Dear Rosa,

I would like to congratulate you for the very extensive and detailed analysis of the EPN-Repro2 dataset. I think that you tackled the major issues related to the processing strategies in the paper, and the assessment of the resulting EPN-Repro2 ZTDs with independent datasets (radiosondes and ERA-interim), both for the biases and the trends is well established in general.

However, I think that the presentation quality (I propose to switch the part on the ZTD trend differences from the conclusions to another subsection in section 4) can be better. You should first investigate some more time in the section that describes the ZTD biases between radiosondes and GPS. The sign of the biases really matters (therefore, you should also explain how the biases are calculated: GPS minus RS or RS minus GPS), and as far as I can extract this information for the paper, the biases are in agreement with the dry biases found for RS with respect to GPS in other studies (you mentioned the Wang et al. 2007 paper). Please be also careful not to mix up different radiosonde types of the same manufacturer (Vaisala) when comparing with the GPS time series at a given station (the BRUS/BRUX example).

Please spend also some time in re-writing a part of the conclusions section, so that the most important findings of your paper are really in there. Also the references in the text and the list of references itself should be made more consistent (also with AMT's recommended abbreviations for journals).

Also the language could be improved at some places, I included a lot of suggestions for modifications in the enclosed manuscript ("Track changes"). And finally, you should also specify at numerous locations in the manuscript which differences, biases are shown, calculated, and every acronym should be spelled out at least once. Be also aware that possible readers of your paper will not all be experienced with GNSS, so try to explain some principles in one sentence (many examples are given in the text).

I hope you take my recommendations into account and I'm looking forward to receive a new version of the manuscript!

With best regards,

Roeland.

P.S. I will send you the docx-file in a separated e-mail, this will be easier to work on.

Dear authors,

I still noted an inconsistency in how the biases are calculated between RS and GNSS. In figure 9, for the site CAGL, these are calculated as RS minus GNSS. In the caption of Fig. 10, it says that the biases are calculated for all sites as GNSS minus RS, but in the text (p9, l291) it says that the overall bias in Fig. 10 is calculated as RS minus GNSS. So, how is it calculated after all? Please check and be correct here, because the exact definition determines whether or not there is an overall dry bias of the radiosondes with respect to GNSS (as found by Wang et al., 2007) or a moist bias of the radiosondes with respect to GNSS (opposite finding in comparison with Wang et al., 2007).
To conclude, to be consistent with the GNSS biases calculated with ERA-interim (GNSS minus ERA-interim as stated in the text, please check), I would also calculate the GNSS biases with RS as GNSS minus RS and therefore adapt either Fig. 9 or Fig. 10. And check another time the discussion of the dry or moist biases of RS with respect to GNSS.

Another remark: please be also consistent when using GPS or GNSS. For instance in Fig. 10, you are using GPS, but as some solutions are GPS+GLONASS (especially after 2008), should you not use GNSS here? Please check at every location in the text where GPS is used that it only applies to GPS only.

Please find below a list of some proposed technical corrections in the final manuscript:
* p1, l18: "(Bruyninx et al., 2015)" instead of "(Bruyninx C. et al., 2015)"
* p3, l89-90: add "are based." after "on which socio-economic response strategies"
* p3, l101: "a climate data record" instead of "a climate data records"
* p4, l103: "this time series still suffers" instead of "this time series still suffer"
* p4, l115: add a space after 1996-2006
* p8, l254: "For those ACs" instead of "For those AC"
* p8, l255: "we have chosen the one carried out" instead of "we have chosen that carried out"
* p9, l274: replace "height component" by "up component"
* p10, l300: Replace "The mean bias ranges from" by "The mean relative bias (GNSS-RS)/GNSS" or the exact definition that has been used [(GNSS-RS)/RS, (RS-GNSS)/RS, (RS-GNSS)/GNSS]
* p10, l302: "The overall mean ZTD bias for all sites is" --> give here also the relative bias (as you did for the extreme values), next to the absolute value.
* p10, l310: "an overall dry bias of about 1.08 mm" instead of "an overall dry bias about 1.08 mm"
* p10, l310: the sentence ends with two .
* P10, l312: add "RS" between "nighttime" and "launches"
* p10, l313: "a small negative (dry) bias of -1.19 mm" instead of "a small negative (dry) bias -1.19 mm"
* p10, l318: replace "remembered" by "kept in mind"
* p11, l335: add the standard deviations in percentages also for the RS-GNSS site distances up to 133 km (third band in Figure 10).
* p11, l338: add "the" before Mediterranean"
* p13, l415: change to "respectively only 1%, 30% and 3% of data pairs available at other stations"
* p14, l454: drop "is" after "as the offset of the specific epoch"
* p15, l494: add the date of GPS week 1055 here.
* p16, l509-l510: explain what you mean by "considering 99% of results roughly". What do you mean by results? Sites? Is roughly referring to 99% (then put it before 99%) or to results (I hope not =:-)
* p16, l523: "a large number of model grid points needs" instead of "a large number of model grid points need".
* p23: please use the last version of the table captions here (e.g. this is not the case for Table 2)
* p25: in Table 2: What does SOFTWAREofte mean?
* p25: in the IONOSPHERE row: "HOI" is not explained (you might consider doing this in the Table caption as well).
* p28: please use the last version of the figure captions here (e.g. this is not the case for Figure 4, in which the explanation for the right part is missing).
* p36: in the figure caption: "repeatability of the final coordinates" instead of "repeatability the final coordinates".
* p38: Show here the ZTD difference GNSS minus RS!
* p39: Show here the ZTD bias GNSS minus RS.
* p40: Specify here how the ZTD biases are calculated (GNSS minus ERA-interim).