Optimised deep convolutional spiking neural network for accurate long-term and short-term rainfall forecasting in climate prediction systems

The rainfall forecast is essential to the fields of hydrology and meteorology. However, the prediction accuracy of existing methods for both shorter and longer-term rainfall forecasting is consistently low. The decreased performance of atmospheric forecasting models under various circumstances causes fluctuations in predicting accuracy. To address these, this paper proposes a novel method called deep convolutional spiking neural network optimised with sandpiper optimisation algorithm fostered long-term and short-term rainfall forecasting (RP-DCSNN-SPOA). The primary source of the long and short-term rainfall (LSTR) data is the Sudan IRA rainfall forecast dataset. Then, the gathered data is pre-processed using anisotropic diffusion Kuwahara filtering to recover the missing values. The DCSNN is used to predict the rainfall forecast. Then, the sandpiper optimisation algorithm (SPOA) is used to enhance the DCSNN classifier that accurately forecasts the rainfall. The proposed method achieves 28%, 22.64% and 28.35%, greater accuracy, 20%, 26.64% and 23.35% greater precision when compared with existing models.