Maritime electromagnetic (EM)‐based communication and detection systems are strongly influenced by meteorological conditions, as they can cause anomalous electromagnetic propagation within the surface layer. To predict the performance of such systems, detailed knowledge of the refractivity profile is required. In recent years, refractivity from clutter (RFC) methods has been developed to estimate this refractivity profile by measuring radar clutter return from the rough ocean surface. The current work proposes an RFC framework that utilizes a novel surrogate model for EM propagation. The surrogate model is based on an offline created library of sparsely sampled field data of clutter returns, compressed into proper orthogonal bases, and indexed on specific surface layer refractive parameters. By exploiting the Riemannian manifold structure of the space that proper orthogonal bases occur in, we are able to interpolate among them. This, then, enables us to use the surrogate model in an inverse problem setting, whose goal is to uncover in situ maritime EM propagation conditions efficiently. We demonstrate the feasibility of our proposed surrogate model‐based RFC approach for evaporation duct by testing it with field data obtained from an experimental campaign.