To my opinion the authors have answer to the basic comments of the reviewers.

I personally have a general comment concerning the data used for the specific area.

Lets start from the fact that the most (or all but one) BSRN stations used for the particular SARAH product evaluation are not representative for the atmospheric conditions of the area under study here. And by representation I mean the aerosol load and variability in AOD and aerosol types.

Aerosols play in this area a much more important role than mid latitude European stations (which are the majority of the evaluation study) where compared with W. African area AOD is much lower and less variable. In addition, total GHI is much lower too.

The three stations used for evaluation in this manuscript provide an evidence on that as statistics show much higher MBEs etc than what is described in the introduction. 5.5 and 17.2 W/m2.

So even if the authors have answered on the reviewer comment about uncertainties I think that the statement "the good correlation and the uncertainty dominated by the bias...supports the suitability of the data to investigate the solar variability" is only partly correct.

More importantly if you talk about trends in the order of 0-2 W/m2 per decade , the AOD differences (real vs modeled/used) will play an important role on these numbers. For example authors can discuss the trends calculated by the 3 the ground based stations. Do they agree ?

Years with higher AOD loads and others with lower are partly taken into account in the analysis using the data as presented here, causing a large uncertainty in the trend calculation. As the aerosol load plays an important role in the GHI in this area.

In addition the area is a unique place concerning aerosol types. Dust, marine, anthropogenic, biomass burning are all present depending on meteorology and season. Other aerosol optical properties (absorption , spectral dependence of AOD) could be also important here.

I would suggest to try to expand the discussion on the trend validity and uncertainty and the representation of aerosol effects in the current data set for the particular area.

best regards
SK

The article is really interesting and deals with an important aspect on an area where not a lot of measurements exist. An area with high population that solar energy studies are important for the energy sector.

The changes from all comments have been included in the new manuscript.
Aerosol related analysis is not the main scope here, however there is a lot of scientific information here for a future study dealing with this aspect.
In general solar radiation changes are mainly due to clouds and aerosols and cloudiness and aerosol changes from different sources could be used in order to investigate the sensitivity of the calculated trends and their validity.

Just a short comment about the new DAOD vs DGHI figure it is that is a nice figure however short of quantitative approach as changes in monthly GHI are due to solar elevation, cloudiness and AOD so in order to directly comment on the aerosol (used) impact, the other factors (solar elevation, cloudiness) have to be eliminated (cloudiness = 0) or normalized (solar elevation to be constant).

congratulations on this very interesting work