Steganographic payload location is primarily used to estimate the embedding location of steganographic messages. Existing JPEG image steganographic payload location methods predominantly rely on analyzing multiple stego images with the same embedding path. However, it is likely that forensics investigators will not be able to obtain various stego photos with the same embedding paths, but only one single stego image, resulting in a very low location accuracy that is close to random guessing. This paper proposes a steganographic payload location algorithm based on Gabor residual fusion of co-frequency sub-images. When only having a single stego image, 2D Gabor filters with different scales, orientations, and phase parameters are used to filter the co-frequency sub-images, generating multiple residuals at each position in the stego image. Then, an ensemble classifier is utilized to fuse the various residuals at the same position to judge whether the position has been modified. Experimental results show that the proposed algorithm can locate the modified coefficients with higher accuracy than others even under the condition of only a single stego image. Specifically, when the modification ratio is as low as 0.1 and the location accuracy of other payload location algorithms is lower than 10%, the proposed algorithm based on co-frequency sub-image Gabor residual can still locate the modified coefficients with accuracy close to 80%.