The ionospheric radio electrical transmissions have multiple paths due to the inhomogeneity of the propagation medium, that is, the ionospheric layers. Tactical applications such as radiolocation and radiocommunications systems need blind, high-resolution identification of multipath channels. This work concerns the separation of the ionospheric paths and is based on recent work done on blind deconvolution which can estimate the impulse responses of a propagation channel. In this way, on the basis of a parametric model of the paths, we propose a blind, spatiotemporal identification of the propagation channel. The parameters that characterize the propagation model are the directions of arrivals (DOA) &thgr;, time delays &tgr;, and complex gains α (also called fading). We propose an algorithm that can both estimate the multipath parameters and test them on real life data. This new method needs fewer snapshots than other methods recently proposed, and thus can monitor more quickly varying channels. Moreover, compared to recent work we have relaxed the problem of making successive estimates of the impulse responses. The proposed method can also identify more paths than the number of sensors. An extension of the algorithm will be presented by including polarization diversity and thus increases the resolution. The proposed methods are illustrated on experimental data. ¿ 2000 American Geophysical Union