Abstract:Theτ-p transform is useful method for seismic data interpolation. But due to lack of enough information of seismic data and limited aspect as well as digitalization etc in practice, the false and incorrect phenomenon often appear in seismic data interpolated inτ-p domain. The studied results showed it is important to reconstruct the absent seismic information when implementing sparse processing of seismic data inτ-p domain, i. e. by cor- rectly choosing the samples of discrete parameters. For that reason, the paper presented a method of seismic traces interpolation based on sparse discreteτ-p transform. On the basis that the posts-tack zero-offset sections can be seen as a composition of a series of linear events, the paper uses the sparse discreteτ-p transform and precondition double conjugate gradient algorithm to interpolate seismic traces. Theoretical data tests and real data processing proved that the method could increase the density of samples of seismic data in spatial direction that effectively removes the spatial aliasing.