In land seismic exploration, the scattering generated by small-scale heterogeneities in the shallow layer can cause strong noise in seismic data, resulting in serious problems in the depth image. By introducing random velocity models to simulate velocity fluctuations in the near-surface layer, and using the point spreading function to characterize the image quality, we examined how the surface scattering can affect the migration image. The results revealed that lacking of intermediate scaled velocity information in the shallow part of the migration velocity model can cause phase or travel time errors in the extrapolated source and receiver wavefields, making them fail to focus. This is the primary reason causing deteriorated image quality in regions with strong near-surface scattering. By investigating the effect of velocity fluctuations of different scales to wavefield focusing, we found velocity fluctuations of 200-500 m scale are vital to provide correct phase and travel times. If this part of the information can be obtained and built into the migration velocity, the subsurface image quality can be largely improved. Adopting the high-precision near-surface velocity model building technique, it is possible to create the near-surface migration velocity with the required accuracy. The current study can be the basis for further numerical investigations and the field experiment.