|The paper consists of an analysis of the trends and variability of integrated water vapor mainly in the period 1995-2010, mainly analyzing ERA-Interim product in comparison with reliable local data (GPS) and another reanalysis, MERRA-2. The manuscript is very well written, and very thorough both in all of its sections. Maybe it can feel dense because of its length, and a shortening could make it easier to read. The statistical treatment is very good to my knowledge. The topic is very interesting and of paramount importance. |
I only have a few questions that I would like the authors to address:
1. Authors claim that the GPS time-series have inhomogeneities. I would like to know if they have considered any kind of statistical treatment to eliminate these inhomogeneities (when it is clear that they are due to GPS and not to ERA-Interim), to see if after correction they match ERA-interim trends.
2. Page 2, line 6: IPCC report is mentioned, but not the year, and it does not appear in references section.
3. Page 2, lines 32-33: could the authors give references to these "several studies"?
4. Page 5, line 1: could the authors provide a references to NASA JPL's product?
5. Page 6, line 4: author claim that interoplation is done with the "4 surrounding grid points". Does this mean that they use the closest north, east, south, west grid points but not the grid point coincident with the GPS location?
6. Page 8, line 24-25: Since the test is not very efficient because of the low number of points, have authors considered using, for instance, monthly averages, in order to have more datapoints? Of course the time-series should be detrended first.
7. Page 9, last line: Authors claim that "to be physically explained, such trends would imply a significant change in the regional and global water cycle". Could this be referenced, or explained?
8. I would suggest that Table S3 have significant trends in bold, or marked in any way to improve readability. But this is up to the authors to decide.
9. Regarding lines 11-12 in page 15, I would like to know what other variables have been considered for "temperature proxy" and why T2m was chosen.
10. Page 16, line 4: CMIP5 is firts used here, so the meaning of the acronym should be provided. A reference here would be appropiate as well.
11. Figures 12 and 15: The figures show the trend spatially averaged. Could the authors explain how the computation of spatial averages was performed?
12. Table S3: p-values do not need so many decimal places, these should be rounded to 3 decimal places.