
A more skilful medium-term rainfall forecast for the present climate can be achieved through the use of artificial neural networks (ANNs). ANN can be used to generate monthly forecasts 3 months in advance. These forecasts can be improved through a weighted linear combination of forecasts. Principal component analysis prior to inputting data does not improve the forecast. An ANN can provide an independent method of GCM validation under future climates with results in reasonable agreement with the averaged values from the GCM ensembles: suggesting a decline in summer rainfall and an increase in winter rainfall at Nebo, a locality in the Bowen Basin, under the 3°C warmer scenario. This represents a smoothing of the annual variability in rainfall for the locality of Nebo rather than more climatic extremes with global warming.
Rainfall, artificial neural network, climate change, model independence