ArDHO-deep RNN: autoregressive deer hunting optimization based deep recurrent neural network in investigating atmospheric and oceanic parameters

Abstract

In recent decades, weather prediction results in a challenging and complex task in various disciplines. However, prediction of the atmospheric parameters is significant in various applications, like climate monitoring, agriculture and production, aviation industry, pollution dispersal, and drought detection. Due to the dynamic nature of the atmosphere, accurate prediction of the weather parameters is more complex. Sea Surface Temperature (SST) is the major factor that affects ocean climate. Hence, it is more important to analyze the atmospheric and oceanic parameters, like SST, Sea Surface Height (SSH), wind velocity, and soil moisture, for forecasting the weather. Hence, this research has developed an effective method named Autoregressive Deer Hunting Optimization (ArDHO)-based Deep Recurrent Neural Network (Deep RNN) for investigating the atmospheric and oceanic parameters. The technical indicators associated with the atmospheric data are effectively extracted. The process of data augmentation enabled to achieve higher dimensionality of data, as the deep learning classifier is more effective in generating optimal results with the high dimensional training samples. Finally, the Deep RNN classifier is employed to make the prediction strategy, and the training process of the classifier is done with the proposed optimization algorithm. However, the proposed method achieved higher performance by obtaining minimal MSE, RMSE, and speed value of 0.0524, 0.2288, and 31.25 and the maximal model size of 23.