Sochastic rainfall generators are important tools for assessing the potential impact of climate change on hydrologic regimes across the world. They are especially valuable when bottom-up or scenario discovery approaches are used, i.e. those where the feasible input is explored. Gaona et al. present here the second version of their STORM tool in this context. The paper is suitable and interesting, but I have a few constructive comments that I believe require addressing. I list them briefly below.

(1) The title of the manuscript is confusing. First, the title mentions a decision-support tool and not a stochastic rainfall model. To me, the title does not fit the content of the paper. A decision support tool help stakeholders during their decision-making process. I see no such tool described in this manuscript. STORM could be part of a decision support tool, but by itself I do not see how it fits this characterization. In fact, the term ‘decision’ is not mentioned anywhere in the document apart from title and abstract. Second, the authors introduce a stochastic rainfall model, which is great. Why do you not state this in the title? Please change the title to be more suitable to the content of the manuscript. Second, the term gauged watershed is confusing given that it is not clear whether gauging refers to rainfall or streamflow.

(2) The introduction section makes essentially not reference to any other stochastic rainfall model than the one discussed here. While I do not expect a full review, I believe it is paramount to understand what other tools exist and how STORM v2.0 compares to it.

(3) The authors use Walnut Gulch as the experimental case study. From the text, I see no mention that STORM v2.0 has been developed specifically for arid regions. So why do the authors test their model only on this very particular case study? The authors should justify this selection. If STORM v2.0 is not specific to arid catchments, then I would expect to see more examples in other climatic domains. I would also expect to see some justification regarding how all new elements of the model can be tested in this watershed, e.g. the orographic effects. Also, hardly any watersheds are as densely gauged for rainfall as Walnut Gulch is. How does gauge density matter for the results obtained by STORM v2.0?

(4) I am a bit confused by the poor performances (e.g. Fig. 5). Why is there no correlation between estimated and observed seasonal rainfalls (Fig. 5)? Do other stochastic rainfall models have the same problem? I miss a discussion of the ability of STORM v2.0 in comparison to what other models of the same type can do. Also, the authors state that “The scatter plot presented in Fig. 5 clearly shows STORM’s innability to depict extreme stormy seasons, either wetter or drier than those in the historical distribution” – is this not a requirement for a tool like this one?

(5) The figure captions would benefit from more detail so that the reader does not have to read the manuscript text before understanding the figures. E.g. Figure 7 shows a particular experiment where increasing deviation is the goal. This is not clear from the caption. So please include some statement regarding the point of the figure in the caption to make it easier for the reader.

(6) The authors conclude that “STORM’s current weakness is its innability to account for other local hydrometeorological patterns, and global teleconnections that may contribute to intra- and inter-seasonal rainfall variability” – However, there is not concluding discussion how this might be rectified in the future and how this influences the application of the software. Is this not a problem?

See attached report