An optimized network for drought prediction using satellite images

The change in climate and the hot temperature environment increased the risk of drought around the workplace. Predicting and forecasting the drought occurrence is essential for managing water resources and agricultural plans. Therefore, in this study, a novel Chimp-based Wide ResNet Prediction Framework (CWRPF) is designed to predict the drought. The key motive of the presented research is to predict the drought and no drought conditions derived from the satellite images. The satellite images are collected from the Bhuvan site. Initially, the satellite images are noise-filtered. The filtered images are then injected into the feature analysis phase to compute the drought indices of a specific area by the fitness function activated in the framework. After estimating the drought indices, the drought condition was categorized. Finally, the designed system is tested in the MATLAB platform and has gained more significant results by providing a 97.68% accuracy rate, R2 as 0.998, and lower RMSE and MAE values of 0.223 and 0.193. The accumulated results are compared with existing techniques to validate the improvement score. The accuracy of the CWRPF is more remarkable than that of other prediction models. Therefore, the system is efficient for drought prediction in satellite images.