Northwestern Bangladesh faces frequent droughts and extreme weather, which seriously affects farming and reduces crop production. To deal with this, we need to understand how temperature changes, rainfall, and plant health interact. Traditional drought indicators like SPI (based on rainfall) and NDVI (which measures plant health from satellite images) help, but they often miss early warning signs and small-area droughts. This study suggests a better way to monitor drought by combining these traditional methods with advanced soil moisture data from GNSS-Reflectometry and the SMAP satellite. We used rainfall data to calculate SPI and NDVI, and added soil moisture data from GNSS and SMAP. Then, using machine learning, we linked these indicators to crop yield data. The results were impressive. Our model became much more accurate R² increased from 0.5 to 0.7–0.8, and prediction errors dropped by 20%. The system detected drought in Bogra and Ishurdi and showed long-term dryness in the Barind area. GNSS and SMAP data were able to detect early drought signs well before any visible damage to crops appeared. This improved method can help Bangladesh better prepare for drought, protect food supplies (SDG 2), and support climate action (SDG 13) for more resilient agriculture.