Extreme weather events such as storms and floods are increasing in frequency and intensity due to climate variability, creating a strong need for accurate and trustworthy prediction systems. This paper proposes a Blockchain-Enabled Convolutional Neural Network (CNN) Framework that leverages EGC2A (Enhanced Grid-Cell Climate Analytics) temporal outputs to improve the reliability, security, and interpretability of hydrometeorological forecasting. The EGC2A model generates high-resolution temporal climate features—such as rainfall intensity, atmospheric pressure shifts, and multi-layer humidity variations—which are fed into a customized CNN architecture to predict storm onset, flood severity, and risk distribution. To ensure tamper-proof data integrity, transparency, and decentralized verification, all processed climate features, model inference logs, and alert events are securely stored on a private Ethereum-based blockchain. The integration of blockchain significantly reduces data manipulation risks, ensures traceability of prediction updates, and enhances the trustworthiness of climate-alert systems used by disaster management authorities. Experimental evaluation demonstrates that the proposed hybrid framework achieves higher prediction accuracy, reduced false-alert rates, and improved temporal sensitivity compared to traditional deep-learning-only models. This work establishes a scalable architecture for future climate prediction systems where AI-driven analytics and blockchain-based trust operate cohesively to support proactive disaster preparedness.