Drought modelling is essential to managing water resources in arid regions to limit its impacts. Additionally, climate change has a significant effect on the frequency and intensity of drought. This research provides a novel approach to forecasting the standardised precipitation index (SPI 3), considering several climatic variables by employing hybrid methods including (i.e., data pre-processing represented by normalisation, cleaning (i.e., outliers and Singular Spectrum Analysis), and best model input (i.e., tolerance technique), in addition to, artificial neural network (ANN) combined with particle swarm optimisation (PSO)). The data on climatic factors were applied to build and evaluate the SPI 3 model from 1990 to 2020 for the Al-Kut region. The result revealed that data pre-processing techniques enhance the data quality by increasing the correlation coefficient between independent and dependent variables; and choosing the optimal input model scenario. Also, it was found that the PSO algorithm precisely predicts the parameters of the proposed model. Moreover, the finding confirmed that the supposed methodology precisely simulated the SPI 3 depending on several statistical criteria (i.e., R², RMSE, MAE).