Thanks for the revisions. The manuscript is almost ready to be accepted for publication. Please, take into account the final suggestions by Reviewer 2 and 3.

* Reviewer 2:
Still, I suggest that the properties of the task (i.e. the CTBTO context with quite sparse measuring network; the assumption of knowledge of release rate boundaries of both, temporal and absolute/day) should be stated (defined) in the Introduction as the definition of your source reconstruction problem since it is quite specific.

* Reviewer 3:
There are two more specific comments I have for this revision.

Line 5, "The Bayesian approach has the advantage of providing credible intervals on the inferred source parameters in a natural way."

Although the authors explained in their responses to my previous comments, "credible intervals" is still too arbitrary to be meaningful. In addition, what can be defined as "a natural way" is questionable as well. This sentence is really too ambiguous to be included in the succinct abstract.

Line 167, "... a uniform prior between 10 and 16"

Based on what the authors stated in their responses to my previous comments ("the prior distribution from which initial samples are drawn is a uniform distribution between 10 and 16), it is mistaken to call it "a uniform prior". Please correct this.