In the paper we consider complex land near surface modeled as random clutter, and we study the depth imaging problem for such media. We consider synthetic clutter models, and we provide a numerical study using various realizations of the cluttered models. We propose and investigate a statistical imaging technique based on path summation allowing to get depth image without reconstruction of deterministic near-surface depth velocity model. Our results demonstrate that statistical imaging can significantly improve and recover reflectors, even with a limited number of realizations. The ability to achieve accurate images despite the absence of an exact velocity model underscores the robustness of this technique. We show that statistical imaging can average out phase distortions caused by near-surface clutter, enhancing the clarity and accuracy of the subsurface image. This method offers a promising direction for imaging in complex near-surface environments, effectively addressing the limitations of traditional deterministic approaches. We further explain why this technique works by leveraging the concept of “seismic speckle,” analogous to optical and ultrasonic speckle noise, to account for phase distortions caused by small-scale heterogeneities. By estimating the statistical properties of the clutter, we generate ensembles of velocity models that perturb the phase of the depth-migrated arrivals symmetrically. In summary, our study highlights the potential of statistical imaging with path summation to significantly improve seismic imaging in challenging near-surface environments, offering a robust alternative to traditional deterministic methods and paving the way for future advancements in subsurface imaging.