This study aimed to develop a Soil Fertility Index (SFI) map for agricultural lands in central Khuzestan Province, specifically focusing on irrigated wheat cultivation. Three scenarios were evaluated using Random Forest (RF) models based on hyperspectral EnMAP and multispectral Sentinel-2 satellite data. The SFI was calculated by integrating eight key soil fertility parameters: Soil Organic Matter (SOM), Total Nitrogen (TN), available phosphorus (P_av), exchangeable potassium (K_ex), Electrical Conductivity (EC), Iron (Fe), calcium carbonate equivalent (CCE), and soil texture classes. In Scenario 1, the RF model using EnMAP data demonstrated excellent prediction accuracy (R² = 0.84, NRMSE = 0.09). Scenario 2, based on Sentinel-2 data, also performed well but with slightly lower accuracy (R² = 0.72, NRMSE = 0.11). The highest model accuracy was achieved in Scenario 3, which fused EnMAP and Sentinel-2 data, improving spatial resolution to 10 m (R² = 0.90, NRMSE = 0.07). This scenario provided the most accurate prediction of soil fertility distribution. The results indicated that 60.6% of the study area fell into the moderate fertility class. Key limitations in soil fertility include high CCE content and low levels of SOM, TN, and P_av. The findings underscore the importance of using integrated remote sensing data for precision agriculture and sustainable soil management, offering valuable insights for policymakers and agricultural managers.