Abstract This paper emphasizes the significance of adopting a robust approach for effectively handling spatially heterogeneous distributed data. We present a dedicated workflow focusing on enhancing the processes involved from acoustic reflectors data to surface and mesh definitions. Results derived from the robust approach are compared to the ones generated following a "reference approach". These distinct approaches are then employed as inputs for separate 3D subsurface models for a reciprocal comparison. The data on which this study is based pertains to the interpretation of specific acoustic reflectors from a geophysical survey typically related for subaqueous and sub-seabed engineering assessment. Generally, these data exhibit a significantly uneven distribution, characterized by abundant data along the acquisition lines and data absence in the space between lines. This distinctive distribution emphasizes the need to carefully consider the appropriate strategy during the surface creation phases. The workflow commences with robust data simplification applied along seismic lines, followed by mesh refinement to improve surface definition. The resulting mesh exhibits a non-uniform resolution, avoiding uncontrolled sub-sampling, accommodating higher data density in areas with greater spatial variability and vice versa. Comparing proposed approaches output reveals a strong level of correspondence, underscoring the reliability and efficacy of the proposed methodology. The benefits of employing proposed ad-hoc meshing criteria are evident in more robust sub-sampling of original interpreted reflector data, resulting in lighter mesh files and reduced computational time during mesh editing and successively 3D modelling phases, all without sacrificing crucial original information.