Throughout the last decades, climate change and its consequential impacts including changes in precipitation patterns and climate extremes have become one of the major areas of concern for researchers, leading to perilous and unforeseeable consequences for humans and the natural world. One of the challenging issues in statistical downscaling of General Circulation Models (GCMs) is to select dominant large-scale data (predictors). In this respect, correlation-based methods have revealed to be efficacious to select the predictors, yet traditional correlation analysis has shown limited ability due to the non-stationary and non-linear nature of climatic time series. Wavelet coherence analysis was employed to assess the high common powers and the multi-scale correlation between two time series as a function of time and frequency. The present study developed a methodology to identify the dominant predictors for statistical downscaling using continuous wavelet transform (CWT), wavelet transform coherence (WTC) analysis and Artificial neural network (ANN). CWT was utilized to identify the potent periodicity for predictands and predictors time series and WTC between predictors and predictands has been used to examine the correlation for the potent periodicity, and in order to determine the dominant predictors, Scale Average wavelet coherency has been applied. The results show that predictors related to geopotential height (zg), eastward wind (ua), temperature (tas), air pressure (psl) and humidity (hus and huss) are the dominant predictors affecting Tabriz precipitation. Three GCMS including CAN-ESM2, BNU-ESM and INM-CM4 has been used in this study. The proposed methodology could decline the dimensionality of large dataset as well as selecting the potential predictors among a numerous number of them.
Anahtar Kelimeler: climate change, general circulation models (GCMs), wavelet transform coherence (WTC), Tabriz city rainfall