Forecasting Victorian spring rainfall using ENSO and IOD: A comparison of linear multiple regression and nonlinear ANN

Mekanik, Fatemeh; Imteaz, Monzur Alam

doi:10.1109/urke.2012.6319591

Cited by 17 publications

(16 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to Schepen et al (2012) a strong relationship between simultaneous climate modes and rainfall does not essentially mean that there is a lagged relationship as well. Of the few studies focusing on the lagged climate-rainfall relationship one can mention , Abbot and Marohasy (2012), Drosdowsky and Chambers (2001), Kirono et al (2010), Mekanik and Imteaz (2012), Schepen et al (2012) and . Thus, the objective of this study is to investigate the relationship of combined ENSO and IOD lags on Victoria's spring rainfall, as a case study.…”

Section: Introductionmentioning

confidence: 99%

Analysing lagged ENSO and IOD as potential predictors for long-term rainfall forecasting using multiple regression modelling

2013

Piantadosi, J., Anderssen, R.S. And Boland J. (Eds) MODSIM2013, 20th International Congress on Modelling and Simulation

View full text Add to dashboard Cite

Several climate indices around Australia were found to have strong correlations with south-east Australian seasonal rainfalls. Any such correlation with lagged climate indices and seasonal rainfall afterwards can be used for forecasting long-term seasonal rainfall. In this study, long-term forecasting of Victorian spring rainfall has been investigated using lagged El Nino Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD) indices using multiple regression analysis. Three stations of Buchan (east VIC), Malmsbury (Central VIC) and Kaniva (West VIC) were chosen as case study. Rainfall was classified according to El Nino, La Nina and Neutral years of ENSO, and also positive and negative years of IOD. It was discovered that categorizing rainfalls based on the years of ENSO and IOD do not have significant effect on its relationship with these climate modes. It was also found that the Pearson correlation coefficient between ENSO and Buchan in east Victoria is very weak; for Malmsbury and Kaniva ENSO indicators are showing higher correlations compared to Buchan. DMI effect is stronger in these two regions as well.Using the non-classified rainfalls, correlation coefficient between spring rainfall at year n and Dec n-1 -Aug n monthly values of ENSO and IOD indicators (Nino3.4, SOI and DMI) were calculated ("n" being the year for which spring rainfall is being predicted); It was discovered that only the three months of June, July and August of Nino3.4, SOI and DMI have significant correlation with spring rainfall. Several multiple regression models were investigated using lagged ENSO and IOD as potential predictors of spring rainfall; the models that satisfied the limits of statistical significance and multicollinearity were used to forecast spring rainfall three consecutive years in advance. Multiple regression analysis showed poor results in regards to forecasting ability in east Victoria, however it was able to forecast spring rainfall three consecutive years in advance for central and west Victoria with a correlation of 0.48 and 0.67 respectively.

show abstract

Section: Introductionmentioning

confidence: 99%

Analysing lagged ENSO and IOD as potential predictors for long-term rainfall forecasting using multiple regression modelling

2013

Piantadosi, J., Anderssen, R.S. And Boland J. (Eds) MODSIM2013, 20th International Congress on Modelling and Simulation

View full text Add to dashboard Cite

show abstract

“…The ANN model used in this study essentially mines historical data, including rainfall, atmospheric temperatures and climate indices, for patterns to produce a desired output that is a quantitative (rather than probabilistic) rainfall forecast for a specific future time period. ANNs have been used to forecast rainfall in many parts of the world [18][19][20][21][22], but rarely used in Australia [23,24] and never to generate official seasonal rainfall forecasts.…”

Section: Forecasting Rainfall Using Artificial Neural Networkmentioning

confidence: 99%

Using lagged and forecast climate indices with artificial intelligence to predict monthly rainfall in the brisbane catchment, Queensland, Australia

Abbot¹,

Marohasy²

2015

Int. J. SDP

View full text Add to dashboard Cite

Brisbane, the capital of Queensland, Australia, has flooded periodically and catastrophically, most recently in January 2011. Official seasonal rainfall forecasts failed to predict the floods. Since winter 2013, the Australian Bureau of Meteorology uses a general circulation model, the Predictive Ocean Atmosphere Model for Australia (POAMA), to make official seasonal rainfall forecasts presented as the conditional probability of rainfall being greater or less than the long-term median rainfall. We show that a more skilful forecast can be made using an artificial neural network (ANN), a form of statistical modelling based on artificial intelligence. A Jordan recurrent neural network with one hidden layer was implemented, using genetic optimization of inputs. For the sites of Gatton and Harrisville, in the Brisbane catchment, monthly rainfall forecasts from the ANN show lower root mean square errors than forecasts from POAMA. These rainfall forecasts from the ANN model were further improved by using inputs of independently forecast values for climate indices including the Southern Oscillation Index, the Interdecadal Pacific Oscillation, Pacific sea surface temperature anomalies (Niño 3.4) and also atmospheric temperature. The results presented here represent a first attempt at independently forecasting climate indices using an ANN model for the Australian east coast.

show abstract

“…Given this, many researchers have investigated this across the Australian continent (Ashok et al 2003a;Cai et al 2012;McBride and Nicholls 1983;Power et al 1998;Risbey et al 2009;Wang and Hendon 2007) or separately for different parts of Australia (Kirono et al 2010;Mekanik et al 2012;Nicholls 2010;Piechota et al 1998;Shi et al 2008). In some regions and seasons, climate indices did not show strong correlation with seasonal and monthly rainfall and were weak predictors for rainfall forecasting (Kirono et al 2010;Schepen et al 2012a).…”

Section: Introductionmentioning

confidence: 99%

Spatiotemporal monthly rainfall forecasts for south-eastern and eastern Australia using climatic indices

Montazerolghaem

Vervoort

Minasny

et al. 2015

Theor Appl Climatol

View full text Add to dashboard Cite

Knowledge about future rainfall is important for agriculture management and planning in arid and semi-arid regions. Australia has complex variations in rainfall patterns in time and space, arising from the combination of the geographic structure and the dual effects of Indian and Pacific Ocean. This study aims to develop a forecasting model of spatiotemporal monthly rainfall totals using lagged climate indices and historical rainfall data from 1950-2011 for south-eastern and eastern Australia. Data were obtained from the Australian Bureau of Meteorology (BoM) from 136 highquality weather stations. To reduce spatial complexity, climate regionalization was used to divide the stations in homogenous sub-regions based on similarity of rainfall patterns and intensity using principal component analysis (PCA) and K-means clustering. Subsequently, a fuzzy ranking algorithm (FRA) was applied to the lagged climatic predictors and monthly rainfall in each sub-region to identify the best predictors. Selected predictors by FRA were found to vary by sub-region. After these two stages of pre-processing, an artificial neural network (ANN) model was developed and optimized separately for each sub-region and the entire area. The results indicate that climate regionalization can improve a monthly spatiotemporal rainfall forecast model. The location and number of sub-regions were important for ranking predictors and modeling. This further suggests that the impact of climate variables on Australian rainfall is more variable in both time and space than indicated thus far.

show abstract

Forecasting Victorian spring rainfall using ENSO and IOD: A comparison of linear multiple regression and nonlinear ANN

Cited by 17 publications

References 20 publications

Analysing lagged ENSO and IOD as potential predictors for long-term rainfall forecasting using multiple regression modelling

Analysing lagged ENSO and IOD as potential predictors for long-term rainfall forecasting using multiple regression modelling

Using lagged and forecast climate indices with artificial intelligence to predict monthly rainfall in the brisbane catchment, Queensland, Australia

Spatiotemporal monthly rainfall forecasts for south-eastern and eastern Australia using climatic indices

Contact Info

Product

Resources

About