Evaluation of Hybrid Wavelet Models for Regional Drought Forecasting

Drought forecasting is essential for risk management and preparedness of drought mitigation measures. The present study aims to evaluate the effectiveness proposed hybrid technique regional forecasting. Multiple Linear Regression (MLR), Artificial Neural Network (ANN), two wavelet techniques, namely, Discrete Wavelet Transform (DWT) Packet (WPT), were evaluated in up a lead time six months. Standard error metrics used select optimal model parameters, such as number inputs, hidden neurons, level decomposition, mother wavelets. Additionally, performance various wavelets, including Haar (db1) 19 Daubechies wavelets (db1 db20), evaluated. results indicated that ANN produced better forecasts than MLR model, whereas models outperformed both models, which failed predict SPI values greater all was found improve timescale increased from 3 12 However, models’ performances deteriorated increased. WPT-MLR best area. findings could be early warning systems