Seismic data inversion plays a central role in exploration geophysics and subsurface medium modeling, yet conventional single-trace methods frequently exhibit limited noise robustness and non-uniqueness, producing unstable solutions. To address these shortcomings, this paper proposes a multichannel basis pursuit inversion that incorporates spatial reflectivity structure constraints. By constructing a structural characterization matrix based on the spatial distribution of reflected wavefields and incorporating it into the regularization term of the inversion equation, the approach couples vertical sparsity priors with lateral structural coherence, forming a synergistic constraint that enhances both geological continuity and spatial correlation. The algorithm is designed to balance convergence and noise suppression. Numerical experiments and field data case studies demonstrate that the proposed method substantially improves noise tolerance, enhances vertical resolution, recovers more thin-bed information, and strengthens lateral continuity and consistency, yielding more stable and reliable results than traditional basis pursuit inversion. The proposed framework thus offers a practical pathway for reflection-structure-based inversion and advances subsurface imaging and interpretation.