Convolutional Bi-GRU-based prediction model for Climate Indicators

Climate prediction is critical to a wide range of fields, from agriculture to public health, but traditional methods face challenges due to the complexity and time-space characteristics of climate data. Machine learning techniques provide a promising solution by effectively extracting features from climate indicator data. In this study, we propose a new method for predicting climate indicators by combining CNN-Bi-GRU models with FNN networks. Through evaluation, our method demonstrates excellent prediction performance in terms of RMSE and MAE on both test and training sets, and robust convergence behavior is observed in model training. In addition, our method achieves high accuracy in predicting key climate metrics such as temperature and rain frequency, providing valuable information for understanding climate dynamics. Comparison experiments with LSTM-FNN, Double-LSTM, BiLSTM, and Bi-GRU models highlight the superiority of the CNNBi-GRU approach, especially in reducing temperature prediction errors. In conclusion, this study presents a promising approach to climate prediction that can enhance our understanding of climate variability and support informed decision making in various industries.