Dynamic SG-SKRDX hybrid framework for precision weather forecasting and crop suitability in the Cauvery Delta

The agricultural sector in Tamil Nadu's Cauvery Delta Region, a crucial contributor to the state's economy and a primary source of livelihood, is experiencing declining productivity due to shifts in weather patterns and soil conditions. Although Indigenous Technical Knowledge (ITK) systems in agriculture are context-specific and resource-based, the evolving climate and soil characteristics necessitate the integration of modern technologies with traditional practices to promote climate resilience and sustainable crop production. Addressing this, the manuscript explores the use of statistical models, Machine Learning (ML), and Deep Learning (DL) for region-specific weather forecasting and crop recommendation. However, predicting future Weather dynamically and offering adaptive crop suggestions remains a significant challenge. This article introduces the Dynamic SG-SKRDX model, a hybrid approach that integrates crop recommendations with futuristic weather forecasts. The model predicts current and future weather conditions in the Cauvery Delta Region using a Support Vector Regressor (SVR)-Gated Recurrent Unit (GRU) (SG) model, which has been enhanced through hyperparameter tuning and cross-validation. This SG model forecasts temperature, humidity, and precipitation for the delta regions using ten years of historical meteorological data. For crop recommendation based on these forecasts, the study employs a dynamic ensemble of ML models - Support Vector Machine (SVM), K Nearest Neighbour (KNN), Random Forest (RF), Decision Tree (DT), and eXtreme Gradient Boosting (XGBoost) - termed Dynamic SKRDX. This dynamic ensemble intelligently selects the best-performing ML models based on changes in the predicted weather variables. Evaluation of the Dynamic SG-SKRDX model reveals strong performance, with the SG weather model achieving 0.65% MSE, 8.07% RMSE, 4.69% MAE, and 65.49% R-Squared. The dynamic SKRDX crop recommendation model attained 93.41% accuracy, 93.72% precision, 93.41% recall, and a 93.33% F1-Score. The proposed Dynamic SG-SKRDX model demonstrates superior capabilities in forecasting future Weather and recommending region-specific crops in the Cauvery Delta Region, offering a pathway towards technology-driven, sustainable agriculture and improved economic stability for the region.