A two- fold application of wavelet analysis is demonstrated in the proposed approach. Firstly, Multi- scale Wavelet Entropy i.e. MWE in conjunction with K- means and Principal Component Analysis (PCA) is used for selecting the suitable input variables. Secondly, Discrete Wavelet Transform (DWT) is employed to obtain the wavelet sub-time series for the selected input variables. The decomposed sub time series of different climate variables are then combined using the MLR or SoV model at each decomposition level and final outcome is added to obtain the simulated time series.
The proposed methodology is explained through the following schematic: