Both referees provide a generally positive assessment, but they suggest some modifications. I expect that the authors can provide a revised version which positively addresses the referees' concerns.

The reviewers give opposite assessment about the completeness and relevance of the corrections introduced with the revision. However, both referees support publication of the paper.
Nevertheless, I think that the paper still requires a careful revision.
In particular:
(1) the inverse problem which is considered in this paper is still poorly defined and described;
(2) the scientific relevance and novelty of the proposed approach should be stressed.
Several other suggestions are annotated in the attached file.

After this revision the paper has been improved, as expected. A few technical corrections are still necessary and they are described in the attached annotated version.
From the scientific point of view, it is necessary to complete some of the revised parts of the paper:
1) to reinforce the description of the relevance of this work in the introduction (lines 40 to 46); it seems that the results are important at the national scale for Algeria, less for an international audience;
2) to reinforce the conclusions to stress what is new in this work for an international scientific audience;
3) to improve the description of the inverse modeling, which is heavily based on the proceudre described by Dai et al. (2006); some sentences seem to be a mere "copy-and-paste" from that paper.

Technical corrections were properly introduced where the last version of the manuscript was annotated; however modifications did not fully considered the suggestions, which I listed in the last review:
1) to reinforce the description of the relevance of this work in the introduction; it seems that the results are important at the national scale for Algeria, less for an international audience;
2) to reinforce the conclusions to stress what is new in this work for an international scientific audience;
3) to improve the description of the inverse modeling, which is heavily based on the proceudre described by Dai et al. (2006).
Therefore, I give the authors a last opportunity to improve the quality of the paper.

I think that the paper has now reached a sufficient quality to deserve publication on HESS.
Only a few minor technical corrections, which do not require further review by Editor or referees, could be included before the final submission and copy-editing.