Synthetic aperture radar (SAR) images acquired over the ocean often show radar signatures of rain, which are not easy to interpret. The scattering mechanisms causing radar signatures are usually attributed to surface scattering due to sea surface roughness variations caused by raindrops impinging onto the sea surface and/or by up- and downdraft winds. In this paper, we address another radar signature of rain, which is often observed in C-band (and also in X-band) SAR images, but whose origin has been a matter of debate in the ocean remote sensing community since long time and has not been solved yet. This radar signature consists of areas of very high radar backscatter (bright patches) at co- as well as well as at cross-polarization. This paper aims at providing evidence that it is not caused by surface scattering, but by volume scattering from wobbling, non-spherical, oblate hydrometeors within the melting layer. To this end, we first review the theory of radar backscattering from the melting layer as developed by D'Amico et al. (1998) and then present historic radar backscatter data from the melting layer carried out by ground-based and airborne radars, which validate this theory. Then we show four representative Sentinel-1 SAR images acquired over the sea area close to Hong Kong and a SIR-C/X-SAR image acquired over the Gulf of Mexico, which show pronounced radar signatures of rain (bright patches) at co-polarization (VV) and cross-polarization (VH). The analysis of the SAR images yields the result that within the bright patches the ratio of the radar backscatter at cross-polarization to the one at co-polarization shows the same characteristics as the linear depolarization ratio (LDR) measured by radar meteorologist in radar backscattering from the melting layer. Furthermore, we show that radar signatures of rain due to volume scattering may interfere with co-polarization radar signatures of rain due to surface scattering. Thus, cross-polarization SAR images are better suited to detect radar backscattering from the melting layer than co-polarization SAR images, This investigation is of relevance for ocean surface wind retrieval using C-band SARs, since scattering at hydrometeors in the melting layer can cause significant errors in ocean wind retrieval. Areas with simultaneously high co- and cross-polarization NRCS values of around −10 dB and − 20 dB, respectively, have to be flagged as areas where the conventional wind retrieval algorithm cannot be applied.