As deep learning-based image recognition integration into precision agriculture, it is promising to promote weather prediction and enable the farmers to take decisions based on crop yield and resource management. In this study, the research explores the use of convolutional neural networks (CNNs) to analyze meteorological imagery (satellite and drone imagery) to increase the precision of weather forecasting models. A new deep learning framework was engineered to capitalize on storage of large collections of historical weather conditions alongside agricultural outcomes to detect atmospheric signature and anomalies that conventional models may overlook. Preprocessing techniques, data augmentation to alleviate over fitting, and transfer learning using pretrained models for better feature representation are included in the methodology. Experimental results show that the proposed model improves prediction accuracy over conventional statistical approaches. Delivered real-time weather insights allow farmers to respond quickly to changing climatic conditions, thereby reducing the risks resulting from weather events. In addition, the implementation of the framework on mobile platforms demonstrates how it can be used on-field for timely decision-making. The findings of this research demonstrate the influence of interdisciplinary approaches on agriculture and lay the foundation for future developments enabled by integration between machine learning and remote sensing technologies. This study improves agricultural systems to be more resilient to climate variability through contributing to sustainable farming, resource conservation, and food security by enhancing weather prediction capabilities through advanced image recognition.