Abstract: Surface-related multiple elimination(SRME) is vital in processing seismic data.At present, the SRME method most commonly used for processing marine data can effectively predict surface-related multiple models, achieving multiple elimination through matching and subtraction.However, this method requires full wave field information, thus placing higher demands on observation systems.The irregular land data acquisition makes it difficult to apply this method for processing land data.Hence, this study presented an improved SRME method, which performs moveout correction on adjacent channels for multiple prediction by constructing the minimum error constraint function, thus enhancing the adaptability to complex observation systems.The improved SRME method was applied to process the actual seismic data of land.Its effectiveness was validated by comparing the gathers, velocity spectra, prestack time migration profiles, and synthetic seismograms before and after SRME.