The authors present a methodology to downscale GHI retrievals from MSG imagery which is validated againts the HOPE field campaign (a set os 99 solar radiation sensors distributed along 10x12 km2). The paper is rather good and the results also are. I think that it has enough quality to be accepted for publication and also interest in the community. I have only a few minor comments and/or doubts.

1) It is not quite clear to me what is the time scale of the GHI retrievals, since Meteosat Second Generation imagery is disseminated in 15-minutes how the authors estimate 5-minute GHI?

2) Eq 2) would need a clearer explanation, I don't see why the use of residuals istead of reflectance and it is not completeley clear (at least to me) how this ethod is used or affect to the GHI retrieval. How afect the reflectance?

3) I think that it would be eneficial a scheme or flow diagrama explaining or indicating the different steps and products used in the methodology in order to get easier for the reader the understanding.