Underwater visual perception is degraded by wavelength-dependent absorption and scattering, the effects of which vary with the camera-to-scene distance. Many underwater image enhancement methods nevertheless apply image-level transformations or use region-wise statistics with discrete boundaries. Such control can make it difficult to preserve near-field appearance while improving far-field visibility, particularly when degradation changes gradually across the scene. We propose a Depth-Aware Underwater Image Enhancement (DUIE) framework that uses relative depth as a spatial prior for continuous enhancement control. DUIE first derives two complementary candidates through foreground- and background-specific statistical color compensation and refines both candidates with a lightweight Nonlinear Activation-Free (NAF)-based network. Its principal learned component jointly maps the input RGB image and an estimated relative depth map to a continuous pixel-wise fusion coefficient. A depth-weighted Laplacian pyramid then blends the refined candidates at multiple scales, replacing the binary mask fusion used by region-wise enhancement with smoother spatial control. The framework therefore combines a physics-inspired depth cue, interpretable statistical candidates, and learned local refinement without explicitly inverting a complete underwater image formation model. Experiments on three real underwater datasets show that DUIE obtains the best overall set of perceptual and edge-based scores among the compared methods, while qualitative results indicate balanced color correction, visibility, and structural detail.