Synthetic aperture radar (SAR) image positioning is commonly affected by factors such as platform instability, aging of onboard instruments, and changing observation environments. Thus, geometric calibration is needed to improve the accuracy of image positioning before mapping and application. An improved geometric calibration model for SAR images was developed, which does not require long-delayed meteorological data and geometric calibration fields. In this method, the standard atmospheric and Saastamoinen models (SAM-S) are combined for atmospheric delay correction, and this integrated method named SAM-S was used for the first time in SAR image geometric calibration. The method was applied to China's Gaofen-3 satellite and a case study was conducted in the Yangtze River Delta, where the weather is cloudy and rainy. By selecting three images, four images, and five images from eight candidate images with a bandwidth of 60 MHz for calibration experiments, where the geometric positioning accuracy was increasingly stable with more SAR images included in the combination. The root mean square of the calibrated SAR images was about 6 m for different combinations and the maximum error was about 12 m for all images. Two selected areas showed that the geometric calibration has reduced the geometric distortion, showing shapes in the SAR images close to the ground targets. The positioning accuracy improvement of Gaofen-3 images can help improve their application potential in global remote sensing mapping, land-use change monitoring, and ground target detection. The proposed geometric calibration method can also be applied to other SAR missions such as ERS, RADARSAT, and TerraSAR-X.