A severe drought may have devastating effects on a region’s hydrological balance, agriculture, animal habitat, and economy, among other things. As a result, a reliable method of drought forecasting is required. Several drought indices purport to measure drought severity, but many of them ignore crucial details. This research presents a fog-based paradigm for drought predictive analysis, made possible by the IoT. Singular vector decomposition is used to compress data at the fog level in this architecture. Future drought conditions are predicted using the Holt-Winters approach, while the severity of individual drought occurrences is evaluated using an Adaptive Neuro-fuzzy Inference System (ANFIS) mechanism. The usefulness of the suggested system is proved by its implementation utilizing the online UCI dataset in terms of forecasting efficacy, and Prediction efficiency.