<![CDATA[Drought is one of the hydrometeorological disasters that has a significant impact on the agricultural sector, water availability, and food security, thus requiring an accurate and adaptive early warning system. This study aims to develop and evaluate a drought Early Warning System (EWS) model based on a combination of multi-variable agrometeorological parameters using a hybrid approach of Singular Spectrum Analysis (SSA) and Adaptive Model-Based Prediction (AMBP). The agrometeorological data used includes rainfall, air temperature, humidity, solar radiation, wind speed, and other supporting variables processed in the form of monthly time series over a period of ten years. The SSA method is used to perform signal denoising and extract dominant components from the data, while AMBP is applied as an adaptive predictive model to generate SPI-6 drought index forecasts. Model performance is evaluated using RMSE and the coefficient of determination (R²) in the model training and evaluation phases. The results show that the hybrid SSA–AMBP model has the best performance compared to single methods, with an RMSE value of 0.149 and R² of 0.983 in the model training phase, and an RMSE of 0.176 and R² of 0.941 in the model evaluation phase. In addition, the 2026 prediction results show a seasonal pattern with indications of a moderate dry period from October to November. These findings indicate that the developed model demonstrates relatively high predictive accuracy and stability based on RMSE and R² evaluation metrics within the SPI-6 dataset used in this study, and it has the potential to serve as a conceptual basis for decision-support in drought risk mitigation, water resource management, and sustainable agricultural planning.]]>
Copyrights © 2026
