Long-term drought forecasting using wavelet-neural networks and wavelet-support vector regression

Authors: Belayneh, A, J Adamowski, B Khalil
Description: Long term drought forecasts can be an effective tool for mitigating some of the more adverse consequences of drought. Data based models are suitable forecast tools due to their minimal information requirements and rapid development times. This study compares the effectiveness of five data based models for forecasting long term drought conditions in the Awash River Basin of Ethiopia. The Standard Precipitation Index (SPI) is forecast using a traditional stochastic model (ARIMA) and compared to machine learning techniques such as artificial neural networks (ANNs) and support vector regression (SVR). In addition, wavelet analysis is used since wavelet transforms have recently shown great ability in modeling and forecasting nonlinear and non-stationary time series in hydrologic forecasting studies. Wavelet transforms are used to pre-process the inputs for the ANNs and SVR models. The performances of all models were compared using RMSE, MAE, R2 and a measure of persistence. The forecast results indicate that the coupled wavelet neural network (WA-ANN) models were the best models for forecasting SPI values over long lead times in the Awash River Basin.
Keywords: drought forecasting, wavelet transforms, support vector regression
Technical field: technical_fields_app7
Session name: Soil and water
Date: 2012
Identifier: CSBE12017
Coverage: Canada

Search a publication

Search for:
Find entries that have:
Publication type: