In reverse time migration, if sources or receivers are buried at certain depths, the interference between the primary wave and the free‐surface reflection causes losing spectral contents in seismic data. If they are not properly compensated, errors will be carried to the target during the reverse time migration and cause a distorted image by so‐called ghost image. By invoking the point spread function, we investigate how distorted seismic data are mapped to the subsurface. Quantitative relations between the free‐surface parameters (i.e. source/receiver depth, free‐surface incident angle and near‐surface velocity) and wavenumber‐domain illumination can be created, which map missing frequency contents in seismic data to missing wavenumber contents in the point spread function. After transformed back to the space domain, we can investigate its effect on the image distortion. Numerical calculations expand the above approach to handle complex overburden structures and more realistic acquisition geometries. The proposed method provides us with a useful tool to investigate deghosting related problems, for example optimizing parameters of acquisition geometry, evaluating capabilities of existing acquisition systems and data processing technologies in suppressing the free‐surface reflection effect in the depth image, or developing new deghosting approaches.