As pointed out by all the reviewers and according to my own reading of the manuscript, a major revision is required to make it publishable. Also, I noticed that the authors did make a major revision according to the comments as in the attached manuscript. The revision is substantial which deserves a careful re-review. I would suggest the authors upload the revisions and a re-review process will be launched as soon as possible.

The authors made a significant improvement according to the comments. Before its final publication on HESS, the following issues are suggested to be addressed:
1) As water withdraw from the river or water transfer from other watershed could alter watershed water balance, please provide the necessary description to exclude such concerns.
2) About the ecosystem service 'urban heat island', as all the discussions on that are mainly guess, I suggest removal of the corresponding sentences while remaining some general words in the introduction part.
3) Please revise the annotations in the attached pdf file.
4) Please revise the manuscript according to the new comments from all reviewers.

comments by Referee #1:
The authors properly addressed major comments as I brought up previously. I believe the manuscript could be accepted by HESS after minor corrections:
1) please provide detailed descriptions of all your data (data coverage, source) in a table;
2) please numerate functions in the text, also there is a typo in Line 201;
3) please rephrase the title of subsection 3.5, the current one is a little vague;
4) please choose a different color to represent the trend 1990-2015;
5) please adjust y-limit of figure 10;

Comments from Referee #2:
This revision has responded to all my comments on its first version. Here I have some minor comments:
Line 195: replace "assumed that that" with "assumed that";
Line 201: the equation has a typo, %ΔQlulc = (1 - ΔQclim/ΔQ) * 100;
Table 1: replace the number "2.36" (i.e. Water->Forest) with "2" since all significant figures should be the same.
Table 3: Need double check the value of "β" (i.e. the regression coefficient on PET), is it really close to zero?
Fig. 10: Need double check the regression equation. It looks not right.

Comments from Referee #3:
1. Abstract indicates “evapotranspiration (ET) decreased by 23% during 1986-2013”:
It is very likely that water storage changed significantly from 1986-2003 due to land cover and land use change. So I do not think it is proper to use “P-R” as an estimate of ET for this period.
2. Fig 3 is not very informative. I suggest dropping it.
3. Fig. 4: please provide the temporal period for MODIS ET. It should not be from 1986 to 2013.
4. Fig. 7: please provide the temporal period for the analysis.