Poststack seismic inversion plays an important role in prediction of oil and gas reservoirs. In poststack seismic impedance inversion, precise determination of the source wavelet remains a critical yet challenging task, particularly when handling field seismic datasets. To address this limitation, we propose a wavelet-independent inversion framework through the development of a convolutional model-driven cost function. The methodology operates by minimizing the discrepancy between two convolutionally derived components: 1) the product of synthetic seismic traces with actual seismic measurements at calibrated well positions, and 2) the convolution of field observations with numerically simulated traces from the same well locations. This reformulated approach inherently eliminates source wavelet dependency through mathematical construction while requiring appropriate seismic data preconditioning. The designed cost function demonstrates enhanced stability against data contamination through two mechanistic advantages: Firstly, the synthetic traces act as spectral filters that suppress out-of-band noise components during convolution operations. Secondly, the linearity of the formulated problem permits efficient resolution via standard conjugate gradient optimization techniques without requiring complex regularization schemes. Validation through comprehensive testing, including both synthetic benchmarks and field case studies, confirms the method's insensitivity to source wavelet inaccuracies and its improved robustness against random noise interference compared to conventional inversion approaches.

Seismic Inversion, Acoustic Impedance, Objective Function, Source Independent